Advances in Intelligent Systems and Computing 457

Munir Merdan
Wilfried Lepuschitz
Gottfried Koppensteiner
Richard Balogh Editors

Robotics in
Education
Research and Practices for Robotics in
STEM Education
Advances in Intelligent Systems and Computing

Volume 457

Series editor
Janusz Kacprzyk, Polish Academy of Sciences, Warsaw, Poland
e-mail: [email protected]
About this Series
The series “Advances in Intelligent Systems and Computing” contains publications on theory,
applications, and design methods of Intelligent Systems and Intelligent Computing. Virtually
all disciplines such as engineering, natural sciences, computer and information science, ICT,
economics, business, e-commerce, environment, healthcare, life science are covered. The list
of topics spans all the areas of modern intelligent systems and computing.
The publications within “Advances in Intelligent Systems and Computing” are primarily
textbooks and proceedings of important conferences, symposia and congresses. They cover
significant recent developments in the field, both of a foundational and applicable character.
An important characteristic feature of the series is the short publication time and world-wide
distribution. This permits a rapid and broad dissemination of research results.

Advisory Board
Chairman
Nikhil R. Pal, Indian Statistical Institute, Kolkata, India
e-mail: [email protected]
Members
Rafael Bello, Universidad Central “Marta Abreu” de Las Villas, Santa Clara, Cuba
e-mail: [email protected]
Emilio S. Corchado, University of Salamanca, Salamanca, Spain
e-mail: [email protected]
Hani Hagras, University of Essex, Colchester, UK
e-mail: [email protected]
László T. Kóczy, Széchenyi István University, Győr, Hungary
e-mail: [email protected]
Vladik Kreinovich, University of Texas at El Paso, El Paso, USA
e-mail: [email protected]
Chin-Teng Lin, National Chiao Tung University, Hsinchu, Taiwan
e-mail: [email protected]
Jie Lu, University of Technology, Sydney, Australia
e-mail: [email protected]
Patricia Melin, Tijuana Institute of Technology, Tijuana, Mexico
e-mail: [email protected]
Nadia Nedjah, State University of Rio de Janeiro, Rio de Janeiro, Brazil
e-mail: [email protected]
Ngoc Thanh Nguyen, Wroclaw University of Technology, Wroclaw, Poland
e-mail: [email protected]
Jun Wang, The Chinese University of Hong Kong, Shatin, Hong Kong
e-mail: [email protected]

More information about this series at http://www.springer.com/series/11156

Munir Merdan ⋅ Wilfried Lepuschitz
Gottfried Koppensteiner ⋅ Richard Balogh
Editors

Robotics in Education
Research and Practices for Robotics
in STEM Education

123
Editors
Munir Merdan Gottfried Koppensteiner
Practical Robotics Institute Austria (PRIA) Practical Robotics Institute Austria (PRIA)
Vienna Vienna
Austria Austria

Wilfried Lepuschitz Richard Balogh

Practical Robotics Institute Austria (PRIA) URPI FEI STU
Vienna Bratislava
Austria Slovakia

ISSN 2194-5357 ISSN 2194-5365 (electronic)

Advances in Intelligent Systems and Computing
ISBN 978-3-319-42974-8 ISBN 978-3-319-42975-5 (eBook)
DOI 10.1007/978-3-319-42975-5

Library of Congress Control Number: 2016946927

© Springer International Publishing Switzerland 2017

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer International Publishing AG Switzerland
Preface

We are glad to present the proceedings of the 7th International Conference on

Robotics in Education (RiE) held in Vienna, Austria, during April 14–15, 2016.
The RiE is organized every year with the goal to provide researchers in the field of
Educational Robotics the opportunity for the presentation of relevant novel
researches in a strongly multidisciplinary context.
Educational Robotics is an innovative way for increasing the attractiveness of
science education and scientific careers in the view of young people. Robotics
represents a multidisciplinary and highly innovative domain encompassing physics,
mathematics, informatics and even industrial design as well as social sciences. As a
multidisciplinary field, it promotes the development of systems thinking and
problem solving. Moreover, due to various application areas, teamwork, creativity
and entrepreneurial skills are required for the design, programming and innovative
exploitation of robots and robotic services. Robotics confronts learners with the
four areas of Science, Technology, Engineering and Mathematics (STEM) through
the design, creation and programming of tangible artifacts for creating personally
meaningful objects and addressing real-world societal needs. As a consequence, it
is regarded as very beneficial if engineering schools and university program studies
include the teaching of both theoretical and practical knowledge on robotics. In this
context current curricula need to be improved and new didactic approaches for an
innovative education need to be developed for improving the STEM skills among
young people. Moreover, an exploration of the multidisciplinary potential of
robotics towards an innovative learning approach is required for fostering the
pupils’ and students’ creativity leading to collaborative entrepreneurial, industrial
and research careers in STEM.
In these proceedings we present the latest achievements in research and devel-
opment in educational robotics. The book offers a range of methodologies for
teaching robotics and presents various educational robotics curricula and activities.
It includes dedicated chapters for the design and analysis of learning environments
as well as evaluation means for measuring the impact of robotics on the students’
learning success. Moreover, the book presents interesting programming approaches

v
vi Preface

as well as new applications, the latest tools, systems and components for using
robotics. The presented applications cover the whole educative range, from ele-
mentary school to high school, college, university and beyond, for continuing
education and possibly outreach and workforce development. The book provides a
framework involving two complementary kinds of contributions: on the one hand
on technical aspects and on the other hand on didactic matters. In total, 25 papers
are part of these proceedings after careful revision. We would like to express our
thanks to all authors who submitted papers to RiE 2016, and our congratulations to
those whose papers were accepted.
This publication would not have been possible without the support of the RiE
International Program Committee and the Conference Co-Chairs. The editors also
wish to express their gratitude to the volunteer students and local staff, which
significantly contributed to the success of the event. All of them deserve many
thanks for having helped to attain the goal of providing a balanced event with a high
level of scientific exchange and a pleasant environment. We acknowledge the use
of the EasyChair conference system for the paper submission and review process.
We would also like to thank Dr. Thomas Ditzinger and Springer for providing
continuous assistance and advice whenever needed.

Vienna, Austria Munir Merdan

Vienna, Austria Wilfried Lepuschitz
Vienna, Austria Gottfried Koppensteiner
Bratislava, Slovakia Richard Balogh
Organization of RiE 2016

Co-Chairpersons

Richard Balogh, Slovak University of Technology in Bratislava, SK

Wilfried Lepuschitz, Practical Robotics Institute Austria, AT
David Obdržálek, Charles University in Prague, CZ

International Program Committee

Dimitris Alimisis, Edumotiva-European Lab for Educational Technology, GR

Julian Angel-Fernandez, Vienna University of Technology, AT
Jenny Carter, De Montfort University in Leicester, GB
Dave Catlin, Valiant Technology, GB
Stavros Demetriadis, Aristotle University of Thessaloniki, GR
G. Barbara Demo, DipartimentoInformatica—Universita Torino, IT
Jean-Daniel Dessimoz, Western Switzerland University of Applied Sciences and
Arts, CH
NikleiaEteokleous, Robotics Academy—Frederick University Cyprus, CY
Hugo Ferreira, Instituto Superior de Engenharia do Porto, PT
Paolo Fiorini, University of Verona, IT
Carina Girvan, Cardiff University, GB
GrzegorzGranosik, Lodz University of Technology, PL
IvayloGueorguiev, European Software Institute Center Eastern Europe, BG
Martin Kandlhofer, Graz University of Technology, AT
BoualemKazed, University of Blida, DZ
Gottfried Koppensteiner, Practical Robotics Institute Austria, AT
TomášKrajník, University of Lincoln, UK
Miroslav Kulich, Czech Technical University in Prague, CZ
Chronis Kynigos, University of Athens, GR

vii
viii Organization of RiE 2016

Lara Lammer, Vienna University of Technology, AT

Martin Mellado, Instituto ai2—UniversitatPolitècnica de València, ES
Munir Merdan, Practical Robotics Institute Austria, AT
Michele Moro, University of Padova, IT
MargusPedaste, University of Tartu, EE
Pavel Petrovič, Comenius University in Bratislava, SK
Alfredo Pina, Public University of Navarra, ES
Pericle Salvini, BioRobotics Institute—ScuolaSuperioreSant’Anna, IT
João Machado Santos, University of Lincoln, GB
Alexander Schlaefer, Hamburg University of Technology, DE
Fritz Schmöllebeck, University of Applied SciencedTechnikum Wien, AT
FrantišekŠolc, Brno University of Technology, CZ
Gerald Steinbauer, Graz University of Technology, AT
Roland Stelzer, INNOC—Austrian Society for Innovative Computer Sciences, AT
DavorSvetinovic, Masdar Institute of Science and Technology, AE
Igor M. Verner, Technion—Israel Institute of Technology, IL
Markus Vincze, Vienna University of Technology, AT
Francis Wyffels, Ghent University, BE

Local Conference Organization

Gottfried Koppensteiner, Vienna Institute of Technology/Practical Robotics Insti-

tute Austria, AT
Wilfried Lepuschitz, Practical Robotics Institute Austria, AT
Munir Merdan, Practical Robotics Institute Austria, AT
Contents

Part I Didactic and Methodologies for Teaching Robotics

Activity Plan Template: A Mediating Tool for Supporting Learning
Design with Robotics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Nikoleta Yiannoutsou, Sofia Nikitopoulou, Chronis Kynigos,
Ivaylo Gueorguiev and Julian Angel Fernandez
V-REP and LabVIEW in the Service of Education . . . . . . . . . . . . . . . . . 15
Marek Gawryszewski, Piotr Kmiecik and Grzegorz Granosik
Applied Social Robotics—Building Interactive Robots
with LEGO Mindstorms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Andreas Kipp and Sebastian Schneider
Offering Multiple Entry-Points into STEM for Young People . . . . . . . . . 41
Wilfried Lepuschitz, Gottfried Koppensteiner and Munir Merdan

Part II Educational Robotics Curricula

How to Teach with LEGO WeDo at Primary School . . . . . . . . . . . . . . . . 55
Karolína Mayerové and Michaela Veselovská
Using Modern Software and the ICE Approach When Teaching
University Students Modelling in Robotics . . . . . . . . . . . . . . . . . . . . . . . . 63
Sven Rönnbäck
Developing Extended Real and Virtual Robotics Enhancement
Classes with Years 10–13 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Peter Samuels and Sheila Poppa
Project Oriented Approach in Educational Robotics: From Robotic
Competition to Practical Appliance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Anton Yudin, Maxim Kolesnikov, Andrey Vlasov and Maria Salmina

ix
x Contents

ER4STEM Educational Robotics for Science, Technology,

Engineering and Mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Lara Lammer, Wilfried Lepuschitz, Chronis Kynigos, Angele Giuliano
and Carina Girvan

Part III Design and Analysis of Learning Environments

The Educational Robotics Landscape Exploring Common
Ground and Contact Points . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Lara Lammer, Markus Vincze, Martin Kandlhofer and Gerald Steinbauer
A Workshop to Promote Arduino-Based Robots
as Wide Spectrum Learning Support Tools . . . . . . . . . . . . . . . . . . . . . . . 113
Francesca Agatolio and Michele Moro
Robotics in School Chemistry Laboratories . . . . . . . . . . . . . . . . . . . . . . . 127
Igor M. Verner and Leonid B. Revzin
Breeding Robots to Learn How to Rule Complex Systems . . . . . . . . . . . 137
Franco Rubinacci, Michela Ponticorvo, Onofrio Gigliotta
and Orazio Miglino
A Thousand Robots for Each Student: Using Cloud Robot
Simulations to Teach Robotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
Ricardo Tellez

Part IV Technologies for Educational Robotics

Networking Extension Module for Yrobot—A Modular
Educational Robotic Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Michal Hodoň, Juraj Miček and Michal Kochláň
Aeris—Robots Laboratory with Dynamic Environment. . . . . . . . . . . . . . 169
Michal Chovanec, Lukáš Čechovič and Lukáš Mandák
UNC++Duino: A Kit for Learning to Program Robots
in Python and C++ Starting from Blocks . . . . . . . . . . . . . . . . . . . . . . . . 181
Luciana Benotti, Marcos J. Gómez and Cecilia Martínez
Usability Evaluation of a Raspberry-Pi Telepresence Robot
Controlled by Android Smartphones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Krit Janard and Worawan Marurngsith
On the Design and Implementation of a Virtual Machine
for Arduino . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Gonzalo Zabala, Ricardo Moran, Matías Teragni and Sebastián Blanco
Model-Based Design of a Competition Car . . . . . . . . . . . . . . . . . . . . . . . . 219
Richard Balogh and Marek Lászlo
Contents xi

Part V Measuring the Impact of Robotics on Students’ Learning

Student-Robot Interactions in Museum Workshops: Learning
Activities and Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Alex Polishuk and Igor Verner
Robot Moves as Tangible Feedback in a Mathematical
Game at Primary School . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Sonia Mandin, Marina De Simone and Sophie Soury-Lavergne
Personalizing Educational Game Play with a Robot Partner. . . . . . . . . . 259
Mirjam de Haas, Iris Smeekens, Eunice Njeri, Pim Haselager,
Jan Buitelaar, Tino Lourens, Wouter Staal, Jeffrey Glennon
and Emilia Barakova
Robot as Tutee . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Lena Pareto
Concept Inventories for Quality Assurance of Study
Programs in Robotics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Reinhard Gerndt and Jens Lüssem
 Part I
Didactic and Methodologies
for Teaching Robotics
Activity Plan Template: A Mediating Tool
for Supporting Learning Design
with Robotics

Nikoleta Yiannoutsou, Sofia Nikitopoulou, Chronis Kynigos,

Ivaylo Gueorguiev and Julian Angel Fernandez

Abstract Although the educational use of robotics is recognised since several

decades, only recently they started being broadly used in education, formal and non
formal. In this context many different technologies have emerged accompanied by
relevant learning material and resources. Our observation is that the vast number of
learning activities is driven by multiple “personal pedagogies” which results in the
fragmentation of the domain. To address this problem we propose the construct of
“activity plan template”, a generic design tool that will facilitate different stake-
holders (teachers, instructors, researchers) to design learning activities for different
robotic toolkits. In this paper we discuss the characteristics of the activity plan
template and the research process employed to generate it. Since we report work in
progress, we present here the first version of the activity plan template, the
construction of which is based on a set of best practices identified and on previous
work for the introduction of digital technologies in education.

Keywords Activity plan template ⋅ Learning design ⋅ Educational robotics

N. Yiannoutsou (✉) ⋅ S. Nikitopoulou ⋅ C. Kynigos

UoA ETL, National and Kapodistrian University of Athens, Athens, Greece
e-mail: [email protected]
S. Nikitopoulou
e-mail: [email protected]
C. Kynigos
e-mail: [email protected]
I. Gueorguiev
ESI CEE, European Software Institute Center Eastern Europe, Sofia, Bulgaria
e-mail: [email protected]
J.A. Fernandez
ESI CEE ACIN Institute of Automation and Control,
Vienna University of Technology, Vienna, Austria
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 3

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_1
4 N. Yiannoutsou et al.

1 Introduction

The educational robotics landscape is vast and fragmented in and outside schools.
In the last two decades, robots have started their incursion into the formal educa-
tional system. Although diverse researchers have stressed the learning potential of
robotics, the slow pace of their introduction is partially justified by the cost of the
kits and the schools’ different priorities in accessing technology. Recently, the cost
of kits has decreased, whereas their capabilities and the availability of supporting
hardware and software has increased [1, 2]. With these benefits, educational
robotics kits have become more appealing to schools. In this context, various
stakeholders—technology providers, teachers, academics, companies focusing on
delivering educational material etc.—invest in the creation of different learning
activities around robotic kits, in order to showcase their characteristics and make
them attractive in and out of schools. Thus, a growing number of learning activities
have emerged. These activities share common elements but they are also very
diverse in that they address different aspects of Robotics as teaching and learning
technology with their success lying in how well they have identified these aspects
and how well they address them. This is partly due to the fact that Robotics is a
technology with special characteristics when compared to other learning tech-
nologies: they are inherently multidisciplinary, which in terms of designing a
learning activity might mean collaboration and immersion into different subject
matters; they are extensively used in settings of formal and non formal learning and
thus involving different stakeholders; their tangible dimension causes perturbations
—especially in formal educational settings—which are closely related to the
introduction of innovations in organizations and schools (i.e. from considering
classroom orchestrations to establishing or not, connections with the curriculum
etc.); they are at the heart of constructionist philosophy for teaching and learning
[3]; they are relevant to new learning practices flourishing now over the internet like
the maker movement, “Do It Yourself” and “Do It With Others” communities etc.
With this in mind, we argue that we need to take a step back from the level of
specific learning activities and create a more generic design instrument i.e. an
activity plan template, which: (a) it will be pedagogically grounded on the partic-
ular characteristics of robotics as a teaching and learning tool (b) it will be adaptable
to different learning settings (formal−non-formal) (c) it will afford generating dif-
ferent examples of learning activities for different types of kits (d) it will focus on
making explicit the implicit aspects of the learning environment and (e) it will urge
designers to think “out of the box” by reflecting its content. In the following
sections we describe the theoretical background supporting the concept of activity
plan template as a design instrument and the method for developing an activity plan
template for teaching and learning with Robotics.
Activity Plan Template: A Mediating Tool for Supporting … 5

2 Theoretical Background

Aiming to explain, in this section, the role of a generic design instrument such as
the activity plan template, in addressing the problem of fragmentation in the
practice of using educational robotics for learning, we will discuss the dimensions
and functions of design in education.
Everyone designs who devises courses of action aimed at changing existing
situations into desired ones [4, cited in 5]. With this definition we aim to highlight
that design is an integral part of the teaching profession. Acknowledging this
dimension in teaching, and with the advent of digital technologies in schools,
design based research has been implemented as an approach to orchestrate and
study the introduction of innovation in education [6]. Furthermore, in the field of
education, design has been introduced as the bridge between theory and practice [5]
because design is expected to play a dual role: (a) to guide practice informed by
theory and (b) to inform back the theory after the evaluation of the design in
practice. Thus, in this context, design is not only an organized sequence of stages,
all of which compose an orchestration of the learning process [7] but it is also a
reflection and an evaluation tool.
Gueudet and Trouche [8] focusing mainly on resources and documents designed
by teachers (e.g. activity or lesson plans), reveal another dimension of design as
they describe it as a tool that not only expresses but also shapes the teacher’s
personal pedagogies, theories, beliefs, knowledge, reflections and practice. The
term they use to describe this process is Documentational Genesis. A core element
of this approach is instrumental theory [9] according to which the characteristics of
the resources teachers select to use, shape their practice on the one hand (instru-
mentation) and on the other hand, the teachers’ knowledge shapes the use of the
resources as teachers appropriate them to fit their personal pedagogies (instru-
mentalization). As a result of the above, teacher designs, according to Pepin et al.
[10], are evolving or living documents—in the sense that they are continuously
renewed, changed and adapted.
Design as expressive medium for teachers and educators, can also function as an
instrument for sharing, communicating, negotiating and expanding ideas within
interdisciplinary environments. This property of teacher designs is linked to the
concept of boundary objects and boundary crossing [11]. The focus here is on the
artefact (in our case activity plan) that mediates a co-design process by helping
members of different disciplines to gain understanding of each other’s perspectives
and knowledge. Educational Robotics for STE(A)M is such an interdisciplinary
environment which involves an understanding of related but different domains (i.e.
Science, Technology, Engineering, Arts, Mathematics) and involves players from
industry, academia and organizers of educational activities.
A problem with all these designs, especially when they involve integration of
technologies, is that they are driven by a multitude of “personal pedagogies” the
restrictions of which result in adapting technologies to existing practices [12].
Conole (ibid) argues that the gap between the potential of digital technologies to
6 N. Yiannoutsou et al.

support learning and their implementation in practice can be bridged with a “me-
diating artefact” to support teacher designs. She continues claiming that such a
mediating artefact should be structured according to specific pedagogic approaches
and should focus on abstracting essential and transferable properties of learning
activities that are not context bound. The activity plan template can play the role of
the mediating artefact equipping professionals with a structured means to describe,
share and shape their practices. This way we can contribute in addressing the
problem of fragmentation in the learning activities regarding the use of Educational
robotics.

3 Developing an Activity Plan Template for Educational

Robotics

The work reported in this paper takes place in the context of the European project
ER4STEM. The main objective of this project is to refine, unify and enhance
current European approaches to STEM education through robotics in one open
operational and conceptual framework. The development of activity plan templates
contributes towards this direction as it provides a generic design instrument that
identifies critical elements of teaching and learning with robotics based in theory
and practice and in that contributes to the description of effective learning and
teaching with robotics. The process through which we develop the activity plan
templates in this project includes the following steps: We create a first draft based
on (a) on identifying and analyzing a set of good practices and (b) previous work on
activity plans that involve innovative use of technologies for teaching and learning.
The next step is to use this first draft to design and implement workshops with
Robotics in different educational settings and systems. During this implementation
we will collect data that will allow us to evaluate, refine and re-design the activity
plan template so as to be a useful and pedagogically grounded instrument for
designing learning activities. In this paper we are at the first stages of our research
and thus we will report on: (a) a set of criteria that we developed in order to identify
good practices and (b) the first draft of the activity plan template.

3.1 Identifying Best Practices

The criteria for selecting best practices in the domain of educational robotics were
formed through a bottom-up empirical process. Specifically, three researchers from
different research teams of the consortium worked independently to select a set of
best practices from robotics conferences, competitions, seminars and workshops
organized by different institutions. This was the first phase of the selection process,
which was not done in a structured way. The second phase included analysis and
Activity Plan Template: A Mediating Tool for Supporting … 7

reflection on phase one. Specifically, the criteria were shaped by (a) an analysis of
the content of five examples of best practices already selected and (b) elaboration of
the criteria that researchers had implicitly applied during the selection of the
specific best practices. Next the items that—from the analytic and the reflective
process—were identified to be part of what could be considered best practice in the
field of educational robotics were synthesized in one document.
The best practice selection criteria are designed to feed into the activity plans
(and not map directly into them) by providing interesting and new ideas for
(a) concepts, objectives, artefacts (b) orchestration (c) teaching interventions and
learning process (d) implementation process and (e) evaluation process.
Criteria.
The criteria developed for identifying best practices are divided in two categories.
One category is mainly a set of prerequisites, which should be covered in order for
an event or activity to be considered. The other category consists of the main
criteria that identify best practice aspects of the activities.
Prerequisites:
• The topic includes concepts related to the following subjects:
Science-Technology-Business-Engineering-Art-Mathematics or something from
another discipline but related to robotics.
• The activity−event shows that it has constructionist elements: i.e. it is not just a
presentation of tools or predefined guidelines.
• The activity−event is innovative, related to student or citizen interests.
• The activity−event includes technology related to educational robotics.
Main Criteria.
In case that the “educational robotic event” is assessed as relevant according to the
aforementioned basic pre-requisites, then the process continues with the assessment
of the following parameters (see Table 1). Not all parameters have to be met in
order for an event or activity to be considered as good practice. On the contrary,
these parameters help us to collect good practices with respect to different
dimensions of robotics activities stemming from different sources.

3.2 First Version of the Activity Plan Template

In this section we discuss the rationale and the main structure of the first version of
the activity plan template. The basic pedagogical theory underlying its design is
constructionism, where learning is connected to powerful ideas inherent in con-
structions with personal meaning for the students. Another aspect underlying our
design rationale is the emphasis on the social dimension of the construction process
aiming to cultivate a specific learning attitude growing out of sharing, discussing
and negotiating ideas. Furthermore, this first version of the activity plan template, is
designed to be adaptable to different learning settings (: i.e. formal−non formal),
8 N. Yiannoutsou et al.

Table 1 Criteria used for the selection of best practices

Parameters Description
Context • Place: provides information about the space where the educational robotic
activity takes place. This information is crucial to determine other aspects of
the learning design such as orchestration issues, formal or non-formal
settings etc. Possible examples can be school, museum, science institutions,
or other educational scientific organizations.
• Participants’ description: provides information regarding issues such as
age, number, culture, background etc. The activity is considered as good
practice if it is aligned to the age of the participants, the number, the prior
knowledge of the participants on a specific subject, etc.
• Theoretical framework: refers to the pedagogical approach used in
implementing the educational activity e.g. DIY (Do It Yourself), DIWO (Do
It With Others), Constructionism, STEM education, Design. In several cases
the theoretical framework is implicit and can be inferred from the way the
activity is orchestrated and designed.
Educational • Connection with a curriculum: This dimension provides information
activity regarding issues of connecting the teaching of robotics to specific topics of
national curricula. It is not expected to apply to all events or activities
identified.
• Motivation for the activity: Provides information on what has motivated
the organization of the specific activity (e.g. introducing girls to robotics,
elaborating on specific STEM concepts, using art to explore robotics etc.).
In identifying good practices we are looking for interesting motivations and
the way the activity is organized to support this motivation. Special focus is
given to events that are designed to motivate young people to learn STEM
disciplines.
• Description of the activity: Provides information regarding the
implementation of the activity. This information helps out in identifying if
the activity matches the context the motivation etc. The activity description
is expected to refer to issues regarding the duration, tasks, orchestration,
grouping, learner interaction (i.e. where is the emphasis concerning the
action, the relationships, the roles in the group and the teacher’s role).
Tools • Technology used—selection criteria: Provides information on the
specific technology used for the implementation of the activity. It is
considered as good practice if the educational robotic event is based on
technology that follows the latest trends, it is compatible with the
background of the participants, facilitates well the objectives and the
motivation of the activity, it is presented in a way that it is understandable
by the specific target group in the workshops and is similar to technologies
used by young people in their everyday life e.g. mobile and cloud solutions
• Type of artefacts produced: This parameter involves the output of the
activity or the event. It is considered as good practice if the artefacts
produced during the educational robotic event are interesting and engaging;
participants are interested to use the artefacts and to apply them in different
domains of their lives.
Evaluation The description of the activity provides information regarding methods and
results of its evaluation, including the perspectives of the participants and
the reflection of the teacher-instructor on aspects that might need
improvement or are going to be changed in next implementations
(continued)
Activity Plan Template: A Mediating Tool for Supporting … 9

Table 1 (continued)
Parameters Description
Sustainability • Cost of the activity: This dimension involves information regarding
mainly costs of the material and organizational costs. It is considered a good
practice if the activity requires materials or tools that are reasonably priced
compared to other related activities.
• Activity Financing: The activity−event is considered a good practice
with respect to this dimension if it has a sustainable model for financing in
mid-term period, e.g. self-financing through fees, wide voluntary base,
partnership with public organizations such as municipalities, schools or long
term sponsorship partners.
• Activity Repetition: An activity−event is considered a good practice if it
is performed sustainably for at least three subsequent periods in close
cooperation with schools or other educational organizations.
Accessibility The information regarding this parameter involves mainly the sharing of
activity related material (i.e. manuals, guidelines etc.), in a way (i.e. open
access, structuring of information) that allows the activity−event to be
replicated by other relevant stakeholders.

thus, its structure is modular and the intention is to allow “selective exposure” of its
elements to different stakeholders (the term “selective exposure” is borrowed from
Blikstein [13] to describe the intentional hiding of some of the template elements,
according to the relevant settings or stakeholders).
This first version discussed here, is informed by an analysis of the best practices
identified and it is based on previous work on activity plan templates that aim at the
integration of digital technologies in learning [14]. The structure of the Activity
plan template is presented in detail in Table 2 and addresses the following aspects:
(a) the description of the scenario with reference to the different domains involved,
different types of objectives, duration and necessary material; (b) contextual
information regarding space and characteristics of the participants; (c) social
orchestration of the activity (i.e. group or individual work, formulation of groups
etc.); (d) a description of the teaching and learning procedures where the influence
of the pedagogical theory is mostly demonstrated; (e) expected student construc-
tions; (f) description of the sequencing and the focus of activities; (g) means of
evaluation.
Future work will focus on refinement of the activity plan template through its
use by ER4STEM partners to create their activity plans and through data collected
during the implementation of these activity plans in realistic situations (workshops).
10 N. Yiannoutsou et al.

Table 2 First version of the activity plan template

Title
Author Teacher, Designer
1. Focus, set up and requirements for the activity
Domain • Primary domain (Select one of the following): Science;
Technology; Business; Engineering; Arts; Mathematics.
• Contextual (Peripheral) domain (provide a rating of the level of
emphasis on concepts in each of these domains): Science (0−10);
Technology (0−10); Business (0−10); Engineering (0−10); Arts (0
−10); Mathematics (0−10).
Objectives Objectives are organized in a set of four different categories:
• Subject matter: i.e. study the angle and position of all materials
(servo motors, circuits, sensors), as well as the construction of the legs
in order for the robot—insect to be autonomous and move correctly.
• Technology use: i.e. Programming with Arduino.
• Social and collaborative skills: i.e. develop collaborative skills, take
roles within groups.
• Argumentation and fostering of maker culture: i.e. practice
making conjectures about how the robot will react to external stimuli
based on the program given.
Time • Duration: i.e. 5 weeks
• Schedule: i.e. 2 h per week
Materials and • Digital artifact: e.g. programming language, visual interface, robot
artifacts simulation etc.
• Robotic artifact: i.e. the technology and the robot form, e.g. an
insect, a car etc.
• Student’s workbook and manual: i.e. a manual with step-by-step
instructions for the electronic and the programming part.
• Teacher’s instruction book and manual: teacher’s notes with a
template of e.g. three incisive stages and five steps for the first two
stages.
2. Space and students
Students (target • Sex and age: e.g. boys and girls, 17 years old
audience): • Prior knowledge: e.g. little if any knowledge of Arduino, good
knowledge of electronics
• Nationality and cultural background: e.g. 5 pupils from Albania
and 10 from Greece
• Social status and social environment: e.g. under-privileged area,
mainstream public school, elite private school
• Special needs and abilities: e.g. ADD, dyslexia, Soc. Em. Behavior
Disorders, gifted, other
Space info • Organizational and cultural context: e.g. in school at the
technology laboratory, during project time in after school established
voluntary club activity.
• Physical characteristics: e.g. indoors, floor
3. Social orchestration
Participants Students: e.g. 15
Tutors: e.g. 2
(continued)
Activity Plan Template: A Mediating Tool for Supporting … 11

Table 2 (continued)
Title
Grouping Setting: students in a normal classroom, around light mobile tables,
in small groups
Grouping criteria: mixed ability, mixed gender
Interaction during the Actions: exchange ideas, dialogue, negotiation, debate.
activity Relationships: collaborative, competitive
Roles in the group: pre-defined roles, emergent roles
Support by the tutor(s): support, intervene, self-regulatory
4. Teaching and learning procedures
Teacher’s role Mentor, consultant, researcher, instructor
Teaching methods Demonstrate, engage by example
Student expected Writing, observing, constructing, discussing, negotiating,
activity
Student learning • Designed conflicts and misconceptions: do the activity designers
processes wish to bring students in conflict with mistaken conceptions
documented in educational research or their teaching practice?
• Learning processes emphasized: e.g. emphasis on analyzing robot
behavior in order to refine and reflect on the code that defines this
behavior.
• Expected relevance of alternative knowledge: e.g. students are
expected to investigate the structure of an insect’s body (biology) in
order to construct their robot.
5. Student productions
Artifacts—robots • Assignment: What tasks shall the robot perform (e.g. entertain,
bring things, call help, vacuum clean etc.)?
• Interaction: What are the means of communication with the robot
(speech, gesture, mind control, buttons, app etc.)?
• Morphology: How does the robot look like? What material is it made
of (e.g. machine-like, zoomorphic, anthropomorphic, cartoon-like
etc.)?
• Behavior: What shall the robot behave like (e.g. butler, friend, pet,
protector, teacher etc.)?
• Material: What parts are needed for the construction of the robot
(e.g. electronics, software, mechanics etc.)?
Programming • Structure of code-commands
• Elements (e.g. iteration, selection, variables)
• Conditionals (e.g. event handling)
Discussion • Descriptive—explanatory: description of a situation, a construct or
an idea for others to understand and/or to implement.
• Alternative: provision of solutions to problems, provision of
alternatives if a dead end is reached.
• Critical—objection: revision of other’s constructs and ideas,
identification of problems, challenge of ideas.
• Contributory—extending: sharing of resources, provision of ideas
towards improving an existing construct or initial idea.
(continued)
12 N. Yiannoutsou et al.

Table 2 (continued)
Title
6. Sequence and description of activities
Phasing • Phase 1: Construction phase—hands on the robot (duration one hour)
• Phase 2: Assembly discussion: All groups present the robots they
have constructed and discuss challenges and problems (Duration
20 min)
• Phase 3: Programming: constructing the robot’s behavior. Groups
can exchange ideas and ask for help from each other (duration 1 h).
• Phase 4: Presentation of the final construct: A short video
demonstrating the robot and its behavior or a blog presentation
including a photograph, a short description and the code.
7. Assessment procedures
Formative • Pupil voice activities (Interviews with students, Questionnaire)
• Observation notes
• Peer assessment
Summative • Essays
• Tests
• Student productions (code-robots-textual discussions)
• Mark sheet

4 Conclusion

In this paper we discussed the role of activity plan templates as mediating artifacts
in harnessing the potential of educational Robotics for learning and in addressing
the issue of fragmentation in the domain. The concept of a mediating artifact was
adopted here to describe a generic learning design instrument that is based on: (a) a
specific pedagogical theory and (b) the particularities of robotics as technologies.
The activity plan template is an abstraction of what we have identified as essential
and transferrable elements of learning with robotics. The work reported here is in
progress, thus the activity plan template presented, is going to be evaluated in
practice by teachers who will use it to create their own activity plans and by
researchers and students during the implementation of these plans in practice.
Feedback generated from this process will be used to inform the activity plan
template so as to achieve (a) a level of abstraction that it will make it adaptable to
different settings and (b) a level of detail that will demonstrate the influence of a
specific pedagogical approach and will address the particularities of Robotics.

Acknowledgments The project has received funding from the European Union’s Horizon 2020
research and innovation program under grant agreement No. 665972. Project Educational Robotics
for STEM: ER4STEM
Activity Plan Template: A Mediating Tool for Supporting … 13

References

1. Alimisis, D.: Robotic technologies as vehicles of new ways of thinking, about constructivist
teaching and learning: the TERECoP project. IEEE Robot. Autom. Mag. 16, 21–23 (2009)
2. Miller, M.: Mobile building blocks 2014. Mobile Cores. PC Mag. (2014)
3. Papert, S.: Mindstorms: Children, Computers, and Powerful Ideas. Basic Books, Inc. (1980)
4. Simon, H.A.: The sciences of the artificial. MIT Press, Cambridge, MA (1969)
5. Mor, Y., Winters, N.: Design approaches in technology-enhanced learning. Interact. Learn.
Environ. 15, 61–75 (2007)
6. The Design-Based Research Collective: Design-based research: an emerging paradigm for
educational inquiry. Educ. Res. 5–8 (2003)
7. Trouche, L.: Managing the complexity of human/machine interactions in computerized
learning environments: guiding students’ command process through instrumental
orchestrations. Int. J. Comput. Math. Learn. 9, 281–307 (2004)
8. Gueudet, G., Trouche, L.: Towards new documentation systems for mathematics teachers?
Educ. Stud. Math. 71, 199–218 (2009)
9. Verillon, P., Rabardel, P.: Cognition and artifacts: a contribution to the study of though in
relation to instrumented activity. Eur. J. Psychol. Educ. 10, 77–101 (1995)
10. Pepin, B., Gueudet, G., Trouche, L.: Re-sourcing teachers’ work and interactions: a collective
perspective on resources, their use and transformation. ZDM Math. Educ. 45, 929–943 (2013)
11. Kynigos, C., Kalogeria, E.: Boundary crossing through in-service online mathematics teacher
education: the case of scenarios and half-baked microworlds. ZDM 1–13 (2012)
12. Conole, G.: The role of mediating artefacts in learning design. In: Handbook of Research on
Learning Design and Learning Objects: Issues Applications and Technologies, pp. 108–208
(2008)
13. Blikstein, P.: Computationally enhanced toolkits for children: historical review and a
framework for future design. Found. Trends®. Hum.Comput. Interact. 9, 1–68 (2015)
14. Yiannoutsou, N., Kynigos, C.: Boundary objects in educational design research: designing an
intervention for learning how to learn in collectives with technologies that support
collaboration and exploratory learning. In: Plomp, T., Nieveen, N. (eds.) Educational
Design Research: Introduction and Illustrative Cases, pp. 357–379. SLO, Netherlands Institute
for Curriculum Development, Enschede, The Netherlands (2013)
V-REP and LabVIEW in the Service
of Education

Marek Gawryszewski, Piotr Kmiecik and Grzegorz Granosik

Abstract The following paper exposes an effective approach of teaching robotics

by applying very efficient and popular set of tools. It demonstrates a combination
of applications such as V-REP and LabVIEW in order to be later utilized for rapid
prototyping, algorithm design and revival of both simple as well as fairly advanced
robotic environments. The described teaching methodology has been successfully
applied to high-school students during their facultative robotics courses which have
been taking place at the Lodz University of Technology. Detailed summary of the
results of this exertion is also provided.

Keywords Educational robotics ⋅ V-REP ⋅ LabVIEW ⋅ Simulation environment ⋅

Rapid prototyping

1 Introduction

According to various statistics [1, 2], robotics market has been exponentially grow-
ing for the last few years and nothing indicates that this could change in the nearest
future. Becoming more and more specialized, robots take over a significant number
of tasks performed by the human so far. Particular growth can be seen not only in the
industrial process automation area, but also, and especially, in consumer robotics.
Even though the prophecies of robots taking over most people’s jobs still seem to
be a fantasy, the transition is clearly visible.

M. Gawryszewski (✉) ⋅ P. Kmiecik ⋅ G. Granosik

Institute of Automatic Control, Lodz University of Technology,
B. Stefanowskiego 18/22, 90-924, Lodz, Poland
e-mail: [email protected]
URL: http://robotyka.p.lodz.pl
P. Kmiecik
e-mail: [email protected]
G. Granosik
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 15

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_2
16 M. Gawryszewski et al.

At the moment, this means that we have to get used to the presence of robotic
devices in our lives and learn to cooperate with them. Even though some of the job
opportunities have already expired, a huge amount of new ones is just opening. How-
ever, the transformation we are witnessing must enforce the change in the way we
perceive robots and in the direction in which potential workforce should be trained.
The conclusion is that robotics should be introduced to the vast majority of students
at possibly early level of education—to get them accustomed to the new technologies
and in the way that is not overwhelming—to let them develop interest in it.
In the following paragraphs we provide an evaluation of two powerful pieces of
software, which when combined together, have a rich potential to significantly facil-
itate the learning process and may also be very useful in further development. Sub-
sequently, we include a set of instructions of how this can be achieved and eventually
we deliver the results of our attempt.

2 Related Work

The demand for having flexible training tools in order to meet the needs of the emerg-
ing robotics market and the potential of virtual reality has been recognized quite a
long time ago. In [3], KUKA’s universal simulation environment, targeted especially
for educational purposes, has been presented in contrast to the other, device-specific
modelling software available at that time. Another virtual visualization environment,
which includes three robotic models and is suitable for teaching robotics at the uni-
versity level has been presented in [4]. A more contemporary approach, based on
Gazebo simulator has been demonstrated in [5]. The summary of the usage of sim-
ilar tools can be found in [6, 7], where the authors share their experiences gained
during the several years of educational activities. The reported results seem to be
satisfactory enough to still take on this subject.

3 Motivation and Background

As an interdisciplinary field of science, robotics requires from its adepts to acquire

an extremely wide range of knowledge in many differed areas in order to be able
to freely express their creativity. It is also worth mentioning, that robotic systems
are usually fairly complex, susceptible to damage or even dangerous when handled
improperly. They are rather expensive as well, which can often result in a limited
access to such devices. Those are only a few reasons why to look for many alternative
ways of targeting the challenges thrown by the world of modern robotics. A short
description of the tools of our choice is presented as follows.
V-REP and LabVIEW in the Service of Education 17

3.1 V-REP

V-REP (Virtual Robot Experimentation Platform) is a versatile, cross-platform, 3D

robotic simulation environment made by Coppelia Robotics [8, 9]. It can be used
for modelling simple robotic components such as sensors and actuators as well as
complex fully functional robotic systems.
The application provides forward and inverse kinematics calculations along with
the four real-time physics engines dedicated for dynamic calculations and simulation
of full objects interactions.
It supports various interchangeable CAD formats such as URDF, COLLADA,
DXF, 3DS, OBJ, STL for easier integration and can be bridged with various external
applications through several remote API’s (C/C++, Python, Java, Matlab, Octave,
Lua, Urbi) and the ROS interface. With its well designed, user friendly and customiz-
able interface it is a very good choice for presenting the basic principles of robotics
as well as designing and testing any robotic device.

3.2 LabVIEW

LabVIEW is a widely spread engineering platform from National Instruments [10].

It comes with a visual programming language named G, which basically means that
the program structure consists of a number of graphical blocks representing indi-
vidual instructions, all connected with wires to ensure a suitable work-flow. Such
an approach guarantees even inexperienced users, unfamiliar with the programming
principles to efficiently learn and operate through the environment. The applica-
tion supports a wide range of data acquisition and instrument control devices which
makes it a great tool for working with the real hardware.

4 Course Description

The main objective of the classes is to enable a group of high school students to
design, simulate and build a factual mobile robot: a Mars rover. By working on this
project, its participants familiarize themselves with a wide range of aspects of the
development process: from making the first design choices to the final testing of a
working machine.
As a part of the introduction, various instruments were presented to them as the
possible ways to solve the whole task: modelling tools, simulators, programming
environments, prototype boards and many others. In this article we focus on the two
aspects of the whole work: simulations and algorithms development.
One of the most important things in the whole process is the fact, that the course
is led by example and try. Students were provided with knowledge of how to begin.
18 M. Gawryszewski et al.

The rest is all about achieving their goals on their own with only a minimal assistance
and supervision coming from a teacher.

4.1 The Role of Selected Software

Harnessing V-REP and LabVIEW enables students to benefit from various, well-
established tools like inverse kinematics solver right out of the box. This means, that
they do not have to understand deeply all aspects of the given problem, e.g. inverse
kinematics. Considering less advanced high school attendees this might be crucial to
preserve motivation, as some problems might be worked around without decreasing
overall complexity of a project.
Up till now, V-REP has been used to demonstrate, how to model and simulate
a robot. The application does not only allow to conduct simulations of a designed
device, but also of the entire surrounding environment.
During the hands-on sessions, students have been creating models of mobile
robots to understand the possibilities and limitations of the whole process. Thanks to
built-in modules, they were able to easily simulate dynamics of their creations and
plan a path to follow, which included taking into account existing obstacles.
Two methods of implementing robot’s algorithms were demonstrated. The first
one was scripting in Lua language [11], which is an internal mechanism of V-REP.
Additionally, the originators of V-REP have established APIs that allow users
to create applications of any kind, which then are able to communicate with the
simulation and control its elements. Thanks to that, it is now possible to create control
algorithm in any modern programming language, but C++, Python, Java, Matlab and
LabVIEW are the preferred ones.
As for the second method, we have chosen V-REP’s LabVIEW interface, which
was originally designed by Peter Mačička [12]. Utilizing his work, we asked the
students to create their own programs to control simulated robots.

4.2 Modelling Basics

Modelling with V-REP involves dealing with three basic groups of elements [13]:

1. Scene objects—twelve types of construction blocks (e.g. shapes, joints, proxim-

ity, force/torque or vision sensors), that can be combined with each other to create
more complex designs
2. Calculation modules—five basic embedded algorithms: forward and inverse kine-
matics, path planning, collision detection, physics and dynamics, minimum dis-
tance calculation
V-REP and LabVIEW in the Service of Education 19

Fig. 1 V-REP’s default user interface running under Linux operating system (with a model of an
authentic Mars rover, built by the students at the Lodz University of Technology [14], imported to
the scene)

3. Control mechanisms—local and remote interfaces like: embedded scripts, plug-

ins, add-ons, remote APIs, ROS nodes and custom (client/server model based)
solutions, that can work simultaneously and cooperate with each other
The main program interface (Fig. 1) is clear and simple. By default, it consists of
standard toolbars offering the ability to access most of the features. A scene pane
which can display multiple views as well as custom user interfaces is available on
the right. On the left, there is a browser with thumbnails of all available models that
can be dragged into the scene and finally a tree view called Scene hierarchy, which
represents dependencies between all objects of the project.
When all elements are finally placed on the scene, they can be easily controlled
via the internal scripting language or provided remote APIs. To illustrate the func-
tionality more accurately, we can imagine creating a rotary joint on the scene and
then setting its velocity by referencing it via the Lua script.
Lua is described as very fast, powerful, yet simple and lightweight scripting lan-
guage ideal for rapid prototyping. In addition to procedural syntax, it provides meta-
mechanisms for implementing elements typical for object-oriented programming
languages, like classes and inheritance [15].
There are several types of embedded scripts that can be applied to the project. The
most important are those used when the simulation is running which divide into four
categories: Main, Child, Joint Control and Contact Callbacks. Then we can also use
the Customization, General callback and finally Dialog and Editor scripts. Below we
20 M. Gawryszewski et al.

provide a short example of a Child script, which demonstrates referencing objects in

V-REP.
Lua script—referencing simulation objects in V-REP

-- Handle of the left motor

leftMotor=simGetObjectHandle(‘‘bubbleRob_leftMotor’’)
-- Handle of the right motor
rightMotor=simGetObjectHandle(‘‘bubbleRob_rightMotor’’)

speed=minMaxSpeed[1]+(minMaxSpeed[2]-minMaxSpeed[1])

if (backUntilTime<simGetSimulationTime()) then
-- When in forward mode,
-- move forward at the desired speed
simSetJointTargetVelocity(leftMotor,speed)
simSetJointTargetVelocity(rightMotor,speed)
else
-- When in backward mode,
-- backup in a curve at reduced speed
simSetJointTargetVelocity(leftMotor,-speed/2)
simSetJointTargetVelocity(rightMotor,-speed/8)
end

(based on: Line following BubbleRob tutorial)

4.3 Interfacing V-REP with LabVIEW

There are many aspects in which the embedded scripting excels. The ease of the inte-
gration, inherent scalability, robustness and compatibility are just only a few of them.
However, considering the general purpose of the course, we decided to present the
remote APIs as well. Among others, LabVIEW seemed to be the best compromise
between the number of available features and intuitiveness.
The interfacing is extremely simple and poses absolutely no problem. On the
server side (in our case this means V-REP simulation), we just have to edit the inter-
nal script to set the available port, include a proper plug-in called
v_repExtRemoteApi.dll (the name may vary depending on the platform) and enable
remote APIs by calling simExtRemoteApiStart(). From the clients perspective, we
use a universal Call Library Function Node (indicated by arrow in Fig. 2). We just
have to make sure, that proper Dynamic-Link Library (remoteApi.dll) is set in the
block properties and select the adequate parameters (indicated by arrow in Fig. 3). A
short tutorial with a wide range of examples has been provided by the author of the
interface [12].
V-REP and LabVIEW in the Service of Education 21

Fig. 2 SubVI handling joint velocity

Fig. 3 Call Library Function Node properties

4.4 Workflow and Educational Values

At the very beginning of the course a modelling software has been presented to
students. It is also worth mentioning, that in addition to V-REP, the presentation
included Autodesk Inventor, which will later be used to create their own models.
The next step was the introduction to LabVIEW. The lecture included environ-
ment basics such as: the distinction between graphs and front panels, basic data types,
kinds of controls, indicators and constants as well as execution structures i.e. Loops
and Cases. Participants learned how to create simple programs, according to para-
digms and patterns used in LabVIEW.
22 M. Gawryszewski et al.

The following four meetings, which are the main scope of this article, were all
about modelling in V-REP and programming in LabVIEW.
During the first one, students were shown the capabilities of V-REP, the basic
concepts, and how to build simulated objects and elements of the environment. They
gained knowledge about its distributed control architecture and the advantages guar-
anteed by this kind of solution (e.g. portability and scalability). Finally, all of them
learned how to associate scripts with individual objects. The possibility of importing
prototypes created in the above mentioned Autodesk Inventor has also been demon-
strated as shown in Fig. 1.
The second and the third meeting were focused on modelling the BubbleRob
robot, which is a part of the official V-REP’s tutorial [16]. All the students, divided
into the groups of two members each, had to go through the whole process of creating
a simulation of a simple mobile robot step-by-step, which definitely allowed them to
improve their practical skills. As a result, they were able to see for themselves, how
robotic models are being designed and how to resolve typical problems.
During the fourth meeting, two previously introduced tools were connected
together: we used LabVIEW to create control application for the robot simulated
in V-REP. This has happened in the two stages. The first one was to set up an exam-
ple control application, which shows how to exchange data in both directions. The
difference between all available interfaces has been outlined emphasizing regular
and remote APIs. The second one involved students in creating their own programs,
to control robot in V-REP (Fig. 4).

Fig. 4 “BubbleRob” robot (back window), controlled by LabVIEW application (front window)
V-REP and LabVIEW in the Service of Education 23

What is important, as LabVIEW was selected as an environment to create algo-

rithms for the real robot, the students were able to test their programs with the sim-
ulation and use it in a real-world application afterwards. They also had a chance to
exploit a variety of Toolkits available for LabVIEW, including those dedicated espe-
cially to robotic applications.
Given the knowledge, that the only thing that differs is a low level communication
layer (i.e. function calls) and that it should be replaced by the one appropriate for the
selected hardware, they were finally enabled to use the same program with both, the
simulated and the actual robot.
The next step of our course is to show the basics of electronics with Arduino [17],
which serves as a controller of robots intended for the course. After it is finished,
students will be ready to start making real robots, as they gain enough skills to know
how to start and they have already seen on simulation that their models can work.

4.5 Survey

A couple of weeks after the last meeting focused on LabVIEW and V-REP, we asked
our students to provide feedback about the workshops. Our goal was to understand,
the students’ opinion about the form of the classes—if they are satisfied with it and
the provided content as well as if they noticed any increase in the level of their knowl-
edge.
Eleven participants completed the questionnaiare. There were 9 statements (col-
lected in Table 1) describing the workshop and the responses were formatted in a
5-point Likert scale: answer 1 means: I strongly disagree and 5: I strongly agree.
In general, our work as tutors has been well evaluated, as all the students indicated
that tutors were helpful and willing to give explanations if something was unclear.
They also agreed that the knowledge was presented in an explicit and understandable
manner as shown in Fig. 5, plot: “Questions 1, 2”.
On the other hand, the presented content did not fit their needs perfectly, as 4
students marked answer 3 which might be considered as a slightly negative opinion
(Fig. 5, plot: “Questions 3, 4”).
For the vast majority of the students our workshops were interesting (9 out of 11)
and most of them declared to take part in more advanced courses, as an extension of
this one (Fig. 6, plot: “Questions 5, 6, 7”).
We also asked about the programming skills, and 9 participants claimed that they
have improved their skills as a result of the course. Unfortunately, the other two
disagreed with this statement.
The question about the further use of the acquired knowledge seems to be unre-
solved: there are 6 students which claim that they see benefits from the workshop as
a way to apply their knowledge in their own projects (Fig. 6, plot: “Question 8, 9”),
yet still 5 people has no clear idea or do not see any profit.
24 M. Gawryszewski et al.

Table 1 Questions
No. Topic Question
1 Tutors Tutors were willing to answer
questions
2 Knowledge was presented in
an explicit and understandable
manner
3 Materials Course materials used were
useful
4 Course materials were easy to
understand
5 Course form Course was interesting
6 Course form allows me to take
part actively
7 My programming skills raised
through participation in the
course
8 Future I would like to take part in
more advanced edition of the
course in the future
9 Knowledge gained during
course will allow me to
implement my own ideas

This was the first iteration of workshops and this survey showed us some space
for improvements. We indicated three crucial areas which should be reworked before
the next edition:
∙ the content—the course was based on step-by-step, written instructions, which
were generally considered as useful, but there is some space for refinement such
as dividing the tutorial into smaller parts or introducing better balanced level of
difficulty.
∙ programming skills—part that introduces LabVIEW shall be reworked to another
form.
∙ students’ ideas—the main goal of the entire project is to build a Mars rover. It is
also very important to create a strong the base to develop students’ own ideas on
basis of this project, and this needs to be emphasized.

5 Final Remarks

The most important aim of our project was to introduce fairly complex robotic prob-
lems to young, inexperienced people and additionally solve these problems in an
V-REP and LabVIEW in the Service of Education 25

Fig. 5 Histogram of answers the the questions: 1—“Tutors were willing to answer questions.”
(left plot, blue), 2—“Knowledge was presented in an explicit and understandable manner.” (left
plot, green), 3—“Course materials used were useful.” (right plot, blue), 4—“Course materials were
easy to understand.” (right plot, green)

Fig. 6 Histogram of answers the the questions: 5—“Course was interesting.” (left plot, blue), 6—
“Course form allows me to take part actively.” (left plot, green), 7—“My programming skills raised
through participation in the course.” (left plot, red), 8—“I would like to take part in more advanced
edition of the course in the future.” (right plot, blue), 9—“Knowledge gained during course will
allow me to implement my own ideas.” (right plot, green)
26 M. Gawryszewski et al.

interesting manner with the available (possibly free) tools. These issues are present
in most projects dealing with kids and teenagers [18, 19].
Among several simulation environments like Webots [20] or Gazebo [21] we have
chosen V-REP as the most engaging and accessible. There are at least five great
premises in favour of this decision:
1. Usability—representation of the scene is done in a similar way as in any other
modelling software, which students are already familiar with way before they get
introduced to V-REP
2. Efficiency—the vast majority of work can be done by simply using drag-and-drop
functionality which considerably improves productivity
3. Scalability and portability—a great variety of ways to add logic to the scene: by
internal scripts or external applications over standardized communication chan-
nels
4. Compatibility—possibility to import/export objects in various CAD formats, so
the elements in simulation can be based on ones created in modelling software
5. Simulation capabilities—it is important to see how the mass and inertia of the
robot influences its ability to move or perform other tasks and V-REP provides
four different models of dynamics, all of them available free of charge for educa-
tional purposes
Our goal was to deliver tools and methods that are fun, easy to use and can be
utilized with limited training, as our final goal is to build a robot, not to learn new
programming environments.
We believe, that our approach will finally succeed, as participants of the course
are now able to build simulations and control algorithms for their own robots with a
very limited supervision.
We also hope, that our experience will be helpful for other tutors that may face
similar issues.

Acknowledgments The authors would like to thank Mr Kacper Andrzejczak for his effort during
the lessons.

References

1. Industrial Robotics Market by Type (Articulated, Cartesian, SCARA, Cylindrical, Par-

allel), Application (Automotive, Electrical and Electronics, and Metal and Machin-
ery), Component (Controller, Sensors, Drive), and Geography—Analysis & Forecast
to 2022. http://marketsandmarkets.com, http://www.marketsandmarkets.com/Market-Reports/
Industrial-Robotics-Market-643.html
2. Grand View Research: Service robotics market analysis by application (Professional, Per-
sonal) and segment forecasts To 2020. http://www.grandviewresearch.com/industry-analysis/
service-robotics-industry
3. Sett, A., Vollmann, K.: Computer based robot training in a virtual environment. In: 2002 IEEE
International Conference on Industrial Technology, Bangkok, Thailand (2002)
V-REP and LabVIEW in the Service of Education 27

4. Yang, X., Zhao, Y., Wu, W., Wang, H.: Virtual reality based robotics learning system. In:
Proceedings of the IEEE International Conference on Automation and Logistics, Qingdao,
China, Sept 2008
5. Canas, J.M., Martn, L., Vega, J.: Innovating in robotics education with Gazebo simulator and
JdeRobot framework. XII Congreso Universitario de Innovacion Educativa en las Ensenanzas
Tecnicas (2014)
6. Lopez-Nicolas, G., Romeo, A., Guerrero, J.J.: Simulation tools for active learning in robot
control and programming. In: EAEEIE Annual Conference (2009)
7. Jung, S.: Experiences in developing an experimental robotics course program for undergradu-
ate education. IEEE Trans. Educ. 56(1) (2013)
8. V-REP project website. http://www.coppeliarobotics.com
9. Rohmer, E., Singh, S.P.N., Freese, M.: V-REP: a versatile and scalable robot simulation frame-
work. In: Intelligent Robots and Systems (IROS), 2013 IEEE/RSJ International Conference on
Robots and Systems (2013)
10. LabVIEW product website. http://www.ni.com/labview
11. Lua project website. http://www.lua.org
12. LabVIEW interface for V-REP. http://www.coppeliarobotics.com/contributions/labview.zip
13. V-REP’s overview presentation. http://www.coppeliarobotics.com/v-repOverview
Presentation.pdf
14. Raptors (a Mars Rover project) website. http://raptors.p.lodz.pl/
15. Lerusalimschy, R., de Figueiredo, L.H., Celes Filho, W.: Lua an extensible extension language.
Softw. Pract. Exp. 26(6), 635–652 (1996)
16. BubbleRob tutorial. http://www.coppeliarobotics.com/helpFiles/en/bubbleRobTutorial.htm
17. Arduino Mega 2560 product website. https://www.arduino.cc/en/Main/ArduinoBoard
Mega2560
18. Petrovic, P.: Having Fun with Learning Robots. In: Proceedings of the 3rd International Confer-
ence on Robotics in Education, RiE2012. September 2012, pp. 105–112. MatfyzPress, Czech
Republic (2012)
19. Swenson, J., Danahy, E.: Examining influences on the evolution of design ideas in a first-year
robotics project. In: Proceedings of 4th International Workshop Teaching Robotics, Teaching
with Robotics & 5th International Conference Robotics in Education, Padova (Italy), 18 Jul
2014
20. Webots project website. https://www.cyberbotics.com/overview
21. Gazebo project website. http://wiki.ros.org/gazebo
Applied Social Robotics—Building
Interactive Robots with LEGO Mindstorms

Andreas Kipp and Sebastian Schneider

Abstract Teaching Social Robotics is a requiring and challenging task due to the
interdisciplinary of this research field. We think that it can not be taught in a solely
theoretical manner. To help students to gain more interest in the topic and to foster
their curiosity we restructured a paper club like lecture to create a bridge between
a theoretical topic and practical applications. This paper describes our approach to
create a lecture covering theory, methods and how to transfer those to applied infor-
matics. Described is the given theoretical input and how students learn to transfer
these to a robot they build on their own. We also evaluate how the new structure was
accepted and what lessons can be learned for this lecture style.

Keywords Social robotics ⋅ Educational toys ⋅ Applied techniques

1 Introduction

In a not so distant future robots might be an integral part of our daily life. The com-
mission for innovation and research of the German government has prognosticated
that the consumer and service market will be the most growing economy in the
next years.1 Hence, there is a requirement for robots that are capable of interacting
with people and working in domains that were up to now staffed by humans. Thus,
engineers designing and building new robots need a broad knowledge from differ-
ent disciplines (i.e. sociology, psychology, computer science). This interdisciplinary
demand is currently brought together in the research on social robotics. However,
there is no common curricula for students that are interested in this field of research

1 http://www.e-fi.de/fileadmin/Gutachten_2016/EFI_Gutachten_2016.pdf, visited 03/11/2016.

A. Kipp (✉) ⋅ S. Schneider

Bielefeld University, 33619 Bielefeld, Germany
e-mail: [email protected]
URL: http://aiweb.techfak.uni-bielefeld.de/
S. Schneider
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 29

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_3
30 A. Kipp and S. Schneider

yet. We see that early stage researchers coming from an engineering background
have to struggle with many different theories, publications and methodologies.
In the past years we have offered courses to teach social robotics. Students read
different publications connected to one of the sub areas of social robots. Those areas
were i.e. robot design, emotion expression, anthropomorphism, applications and
evaluation methods. The students were preparing on of the seminar sessions includ-
ing a presentation of the publication, a discussion and a handout. We encountered
that the seminar sessions often did not match the actual details of the topics. Thus,
the association between different topics and the general scope of the seminar were
difficult to grasp for the students and obstructs students to continue with this line of
research. We encountered that a theoretical approach to teach social robotics is not
sufficient to understand the concepts and a more hands-on approach is needed.
To overcome the difficulty of the materials and the complex access to knowledge
about social robotics, we restructured the seminar in a modern fashion. We have
changed the seminar to a course that covers topics on social robotics in an interactive
lecture like structure. Each lecture is accompanied by an exercise where a given
technique or method is directly integrated into a practical hands-on part. A group
project concludes the lecture part by applying all learned elements into one practical
and functional social robot. Thus, we wanted to teach how the theoretical input can
be explored using a practical approach.
For the practical part the LEGO© Mindstorms EV32 educational set is used,
allowing students to easily create robotic systems that can act and be perceived
as small social entities. The robot scenario is based on the Tamagotchi devel-
oped by Bandai. This small toy uses different needs and actions to mimic a small
autonomously behaving pet the user has to care about. The idea for the project was
to take basic elements and behaviors from the Tamagotchi and let the students under-
stand and transfer those to a socially perceivable LEGO robot.
In this work we want to present our concept for teaching social robotics on a uni-
versity level. We will present our lecture structure, content, and methods we have
used. At last, we will give a discussion about our experiences teaching social robot-
ics, the effectiveness of our lecture paradigm, and reflect on comments and feedback
from our students.

2 Related Courses and Teaching Methods

In this section we want to describe how other universities are teaching social robotics.
We have searched for courses using the keywords ‘human-robot interaction’, ‘social
robotics’ and ‘lecture’.
The social robots lab in Freiburg offers a seminar on social robotics.3 Student’s
learn how to conduct a literature review, read papers, and learn about state-of-the-

2 http://education.lego.com/MINDSTORMS-EV3.

3
http://srl.informatik.uni-freiburg.de/ss15seminarsocialrobotics, visited 03/10/2016.
Applied Social Robotics—Building Interactive Robots with LEGO Mindstorms 31

art methods. Finally, they give a presentation of their results during a block seminar
and write a summary about a paper. The content of the paper were mostly tracking,
motion, and path planning.
The GeorgiaTech has offered a course on HRI.4 The lecture covers a wide range
of topics on the emergence of social intelligence and the state-of-the-art on build-
ing systems with social intelligence (e.g. Anthropomorphism, Embodiment, Exper-
imental Design, Intentional Action, Collaboration, Teamwork, Turn taking, Dialog,
Emotional Intelligence, Social Learning, Telepresence, Assistance). The lecture is
accompanied by a final group project.
The Indiana University offered a course on HRI Design.5 Topics were Classifying
HRI, Evaluating HRI, Autonomy and Perception, Interfaces, Enhancing Interfaces,
Robot Teams, Museum Robot, and Search and Rescue. The lecture was followed
by a final project. During the course, students had to complete readings, quizzes,
and labs on how to design a HRI system. Prerequisite were programming knowledge
in C and JAVA. The assignments were a discussion where students have to submit
half a page summary of the class paper, pros and cons, and questions. Quizzes cover
the reading material using multiple choice and true-false questions. Lab assignment
were conducted on a mobile robot outside of class.
At last we want to mention the lecture Principles of Human-Robot Interac-
tion from Carnegie Mellon University6 . Topics are: Social Robotics, Multi-modal,
human-robot communication, Human-robot interaction architectures, Sensors and
perception for HRI, Museum robotics, Educational robotics, Urban Search and Res-
cue, and Quality of Life Technologies. Students have to attend the course, read
papers, answer questions related to the papers, and do a semester-long group projects.
All presented courses, except the seminar taught at the University of Freiburg,
have a similar structure and similar topics. However, the description of the topics
for the course are still a bit broad. This makes it hard to compare the content of i.e.
‘Autonomy and Perception’ to “Sensors and Perception for HRI”. The difficulty to
distinguish the different topics of social robotics reflects the interdisciplinary of this
research field. Hence, all lectures use different books or publications as the reading
material for the students (there is only one publication that all of them are using as
a Ref. [1]). This leads to the fact that students studying at different universities will
read different references for the same subjects. We do not think that every seminar
should have the same content at every university. The diversity on topics is important
to give students the choice to think about to which university they apply. However,
we see a demand to define a core set of topics that should be taught in a introduction
course on social robotics.
Therefore, we want to report how we went from a reading-based seminar to a
lecture based-hands-on course. Using this process we generated ideas how to capture
the different topics of social robotics into a new curricula.

4 http://www.cc.gatech.edu/~athomaz/classes/CS7633-HRI/, visited 03/10/2016.

5https://www.rose-hulman.edu/~berry123/Courses/HRI/HRI%20Syllabus.pdf, visited 03/10/2016.
6
http://www.cs.cmu.edu/~illah/ri899.html, visited 03/10/2016.
32 A. Kipp and S. Schneider

3 Previous Courses

In the past years we offered several courses for students to start gathering knowledge
in the field of social robotics. Most of the courses were held in a paper club style to
introduce to the different subtopics of social robotics.
Throughout the lecture each student had to prepare a given topic and to present it
to its fellow students. To fulfill the course the topic had to be documented in a written
style and handed in at the end of the term. In the first session of the course details
on the structure of the lecture as well as a basic introduction on the topic of social
robots were provided. Afterwards the different subtopics got briefly introduced and
distributed between the students. The common topics used for the paper club lecture
can be seen in Table 1.
Each student prepared their topic themselves and presented it in the paper club.
The presentation concluded with a discussion allowing fellow students to ask ques-
tions on the topics or to discuss possible conjunctions with following topics. After-
wards the students wrote a short documentation for their topic to be handed in at the
end of the term. The documentation was the main element to successfully complete
the course. There was no written or oral exam due to the fact, that the course was
part of an individually chosen block of the students study regulations.
One big problem with this structure was the decreasing motivation to participate
in later presentations, especially for those who had already presented their topic.
The seminar started with around 14 students in the first sessions. Throughout the
term this number decreased to about five to six students being present for the last
presentation.
Another problem was the lack of fortification of knowledge and the transfer to
other topics. Due to the fact that the presented information where not prompted after-
wards in an exam style, many students did not tend to foster their interest.

Table 1 Topics covered in the paper club on the topic of social robotics
Topic
Anthropomorphism [2]
Forms and function of robots [3]
Uncanny valley [4]
Relations between forms and robots [5]
Perception of behaviors [6]
Attitudes towards robots [7]
Mental models for robots [8]
Socially assistive robotics [9]
Evaluation of HRI [10]
Long-term interaction with social robots [11]
Emotion models [12]
Applied Social Robotics—Building Interactive Robots with LEGO Mindstorms 33

4 The Idea with a Goal: Applied Social Robotics

Based on the experiences from the previous course, we decided to change the struc-
ture for the next term and to help students in finding more meaningfulness. We
wanted to foster the transfer of gathered information and knowledge into a more
practical approach.
To achieve this we decided to build a new course structure around a project in
which students themselves build small social robots. Because the course is open for
Bachelor students that had just started their university career as well as for Master
students that already had come in contact with more complex topics, we decided to
use the LEGO Mindstorms platform. This technology allows, even without deeper
technological understanding, to easily create small robotic systems. Nevertheless it
also offers a wide range of exploration and artistic freedom for users with more exper-
tise.
To provide a base scenario we decided to let the students create robots behaving
equally to the Tamagotchi toy pet (see Fig. 1). These small devices mimic behaviors
of pets by demanding attention and care (see Sect. 5.1). Hence, our teaching idea was
to help students understand how the Tamagotchi can be seen as a social entity, how
its behaviors can be described, and how such elements can be transferred to small
robots build with LEGO.
Our overall goal was to evaluate how students deal with the new structure. We
proposed that creating a conjunction between a dry theory and a practical approach

Fig. 1 The Tamagotchi

device. A small child’s toy
capable to mimic behaviors
to foster engagement. It can
request attention and care
due to visual and audio
feedback
34 A. Kipp and S. Schneider

can help students to gain an easier access to such a topic. Additionally, such a course
structure can promote creativity and result in an interesting outcome concerning the
created robots.

5 Structure of the Course

For the structure of the course we decided to use three different parts: a lecture on
topics of social robotics, an exercise to introduce the technological part and give
some transfer ideas between topics and technology, and a project in the end of the
term with focus on applying learned elements into a real robot behaving in a social
manner.
The first part is a lecture. Throughout each session the lecture covers a given topic
(see Table 2) that is presented by a lecturer. At the beginning of a new topic the last
topic is shortly reviewed and students have to solve simple tasks like answering ques-
tions. To keep the students interested we used some interactive teaching methods that
should encourage students to directly work with the knowledge. For example we use
a method called Mumble Time. This method helps students to exchange their knowl-
edge on the current topic with a partner and to collect ideas for a discussions with
the whole group. Students that don’t like to discuss in a broad group can exchange
ideas with the partner and let them forward these to the group. The techniques used
are based on smaller and bigger groups and work as well as working with shared
contents of the presented topics. At the end of each lesson the topic is summed up
and a discussion to ask questions, on possible ideas on how to apply the topic, or how
the conjunction to a follow up topic can be is started. The lesson concludes with an
impulse outlook for the following exercise and the upcoming topic.
The second part is the exercise. These exercises mostly cover how the technol-
ogy of the LEGO Mindstorms can be used to create small robotic devices. Here
the programming part gets explained as well as direct practical realizations by the
students. As programming language we selected JAVA and the LEJOS7 API. This
API allows to develop control algorithms for the LEGO EV3 control brick with easy
to understand elements. Throughout each exercise students work together in groups
with up to three people and try to solve given tasks. These tasks cover building small
moving vehicles or more complex tasks like creating emotions with the given parts.
The third and most practical part is the group project at the end of the term. Each
group consists of up to four students. For every group a complete LEGO Mindstorm
education set is issued to create their own robot. As project goal we defined that each
robot needs to include behaviors according to those of the Tamagotchis. For the full
project every group has time span of six weeks. Throughout this time frame the group
has to create their own robot, program its behaviors, and to create a documentation
of the project process. In the last session of the term, the groups have to present their

7
http://www.lejos.org/.
Applied Social Robotics—Building Interactive Robots with LEGO Mindstorms 35

Table 2 Topics to be covered in the lecture part of the seminar. Each topic is presented by a lecturer.
Throughout the lecture students use group work to discover and discuss the different topics
Topic Content and lecture Teaching method
Introduction to the course Information about the lecture, –
exercise and the group project
Introduction to social robotics Outlook on research, MumbleTime, Open
platforms, What makes a robot discussions
social? [1, 13]
Social agents and control Introduction to agents, models From mumble time to group
structures for agents, Definition of presentations
control structures [14]
Anthropomorphism and social What means Group work and cross
actors Anthropomorphism? How do presentation
humans perceive robots and
agents? What means “to act
socially”? [2, 15]
Form and design for robots From technical to human-like Mumble time and
robots? Design choices for collection/discussion of good,
robots, Examples of research bad, ugly designs
platforms [4, 5]
Internal models and emotions How to model behaviors, Hands on: How to build
controlling internal states, emotional elements for robots
How to express emotions? [12,
16, 17]
Applications for social robotics Information on ongoing Group posters and presentation
research, already applied
robotic systems
Studies: How to for evaluation Basics about evaluations, What Group discussion about what
is a research question? How to to evaluate based on the
create a study? [18, 19] upcoming project

robot by providing information on their idea behind their robot, how they build and
developed it, and give a prospect about possible enhancements or additions.
To fulfill the course each group has to prepare a documentation containing the
information provided in the final presentation as well as a documentation of the
building and programming process.

5.1 Tamagotchi: Behaviors of a Small Toy

To provide a scenario that defines a broad range of behaviors but leaves space for
individual ideas, we decided to use the Tamagotchi toy device and its features as
basis.
36 A. Kipp and S. Schneider

The Tamagotchi published by Bandai in the 1990th is a small digital toy that was
created in Japan by Akihiro Yokoi. The small egg like toy has a digital display and
three buttons. The toy is programmed to mimic behaviors of a small pet. With the
buttons the user can manipulate and interact with the digital pet. The pet itself is
based on simple wishes and desires. It needs to be fed with different food, cleaned,
disciplined, cheered, or cared about in case it got ill. With the different buttons the
user can initiate different actions according to the needs of the pet. The goal is to
understand the different needs and to satisfy them. If not cared about correctly the
pet may die and a new pet needs to be bred.
Throughout the lecture the Tamagotchi is used to compare topics of social entities
with the behaviors of the toy. For this students got introduced to the Tamagotchi and
were advised to understand how the digital pet works. Our goal was to teach stu-
dents how to transfer the Tamagotchis behaviors into ideas for a social LEGO robot.
Based on understanding the toys background the students should reverse engineer
the behaviors and then transfer them to their robots. For the project and the trans-
fer we focused on the behaviors concerning feeding, cleaning, cheering, and taking
care.
Each group could themselves decide how such behaviors are expressed with the
available parts of the provided set. We only advised each group that their ideas should
be understandable by other persons that only interact with the robot, but who had not
programmed the behaviors themselves.

5.2 The Group Project and the Results

For the project phase a total of six weeks was determined. In these weeks the student
groups should think about how they could create the Tamagotchis behaviors, how
to build their robot, and how to program the control mechanisms. Each team was
free to explore how the given LEGO elements can be used and how their ideas could
come to life. Every group was advise to document the steps taken from starting with
ideas until the final robot was created. This should help to generate content for the
documentation to be handed in afterwards.
Four student groups with up to three students were formed for the project phase.
Each group got one full set of LEGO Mindstorms. The sets could be taken home and
there was no need to bring them in or to only work in predefined slots. This allowed
the groups to work freely on their own behalf. Before the final presentation all groups
were randomly visited by the course supervisors to get an idea on how the groups
come along. Also each group was free to ask the supervisors for help.
For the final presentation each group created a small digital presentation men-
tioning the ideas on how they approached the project, how they build their robot,
and how they implemented the requested behaviors as well as their own. Addition-
ally they should mention what problems occurred throughout the free working time.
After all groups had presented their LEGO robot, the different robots were demon-
Applied Social Robotics—Building Interactive Robots with LEGO Mindstorms 37

Fig. 2 Two examples of the different robots created by the project groups. Image (a) shows robot
duck with ears to show emotions. Image (b) shows a dog like pet with eyes presented on the display

strated between the project groups. In this live demo the different behaviors were
shown and fellow students could test the robots themselves and asks questions.
The resulting robots were quiet impressive and fully functional. Every group was
able to create robots that showed the intended behaviors. Some groups focused on
more complex technique robots while other focused more on simple movements with
most impact on understanding the behaviors. To give an example one group created a
robot that looks like a duck (see Fig. 2a). The robot is capable to move its ears up and
down. This feature is used to show the emotional state of the robot. Whenever the
robot is sad the ears are moving down. For the predefined states this feature is used to
support the robots needs, for instance if the robot feels dirty and needs to be cleaned.
This emotional state is supported by using audio files in the corresponding situation.
The behavior of feeding / eating is supported by opening the mouth and by chewing
whenever a given item is placed inside the mouth and therefore above the placed light
sensor. In total the robot duck was programmed by using a subsumption architecture.
Needs and beliefs are predefined and get selected according to values stored in the
background. The designed and implemented behaviors are understandable to users
and fulfill the course definitions. The second example is a dog like robot called little
dog (see Fig. 2b). It can turn around, move its head, and use audio files to express
its mood. One interesting feature of this robot is the cooperation needed with the
human. For example if the robot gets tired, it uses audio and head rotation paired
with eyes shown on the display to announce this state. The user then needs to flip
the robot to the side. By using the gyro sensor this state gets detected and the robot
starts sleeping and snoring. A video showing the behaviors of the little dog can be
found at the CITEC video channel.8

8
CITEC, https://www.youtube.com/citecbielefeld.
38 A. Kipp and S. Schneider

5.3 Evaluation of the Course

The Technical Faculty of Bielefeld University offers an evaluation for each course by
providing questionnaires. These can be handed out to the students and will be eval-
uated by staff from the faculty itself. The results can then be compared between all
provided courses. Unfortunately these questionnaires are optional and were not con-
ducted for the previous lecture style. Nevertheless from colleagues we were assured
that the amount of students decreases until the last sessions and that the motivation
was not that high.
For our new lecture we handed out the questionnaire on the date of the presen-
tation session. In total eleven students participated in the evaluation. We had three
bachelor students, five master students, and three Ph.D students.
The results showed a positive response to the new designed course structure. One
question concerns why students visit the course (see Fig. 3a). For this question mul-
tiple answers were possible. Ten students marked their interest on the topic as a
reason to participate. Three students liked the idea to work in a more practical man-
ner. Nearly all students participated in every session until the final presentation. This
shows that we could keep the students interested on the topic. For the lecture part
the slides used were marked as detailed and interesting (see Fig. 3b). Nevertheless an
additional script would be appreciated and could help foster the transfer of gathered
knowledge.
We tried to apply now teaching methods to help those students that normally
would not discuss in larger groups. This commitment was positively assessed by the
students group (see Fig. 4a).
For the practical part the votes showed that students liked to be creative and to
transfer the topics to the LEGO robot. The results also show that the new structure
fosters interest on the topic of social robotics beyond the course (see Fig. 4b).
Additionally to the questionnaire we conducted an open discussion round at the
end of the presentation session. We used a method called Five Finger Feedback,
allowing students to give one feedback concerning a scope matching a finger of the

(a) (b)

Fig. 3 Evaluation results provided by the technical faculty. a Why do you participate. b Did the
provided material support the content of the lecture?
Applied Social Robotics—Building Interactive Robots with LEGO Mindstorms 39

(a) (b)

Fig. 4 Evaluation results provided by the technical faculty. a The lecturers are interested in teach-
ing the topic. b Students are actively involved

hand: What was good? (thumb), What needs to be enhanced?, What was not good?,
What do you take with you?, and What was too short? (little finger).
From the verbal feedback we can conclude that the new structure is very helpful
to transfer theories to practical applications. It helps students to understand the asso-
ciations between topics. Students reported that they liked to get a psychological point
of view on computer science which was new and interesting for them. For ourselves
we learned, that we need to provide more time for the exercises to foster creativity.
Also we may need to hand out a script with more details on the given topics. Another
point mentioned was the wish to deploy more LEGO sets to allow smaller groups
and therefore to build more robots.

6 Lessons Learned

Retrospective we are happy with the new structure and we are glad we switched to
the more practical part. This was our first approach to this new type and we learned
much about what else could be enhanced. We will work on the parts mentioned in the
evaluation and given by the students feedback to make the seminar more interesting
and understandable. Also we will extend the number of LEGO sets to increase the
number of further students. Additionally, we will define some criteria to rate the
results and to evaluate if the provided lecture style and the given specifications match
the outcome.
In general for teaching a topic that is complex and that combines many subtopics,
the idea to create a bridge between theory and practical applications offers a big ben-
efit for both, students and lecturers. With the direct transfer from lecture to exercise
students can more easily adapt to gathered knowledge. Also the connection between
subtopics becomes more clear. At some point for our teaching method this conjunc-
tion resulted in interesting discussions. This also helps us, the lecturers, to question
the theory ourselves.
40 A. Kipp and S. Schneider

Nevertheless a practical part can help to understand theory more easily, the lecture
part should also offer a good proportion of knowledge. This lecture part should also
offer some impulses for the students to employ themselves even more and to create
more transfer ideas for the practical parts and even other related topics.

Acknowledgments This research/work was supported by the Cluster of Excellence Cognitive

Interaction Technology CITEC (EXC 277) at Bielefeld University, which is funded by the German
Research Foundation (DFG).

References

1. Fong, T., et al.: A survey of socially interactive robots. Robot. Auton. Syst. 42(3), 143166
(2003)
2. Epley, N., et al.: On seeing human: a three-factor theory of anthropomorphism. Psychol. Rev.
114(4), 864 (2007)
3. DiSalvo, C.F., et al.: All robots are not created equal: the design and perception of humanoid
robot heads. In: Proceedings of the 4th Conference on Designing Interactive Systems:
Processes, Practices, Methods, and Techniques, pp. 321–326. ACM (2002)
4. Mori, M., et al.: The uncanny valley [from the field]. Robot. Autom. Mag. IEEE 19(2), 98–100
(2012)
5. Powers, A., et al.: Matching robot appearance and behavior to tasks to im- prove human-robot
cooperation. Human-Computer Interaction Institute, p. 105 (2003)
6. Moshkina, L., et al.: Human perspective on affective robotic behavior: a longitudinal study.
In: 2005 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2005 (IROS
2005), pp. 1444–1451. IEEE (2005)
7. Bartneck, C., et al.: The influence of peoples culture and prior experiences with aibo on their
attitude towards robots. Ai Soc. 21(1–2), 217230 (2007)
8. Lee, S.-l., et al.: Human mental models of humanoid robots. In: Proceedings of the 2005 IEEE
International Conference on Robotics and Automation, 2005. ICRA 2005, pp. 2767–2772.
IEEE (2005)
9. Feil-Seifer, D., et al.: Defining socially assistive robotics. In: 9th International Conference on
Rehabilitation Robotics, 2005, ICORR 2005, pp. 465–468. IEEE (2005)
10. Burghart, C., et al.: Evaluation criteria for human robot interaction. In: Companions: Hard
Problems and Open Challenges in Robot-Human Interaction, p. 23 (2005)
11. Leite, I., et al.: Social robots for long-term interaction: a survey. Int. J. Soc. Robot. 5(2), 291308
(2013)
12. Picard, R.W., et al.: Affective Computing, vol. 252, ch. 2. MIT press, Cambridge (1997)
13. Dautenhahn, K.: Socially intelligent robots: dimensions of human robot interaction. Philos.
Trans. R. Soc. Lond. B Biol. Sci. 362(1480), 679704 (2007)
14. Russell, S., et al.: A modern approach. In: Artificial Intelligence, vol. 25, p. 34. Prentice- Hall,
Egnlewood Cliffs (1995)
15. Reeves, B., et al.: The media equation: how people treat computers, television,? new media
like real people? places. Comput. Math. Appl. 5(33), 128 (1997)
16. Norman, D.A.: Emotional design: Why we love (or hate) everyday things. In: Basic books
(2005), Chap. 6
17. Marsella, S., et al.: Computational models of emotion. In: A Blueprint for Affective
Computing—A Sourcebook and Manual, pp. 2146 (2010)
18. Bartneck, C., et al.: Measurement instruments for the anthropomorphism, animacy, likeability,
perceived intelligence, and perceived safety of robots. Int. J. Soc. Robot. 1(1), 7181 (2009)
19. Feil-Seifer, D., et al.: Benchmarks for evaluating socially assistive robotics. Interact. Stud. 8(3),
423439 (2007)
Offering Multiple Entry-Points into STEM
for Young People

Wilfried Lepuschitz, Gottfried Koppensteiner and Munir Merdan

Abstract Enrollment in the STEM fields (science, technology, engineering and

math) is not keeping pace with the need. Recent reports indicate a decrease in the
number of graduates from STEM fields and a shortage on the job market. Consid-
ering these issues, particular attention has been paid developing innovative methods
and tools for improved teaching of STEM themes. This work presents an approach
involving multiple entry points for young people to engage in the STEM fields. This
approach is manifested in the non-profit association Practical Robotics Institute Aus-
tria (PRIA) with its activities designed to fill STEM gaps in the Austrian education
system and to bring innovative engagement that cannot be found in the classrooms.
Thus, STEM literacy is fostered as well as the development of systems thinking,
problem solving, and teamwork skills.

Keywords STEM ⋅ After-school program ⋅ Camps ⋅ Workshops ⋅ Non-profit asso-

ciation ⋅ Research ⋅ Education

1 Introduction

Science, Technology, Engineering, and Mathematics (STEM) education is a recently

emerged paradigm that focuses mostly on science and mathematics disciplines, but
also includes technology and engineering [1]. STEM education has evolved into a
meta-discipline and thus an integrated effort that removes the traditional barriers
between these subjects. Instead it focuses on innovation and the applied process of
designing solutions for complex contextual problems using current tools and tech-

W. Lepuschitz ⋅ G. Koppensteiner ⋅ M. Merdan (✉)

Practical Robotics Institute Austria, Wexstrasse 19-23, A-1200, Vienna, Austria
e-mail: [email protected]
URL: http://www.pria.at
W. Lepuschitz
e-mail: [email protected]
G. Koppensteiner
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 41

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_4
42 W. Lepuschitz et al.

nologies [2]. It shall help individuals in developing different strategies in order to

solve interdisciplinary problems and gain skills and knowledge as they are engaged
with STEM related activities through formal and informal learning programs [3, 4].
Currently, there is a shortage of student interest in STEM, and a lack of a well-
trained teaching force in those areas [5]. Austria, like many other industrialized coun-
tries, suffers under the known phenomenon that the affinity of the youth to STEM
professions decreases with the degree of achieved wealth in the country [6]. Almost
52 % of Austrians rate it “as unimportant for a personal daily routine to know some-
thing about science and research”, and Austria therefore represents a front-runner in
a negative sense out of 32 European states [7]. Already today in Austria, eight out
of ten industrial companies have problems to find qualified personnel in the fields of
engineering, production, research and development [8]. In a direct European com-
parison regarding the youth interest in a STEM career Austria holds one of the last
places, although two thirds of university graduates, which studied in a technical field,
found qualified jobs within three months after graduation [9]. Clearly, there is a need
to educate large numbers of students who can fill these well-paid STEM jobs. How-
ever, engaging students in high quality STEM education requires programs to include
rigorous curriculum, instruction, and assessment, to integrate technology and engi-
neering into the science and mathematics curriculum, and also to promote scientific
inquiry and the engineering design process [2].
There are a number of strategies and programs targeted to increase the number of
scientists, engineers, technologists and mathematicians. Numerous initiatives from
USA call for an extensive effort to reform STEM education and cultivate the next
generation of skilled scientists, engineers, technicians, and science and mathematics
educators [10, 11]. In Europe the EU initiative inGenious1 aims to improve the image
of the STEM fields and careers among young people and encourage them to think
about the wide range of opportunities provided by working in science. It is repeatedly
argued that long-term economic growth depends on a country’s success at fostering
young people’s achievement in the STEM subjects [12].
However, increasing student competencies in science subjects is not an easy task.
Several reports have stated that deductive, teacher-led lessons are still the norm [13].
This traditional approach is criticized and there are many calls for a more “active”
familiarization with scientific contents [14, 15]. Research indicates that a large per-
centage of students’ STEM learning has the potential to happen in places outside
the formal classroom such as homes, after-school programs, camps, science clubs
and fairs, museums, outreach labs as well as everyday experiences. Positive impli-
cations for the students’ STEM learning in this context is proven as students who are
engaged in informal STEM activities demonstrate higher reasoning abilities than
students who do not participate in such activities [16, 17]. The report of the Euro-
pean Commission views the resurgence of science education as dependent on more
inquiry-based teaching and, significantly, increased opportunities for the cooperation
between the formal and informal sectors [18]. Consequently, emerging, innovative
programs are providing engineering students with an increasingly wide variety of

1 http://www.ingenious-science.eu/.
Offering Multiple Entry-Points into STEM for Young People 43

learning opportunities outside of the formal academic curricula enabling students

to “experience engineering” in an authentic environment. Generally, effective out
of school programs, as a one promising avenue to support students in learning and
pursuing STEM careers, should be student-centered, employ cooperative learning
strategies, and foster skills and attitudes towards STEM through “authentic, hands-
on activities” [19].
A variety of programs alternative to traditional schooling have gained promi-
nence in education communities [20] and a growing number of informal programs
have been created to take advantage of this opportunity while addressing the need
to increase participation in science and engineering [21]. Besides large, dedicated
blocks of time, these informal programs have other unique benefits over formal
schooling. After-school program activities have become a means for students to bet-
ter understand scientific concepts and processes and allow them to acquire scientific
inquiry skills and develop scientific reasoning [22, 23]. Thus, opportunities are pro-
vided for student-directed, collaborative projects where the participants learn exper-
imentally, creating knowledge through the transformation of their own experience,
and where they are immersed within the world they are learning about [24]. Fur-
thermore, studies have shown that after-school programs can offer opportunities not
available during day school and that engaging in after-school activities fosters the
students’ intrinsic motivation putting forth more concerted efforts in these settings
[25]. After-school instructional environments offer also a greater variety of topics
to explore, making STEM concepts more real to students and providing a variety
of engaging and interesting activities to develop STEM skills. In this kind of envi-
ronment, students learn how to collaborate and communicate with their peers and
teachers in ways different from their interactions in regular classrooms [4, 26].
This work presents the non-profit association Practical Robotics Institute Austria
(PRIA) as an approach for offering multiple entry points for students of different
ages to engage in the STEM fields and entrepreneurship. A broad range of activ-
ities is provided for teaching and learning robotics and ICT thus supporting flexi-
bility, experimentation, and playing with technology. The paper is structured as fol-
lows: The following section briefly introduces PRIA and its aims and organization.
Section 3 describes the possibilities of PRIA offered to the students of the different
school levels. Finally, the impact of the PRIA is described in the fourth section and
a conclusion is given in the fifth section.

2 Practical Robotics Institute Austria

The Practical Robotics Institute Austria (PRIA) is a non-profit association with the
vision to prepare and motivate next generations of researchers, engineers, and scien-
tists. The aim to promote scientific and technical excellence in schools using robot-
ics and ICT is reached through extensive educational and research programs. The
educational activities focus on the involvement of pupils and students into com-
plex projects and problem solving processes. This involves the development of new
44 W. Lepuschitz et al.

teaching methods, which stimulate learning and experimenting as well as realizing

one’s own ideas, for increasing the interest of children and adolescents in science,
technology and innovation as well as the encouragement of entrepreneurial think-
ing. The research activities are focused on executing projects concerned with the
development of innovative control architectures and technologies for robotics and
industrial automation as well as product and prototype development in the fields of
electronics and ICT. Generally, PRIA merges the fields of research and education
integrating the achievements and findings from research in education and vice versa.
PRIA uses robotics as one of the basic tools for introducing students to the STEM
fields as it is an interdisciplinary domain promoting the development of systems
thinking, problem solving, self-control, and teamwork skills [27].
PRIA was founded in 2012 by Dr. Gottfried Koppensteiner together with Dr.
Munir Merdan and the students Christoph Krofitsch and Reinhard Grabler. Before
founding PRIA, Dr. Koppensteiner already carried out several projects involving
high school students in science and research during his employment at the Vienna
University of Technology. Also, he organized the European Conference on Educa-
tional Robotics (ECER), a scientific conference for high school students involving
also a robotics competition, for the first time in 2012. To intensify the involvement of
students into research and to have a legal body for acquiring the necessary budgetary
means, PRIA was founded. While both Dr. Koppensteiner and Dr. Merdan could look
back already on a successful scientific career, the other two founding members Mr.
Krofitsch and Mr. Grabler just had graduated from high school. Nevertheless, due
to their former engagement in scientific projects during their school time, they were
eager to support the creation of PRIA and remained as employees ever since. PRIA
is located at their former school called Technologisches Gewerbemuseum (TGM),
which is Vienna’s largest technical high school. Around 2400 high school students
aged generally between 14 and 19 years old attend this school. Additionally 400 stu-
dents attend the evening school for adults.

3 Young People Engaging in STEM

PRIA offers multiple entry points for young people to become acquainted with
STEM and entrepreneurship. PRIA provides a program with activities for students
from primary, middle and high school as well as for those attending university (see
Fig. 1).
Offering Multiple Entry-Points into STEM for Young People 45

Fig. 1 Entry points offered by PRIA for young people of different school levels

3.1 Entry Points for Primary and Middle School Students

3.1.1 Workshops

Primary and middle school students can engage themselves in robotics workshops.
At the very beginning the students built up their first simple robot in electronic
lessons to get in touch with an easy level of robotics. This robot consists of a body
and batteries along with a simple electronic circuit with two motors and two light
transistors. The robot is an intelligent light follower, strongly derived from Braiten-
berg’s simple robot [28] and the BYO-Bot.2 In this workshop the students learn about
electrical and mechanical aspects. In further workshops the students are exposed to
the technologies used in LEGO Mindstorms or Botball. The educational repertoire
includes graphical programming but also easy programming guides for C-language,
teaching them about vision systems as well as sensors, motors, and controllers.
Depending on the project and the demands of a school, workshops are offered in var-
ious lengths. A university graduated professional of PRIA serves as primary contact
for the courses, facilitating activities and managing content and organization issues.
Extraordinary undergraduate students employed by PRIA provide instructional sup-
port. They serve as mentors and role models for the participants in the workshops.
During the school year, PRIA offers workshops for schools in the frame of publicly
funded projects (such as ER4STEM3 ).

2 https://bbstore.mycafecommerce.com/product/hand-assemble-byo-bot.

3
http://www.er4stem.com.
46 W. Lepuschitz et al.

3.1.2 Summer and Winter Camps

Apart from the publicly funded workshops, so-called summer and winter camps are
offered in the official school holidays. This program integrates successful ideas from
similar other programs [29]. Mobile robots are used as a motivating and interesting
tool for studying machines design and construction, software, as well as commu-
nications systems. The program allows participants to connect theory with practice
and exercise team work, project management, problem solving, and communication
skills in a stimulating setting. The students are mentored in different ways of per-
forming their tasks and how to solve the problems from stage to stage. Throughout
these activities, the student groups develop different ideas and concepts in order to
accomplish their goals, issues, and problems. Currently the winter camps are offered
during one week in February and summer camps are offered during one or two weeks
at the end of August. Furthermore, PRIA is regularly carrying out workshops in var-
ious holiday camp activities organized by other institutes or companies.

3.1.3 Events

PRIA regularly organizes events for bringing science to the school students. In the
frame of the regionally funded project STEMoFuture, PRIA organized two times
the “Long Night of Technology” at the TGM. Students from primary and middle
schools were informed and encouraged to visit these events as they incorporated
exhibitors from several institutes such as the Vienna University of Technology or
the Austrian Institute of Technology. Furthermore, the TGM itself presented some
of its departments to the possible future high school students. Moreover, primary
and middle schools are invited every year to attend the ECER as guests. By seeing
older school students engaging themselves in robotics, the younger ones realize that
technology is not only for adults but can be pursued also at a younger age at an
advanced level. Likewise to the primary and middle school students, of course also
older students are invited to attend events such as a “Long Night of Technology”.

3.2 Entry Points for High School Students

3.2.1 Competitions

Each year PRIA organizes the European Conference on Educational Robotics

(ECER4 ). It is a well established international scientific conference for high school
students having the official European Regional Botball Competition as core activ-
ity. Botball5 is a team-oriented robotics competition, at which each team receives a

4 https://pria.at/en/ecer/.

5 http://www.botball.org/.
Offering Multiple Entry-Points into STEM for Young People 47

standardized robotics set consisting of metal and Lego parts, sensors and actuators
as well as two robot controllers. New competition tasks are developed every year and
published at the Botball workshop at the beginning of a new season, giving every
team the same time for developing their strategies and robots. In the preparation
for ECER, PRIA offers annually a large Botball workshop free to attend for ECER
participants. Technical details of Botball are discussed and the game rules for the
competition at ECER are explained. Furthermore, the workshop also contains a ses-
sion concerning scientific writing. Generally, students develop their robots to solve
the tasks of the game table having about three months to find a good strategy and
to build up and program their robots for this strategy. At the conference the students
can present their findings in engaging talks, show their robots live, and take part in
the exciting robot competitions. A special focus is given to planning, documentation,
and the quality of the technical solutions. Apart from the competition, special awards
are granted for outstanding achievements in programming, mechanical engineering
as well as documentation.

3.2.2 Research Projects and Pre-scientific Works

PRIA integrates high school students into the ongoing research projects carried out
by actual researchers. Typically these projects are related to the topics of Industry
4.0, which fits well to the abilities of high school students from technical high schools
such as the TGM. In such schools, the students have to compose a pre-scientific work
denoted as diploma thesis. By knowing that they contribute to an actual research
project, the students are assured that their work is meaningful. Real projects with
meaningful outcomes have been shown to engage students, especially when real sci-
entists are involved [30]. In case the students would like to carry out a project of their
own, PRIA considers if it fits into the association’s goals and gives support even if it is
not directly part of a research project. Projects and pre-scientific works represent first
steps for the high school students to become entrepreneurs as they have to manage
their work themselves (of course with assistance of PRIA staff). This kind of partic-
ipation is of advantage for all involved stakeholders of the project. The evaluation
of a project in which scientists partnered with teachers found considerable benefits
for teachers and scientists as well as the students; benefits that included increased
knowledge and considerable enjoyment for all persons concerned [31]. Moreover,
many students dont envision themselves as scientists, in part because they dont see
scientists as “real people.” Formal and informal opportunities to connect with scien-
tists can help students recognize that those are “regular people”, who have hobbies,
families, and out-side interests [32].

3.2.3 Internships

During the summer holidays PRIA offers internships, which can be on the one hand
in the frame of research projects offering an insight into ongoing research activities.
48 W. Lepuschitz et al.

On the other hand the internship can be in the frame of the robotics workshops in
PRIA’s summer camps as an assistant supervisor supporting the attending young
people in their endeavors with robotics.

3.3 Entry Points for University Students

3.3.1 Employment

After graduation from high school, talented young people can become employees
of PRIA. Most of the current university students employed at PRIA already were
involved in previous PRIA activities during their school time. The possible tasks are
manifold and the work requires technical as well as social skills. Students can engage
in the research as well as in the educational activities of PRIA (e.g. supervising at
workshops or at summer and winter camps). Moreover, when a group of high school
students carry out their diploma thesis project, university students employed at PRIA
often act as their primary contact person (even though supervised by one of the PRIA
researchers with university degree). Thus, they improve both their technical and their
social skills. Furthermore, they can take part as co-authors at the composition of
scientific papers.

3.3.2 Masters Thesis, Bachelors Thesis, Term Paper

A further opportunity for university students is to write a thesis or a term paper.

Either it is a topic related to work conducted at PRIA or the student can suggest a
topic. In either case an employee of PRIA with a university degree can act as an
external supervisor.

4 Young People Engaging in Research

As mentioned earlier, high school students as well as university students have the
opportunity to participate in ongoing research projects at PRIA. One of these projects
entitled “Smart Phone Control of Robots for Education & !ndustry” (SCORE!) is
concerned with the use of mobile devices for controlling robots in an easy and afford-
able way.
Modularity and flexibility in robot-based training and education while keeping
the work-setting simple and intuitive is a key factor for involving teachers and stu-
dents. Mobile devices such as smartphones offers great potential as most people are
familiar with them. The smartphone has remarkable power in computation, involves
convenient operations, such as camera monitoring or wireless internet, and encom-
passes various components that are useful for controlling robots. Part of the project
Offering Multiple Entry-Points into STEM for Young People 49

Fig. 2 Robotics controller Hedgehog and according smartphone app for programming and con-
trolling robots

SCORE! was the development of a robotics controller denoted as Hedgehog consist-

ing of hard- and software, which offers flexibility and extension possibilities for in
the domain of robotics education. The controller works with a two-layer-architecture
based on reflective (low-level control) and planning (high-level-control) paradigms,
which, besides supporting educational features, makes it usable for applications in
robotics research as well. An app for mobile devices is used as interface to the user,
which enables robot testing by accessing sensors, motors and servos, as well as devel-
oping programs that can be downloaded to the controller to get executed (Fig. 2).
During the project, 25 high school students participated extensively in the devel-
opment of the Hedgehog controller. A total of five high school diploma theses and
four yearly projects were finished. Additionally, several high school students partici-
pated in the tests of the Hedgehog controller. Furthermore, three university students
wrote their bachelor and diploma theses in the frame of this project [33–35] and
three university students as well as one high school student, all of them employed by
PRIA, took part in the publication of research papers [36–38].
A further project, entitled “Batch Process Automation with an Ontology-driven
Multi-Agent System” (BatMAS), is concerned with the application of knowledge-
based systems in the domain of Industry 4.0. In the frame of this project, a group
of four high school students worked on their high school diploma thesis and one
university student employed by PRIA contributed to a scientific publication [39].

5 Impact

PRIA’s impact is significantly rising from year to year. In 2015 more than 1050
children and adolescents were reached by the activities carried out by PRIA. 35 %
of them were girls and 33 % had migration background. Due to the funded projects,
50 W. Lepuschitz et al.

80 % were able to participate without costs. Additionally, nearly 400 parents were
also reached as they represent an important part in the decisions of the children and
adolescents regarding education and career.

6 Conclusion

This paper presented the entry points provided by PRIA for young people to the
STEM fields. In principle the fields of research and education are merged offering
multiple possibilities for young people to become acquainted with STEM encom-
passing workshops, camps, events, followed by competitions, research projects, and
finalized with thesis, mentoring and employment. This provides opportunities for
students on the one hand to increase their STEM literacy and on the other hand to
acquire both problem-solving skills and experience relevant for their future develop-
ment. The PRIA approach highlights how innovative and effective ongoing activities
can improve the opportunities for students and help fix the broken pipeline in STEM
education.

Acknowledgments The authors acknowledge the financial support from the European Unions
Horizon 2020 research and innovation program under grant agreement No. 665972.

References

1. Bybee, R.W.: What is STEM education. Science 329, 996 (2010b). doi:10.1126/science.
1194998
2. Kennedy, T.J., Odell, M.R.L.: Engaging students In STEM education. Sci. Educ. Int. 25(3),
246–258 (2014)
3. Lacey, T.A., Wright, B.: Occupational employment projections to 2018. Mon. Labor Rev. 82–
109 (2009)
4. Sahin, A., Ayar, M.C., Adiguzel, T.: STEM-related after-school program activities and associ-
ated outcomes on student learning. Educ. Sci. Theor. Pract. 14(1), 13–26 (2014)
5. Mataric, M.J., Koenig, N.P., Feil-Seifer, D.: Materials for enabling hands-on robotics and stem
education. In: Proceedings of the AAAI Spring Symposium on Robots and Robot Venues:
Resources for AI Education, Stanford, CA (2007)
6. Sjberg, S., Schreiner, C.: The next generation of citizens: attitudes to science among youngsters.
In: Bauer, M., Allum, N., Shukla, R. (eds.) The Culture of Science How does the Public Relate
to Science Across the Globe? Routledge, New York (2010)
7. EUK, Science and Technology Report, Eurobarometer (2010)
8. Industriellenvereinigung: ZAHLEN, DATEN & FAKTEN Arbeitsmarkt und Karrierechancen
in Mathematik, Informatik, Naturwissenschaften und Technik (2013). http://www.iv-net.at/iv-
all/publikationen/file_610.pdf, Accessed 19 Jan 2016
9. Wanek-Zajic, B.: Wohin nach der Ausbildung? Bildungsbezogenes Erwerbskarrierenmonitor-
ing (bibEr), AMS info. No. 222 (2012)
10. Rising Above the Gathering Storm: Energizing and employing America for a brighter eco-
nomic future. National Academy Press, Washington, DC (2007)
11. Obama Educate to innovate. https://www.whitehouse.gov/the-press-office/president-obama-
launches-educate-innovate-campaign-excellence-science-technology-enon, 23 Nov 2009
Offering Multiple Entry-Points into STEM for Young People 51

12. Sawyer, R.K.: Optimising learning implications of learning sciences research. In: OECD (Ed.)
Innovating to learn, learning to innovate, pp. 45–65. OECD Publishing (2008)
13. Andres, S.H., Steffen, K., Ben, J.: TALIS 2008 technical report. OECD, Teaching and Learning
International Survey, Paris (2010)
14. Itzek-Greulich, H., Flunger, B., Vollmer, C., Nagengast, B., Rehm, M.: Trautwein U: effects
of a science center outreach lab on school students’ achievement are student lab visits needed
when they teach what students can learn at school? Learn. Inst. 38, 43–52 (2015)
15. Schroeder, C.M., Scott, T.P., Tolson, H., Huang, T.-Y., Lee, Y.-H.: A metaanalysis of national
research: effects of teaching strategies on student achievement in science in the United States.
J. Res. Sci. Teach. 44, 1436–1460 (2007)
16. Gerber, B., Edmund, M., Cavallo, A.ML.: Development of an informal learning opportunities
assay. Int. J. Sci. Educ. 23(6), 569–583 (2001)
17. Nite, S., Margaret, M., Capraro, R., Morgan, J., Peterson, C.: Science, technology, engineering
and mathematics (stem) education: a longitudinal examination of secondary school interven-
tion. IEEE Front. Educ. Conf. (FIE) (2014)
18. Commission, European: Science Education Now: A Renewed Pedagogy for the Future of
Europe. European Commission, DirectorateGeneral for Research, Brussels, Belgium (2007)
19. Hussar, K., Schwartz, S., Boiselle, E., Noam, G.G.: Toward a systematic evidence-base for
science in out-of-school time: The role of assessment (2008). Accessed 11 Sept 2008
20. Bucknavage, L.B., Worrell, F.C.: A study of academically talented students in extracurricular
activities. J. Second. Gift. Educ. 6(2/3), 74–86 (2005)
21. Cardella, M.E., Wolsky, M., Paulsen, C.A., Jones, T.R.: Informal pathways to engineering:
preliminary findings. Paper presented at 2014 ASEE Annual Conference, Indianapolis, Indiana
(2014). https://peer.asee.org/20638
22. Fisanick, L.M.: A descriptive study of the middle school science teacher behavior for required
student participation in science fair competitions. Published Doctoral Dissertation, Indiana
University of Pennsylvania, Pennsylvania) (2010)
23. McGee-Brown, M., Martin, C., Monsaas, J., Stombler, M.: What scientists do: Science
Olympiad enhancing science inquiry through student collaboration, problem solving, and cre-
ativity. Paper presented at the annual National Science Teachers Association meeting, Philadel-
phia, PA (2003)
24. Kolb, David A., Learning, Experiential: Experience as the Source of Learning and Develop-
ment. Prentice Hall, Englewood Cliffs, NJ (1984)
25. Vandell, D.L., Shernoff, D., Piece, K., Bold, D., Dadisman, K., Brown, B.: Activities, engage-
ment, and emotion in after-school programs (and elsewhere). New Dir. Youth Dev. 105, 121–
129 (2005)
26. Mahoney, J.L., Cairns, B.D., Farmer, T.W.: Promoting interpersonal competence and educa-
tional success through extracurricular activity participation. J. Educ. Psychol. 95(2), 409–418
(2003)
27. Johnson, J.: Children, Robotics and Education. In: Artificial Life and Robotics, 16–21. Springer
(2003)
28. Braitenberg, V.: Vehicles: Experiments in Synthetic Psychology. MIT Press (1986)
29. Stapleton, W., Asiabanpour, B., Stern, H., Gourgey, H.: A Novel Engineering Outreach to High
School Education, Imagining and Engineering Future CSET Education, 2009 Frontiers in Edu-
cation (FIE) Conference. San Antonio, TX (2009)
30. Stocklmayer, S.M., Rennie, L.J., Gilbert, J.K.: The roles of the formal and informal sectors in
the provision of effective science education. Stud. Sci. Educ. 46(1), 1–44 (2010)
31. Rennie, L.J., Howitt, C.: Science has changed my life!: Evaluation of the scientists in schools
project (A report prepared for CSIRO.). Department of Education, Employment and Workplace
Relations, Canberra, Australia. Accessed 25 Nov 2009
32. Cohen, C., Patterson, D.G.: The emerging role of science teachers in facilitating STEM
career awareness (2012). Retrieved from http://nwabr.org/sites/default/files/pagefiles/teaching-
STEM-career-awarenessPRINT.pdf
52 W. Lepuschitz et al.

33. Novotny, L.: Mensch-Maschine Schnittstelle im Bereich der mobilen Robotik; supervisors: M.
G. Koppensteiner, PRIA, Tesar, FH Technikum Wien (2013)
34. Krofitsch, C.: Remote Debugging Environment for Educational Robotics; supervisor: P. Vienna
University of Technology, Puschner (2015)
35. Chua, D.: Entwicklung und Umsetzung von Augmented Reality im Bereich der Industriero-
botik; supervisors: C. W. Lepuschitz, PRIA, Engelhardt-Nowitzki, Technikum Wien (2015)
36. Krofitsch, C., Hinger, C., Merdan, M., Koppensteiner, G.: Smartphone driven control of robots
for education and research. In: IEEE 2013 International Conference on Robotics, Biomimetics,
Intelligent Computational Systems, Yogyakarta, Indonesien (2013)
37. Krofitsch, C., Lepuschitz, W., Klein, M., Koppensteiner, G.: Flexible development environment
for educational robotics. In: International Conference on Control, Automation and Robotics,
Singapore (2015)
38. Koza, C., Krofitsch, C., Lepuschitz, W., Koppensteiner, G.: Hedgehog: a smart phone driven
controller for educational roboticspaper. In: 6th International Conference on Robotics in Edu-
cation, Yverdon-les-Bains, Switzerland (2015)
39. Lepuschitz, W., Lobato-Jimenez, A., Axinia, E., Merdan, M.: A survey on standards and
ontologies for process automation. In: Proceedings of the 7th International Conference on
Industrial Applications of Holonic and Multi-Agent Systems, pp. 22–32 (2015)
 Part II
Educational Robotics Curricula
How to Teach with LEGO WeDo at Primary
School

Karolína Mayerové and Michaela Veselovská

Abstract In this paper we present a set of activities with robotics kit Lego WeDo.
Activities are designed for ordinary pupils of the third and fourth grade of primary
school, and not only for robotic fans. We briefly describe every task of each activity.
These eight activities are iteratively confirmed sequence which we recommend to
use for familiarization with the elements of the kit and also software. Set of activi-
ties and recommendations to them is the result of several iterative verifications and
editing, based on observations of teaching according to our materials during infor-
matics lessons. These activities follow the requirements of our national education
program and develop several 21st century skills for successful life.

Keywords Activities ⋅ LEGO WeDo ⋅ Primary educational ⋅ Robotics ⋅ Introduc-

tion to programming

1 Introduction

In this article we present our final curriculum for educational robotics, which were
inspired by many ideas. For example we followed ideas of constructionism [1, 2],
managing the robotics classroom [3] or development of educational goals in psy-
chomotor, affective and cognitive domain [4]. This pedagogical intervention is the
output of our iterative research developments in the field of educational robotics. In
this article we present the eight activities which we designed for pupils of primary
school. We created this curriculum for the purpose of implementing educational
robotics into compulsory subject Informatics in Slovakia. Therefore we based on
the experiences we have gained during the realization of our dissertation project [5],
and also with respect to the existing national educational curriculum for informatics

K. Mayerové (✉) ⋅ M. Veselovská

Faculty of Mathematics, Physics and Informatics, Comenius University in Bratislava,
Mlynská dolina, 842 48 Bratislava, Slovakia
e-mail: [email protected]
M. Veselovská
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 55

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_5
56 K. Mayerové and M. Veselovská

[6]. So that they are now usable in average class at primary school and allow devel-
opment of 21st century skills in appropriate and playfully way [7]. For our research
goals we were searching for robotic kit, with which pupils can develop skills to con-
struct robotic models and also skills to program it. Therefore we chose robotic kit
LEGO WeDo. There are many another robotic activities, for example [1, 3, 8] but
they are not usable for primary school subject Informatics in our country.

2 Research Methodology

One of the aims of our dissertation research was to iteratively design, implement
and verify curriculum, which can introduce one of possible ways of implementa-
tion of educational robotics into informatics in primary school. In our research we
specified two research questions. In this article we deal with one question: What
are the aims, form and content of curriculum for educational robotics in context of
primary school subject Informatics in Slovakia? So, in our dissertation research we
focused on the creation, implementation and iterative verification of our curricu-
lum, which consists from activities for primary school pupils and methodological
materials for teachers. These activities are destined for informatics lessons using
the LEGO WeDo construction set with its integrated programming environment for
8 to 12 years old pupils. In our research we used qualitative methods of data col-
lection and data analysis [9], including observation (field notes, transcriptions and
drawings), focus groups, audio-visual materials (pictures, photographs and recorded
videos) of pupil’s products. As a research strategy, we used design based research
[10], which allows researchers to enter and interfere the process and thus create an
enhanced intervention. We verified our activities with more than one hundred lessons
in five stages during few years. We conducted our research in three different primary
schools in Slovakia. In first stage of our research one researcher taught designed
activities and other researcher was collecting data. In the other stages of research
our activities were taught by three different teachers in three different schools and
both researchers were collecting data. We used different methods for data analy-
sis, for example open coding and constant comparison. For validation of achieved
learning goals we used analysis of recorded videos, which included pupils work and
reactions during realization of our activities. This way we could compare their skills
and knowledge in the beginning and in the end of our intervention.

3 Our Eight Activities

For each of the eight activities we have developed methodical material for teachers.
For some activities we also designed and verified worksheets for pupils. We recom-
mended that it is beneficial to implement the first four activities in third grade and
another four activities in fourth grade. The full version of our curriculum in final
How to Teach with LEGO WeDo at Primary School 57

form, along with worksheets for pupils can be found in our doctoral thesis1 [11].
Each methodological material contains the following seven sections: Preconditions,
Aids, Goals, Competencies,2 Time, Assignments and The recommendations. But for
lack of space we mention only some of them. In our activities pupils work in small
groups (two or three members). Each activity also contains the correct solutions for
programming problems from worksheets, as well as examples of expected final mod-
els. In the next part of article there is detailed description of the eight activities.

3.1 Hello Robot—Activity 1

The learning object of first activity is to organise pupils’ previous knowledge about
robots, using discussion. These activities were described in our earlier papers [14].
At the end of the lesson, pupils should have learned and know that robots’ main
purpose is to help people. Robots need some energy source to move (battery, adapter
and so on) and they consist of various components; robots cannot think on their own,
only if we “tell them to”, that is if we program them. Another goal of the lesson is
to introduce pupils to working with the Lego WeDo construction kit, and thus help
develop their fine motoric capacities and communication skills.
Description of Activity: In the beginning of activity we conduct discussion with
pupils. This way we can easy identify their experiences, representations and concepts
which they usually connect with theme robot. In the end of discussion we explain
next task in which they should create their own robotic model. In the end of whole
activity they introduce their robots with short presentation to other classmates. They
use robotic kit LEGO WeDo and program motor to move with software in computer.
In the end of activity every pair present their robot, so they state its mane and explain
its functions. The rest of pupils come close to presenting pair and they listen carefully.
They can also ask some questions about presented robot.

3.2 Building and Programming Small Airplane—Activity 2

This lesson’s learning object is to make pupils familiar with the Lego WeDo soft-
ware environment and its basic icons. It is their first encounter with commands such
as shape wait for, motor on for, motor power or possibly with the shape repeat3 com-
mand. The aim is, using attractive approaches, to allow pupils develop in them the
habit of following instructions, to observe the difference between commands shape
wait for and shape motor on for and to be able to apply commands in the correct

1
so far only in Slovak version.
2 are derived mainly from the skills for the 21st century skills [12, 13].
3 The shape repeat command appears in the end of activities only and not every pupil necessarily

manages to go through all of the activities.

58 K. Mayerové and M. Veselovská

order. Another objective is to improve the pupils’ fine motoric skills by constructing
a model and also develop social behaviour, communication and cooperation skills
using group work exercises.
Description of Activity: Pupils build airplane according to instructions.4 Then
they program it several different ways. For example turn motor on (to spin one way),
stop motor and turn it on to spin other way, simulate starting of motor of airplane,
and so on.

3.3 What Will Transport Look Like in the

Future?—Activity 3

This lesson aims at strengthening the knowledge of motor control. Furthermore, it

focuses on the development of creativity and skill to put one’s own ideas in prac-
tice using creative activity. Through fulfilling the assigned tasks, children look for
suitable methods to find solutions, later on for effective ways for improving their
models.
Description of Activity: Pupils should design, build and program model of vehi-
cle, which people will be used in 200 years. About 15 min before the end of activity
we interrupt pupils work and open presentations. This way every group of pupils can
introduce their model. We designed and confirmed two types of assessment for this
activity.

3.4 Rafting—Activity 4

The learning object of this activity is to develop, through playful forms, the capacity
to express oneself intelligibly and realise the importance of the sequence of order, i.e.
if we change the order of steps while building a particular model what consequences
it may have. Pupils should comprehend the importance of the details described, in
particular in the case of varied shapes and colours.
Description of Activity: In the beginning we talk about times when people were
rafting on the river differently as today, etc. We can ask pupil if they know what
is raft and if someone want to explain it to others. Then we explain task to pupils:
they should build model of raft but in different way. We prepared model of raft, which
consists from several smaller models, see on the left side Fig. 1. This model is hidden
from pupils. Pupils create several groups with 3 or 4 members. Each group choose
one member as observer. Only observer can see hidden model (which is situated in
marked place) and he can explain its structure to other members of group. Observer
can only describe mentioned model verbally, but he/she cannot show components of

4 Availableat Lego Education page: https://education.lego.com/en-us/lesi/support/product-

support/wedo/wedo-base-set-9580/building-instructions.
How to Teach with LEGO WeDo at Primary School 59

Fig. 1 Left side—Models from Activity 4. Right side—Example of the photographs of models
from Activity 7

model with any part of his/her body. We have prepared four models, so each member
of group can be observer. Other members of group should build hidden model based
on observers’ verbal explanation and instructions. If they do not understand, they can
ask questions to observer. When pupils build four smaller models, they can merge
them into one larger model.

3.5 Ventilator—Activity 5

The learning objective is to develop ability to determine specific elements based on

their identification in the final product and following the instructions. It provides
training of the capability to transfer information from a graphical representation to
reality, introduces pupils to the tilt sensor functionality and related software icons.
The activity involves revision of basic commands by means of gradual sequence of
programming tasks and further works with the shape wait for command in combina-
tion with shape numerical parameter and later links it with shape sensor parameter.
Description of Activity: At the beginning of the lesson, we ask the pupils to say
what they remember from previous robotics lessons. We should focus on two areas:
design and programming. Pupils should say something interesting and different from
what was said by previous classmates. Thereafter pupils will build ventilator accord-
ing to the instructions. When pupils finished they work at model, they start to pro-
gram it. At the end of the worksheet are instructions, how to move ventilator. We can
solve these tasks together with all pupils, or let pupils work individually. Each group
of pupils then can progress at their own pace.
60 K. Mayerové and M. Veselovská

3.6 Remembering the Coding the Ventilator—Activity 6

The learning objective is to strengthen knowledge learned during the previous les-
son and clarification of commands provided these have not been understood properly,
such as parameter values when setting shape the motor power, unproductive com-
mand duplicity, shape the wait for command in combination with shape sounds or
shape tilt sensor and shape parameter used in shape repeat.
Description of Activity: The tasks in the worksheet are efficiently ordered to
lead pupils to reflect on the meaning of each icon, or to reflect effectiveness different
combinations of icons. So that the pupils can test programs in a worksheet, first they
must build a simple model of the ventilator. When pupils solved the problem in a
worksheet, they start to build their stable base for ventilator, i.e. tower which has on
top the ventilator. It is possible that this activity can take two lessons.

3.7 The Administrator of Windmill—Activity 7

The learning object of this activity is to create a model based on information dis-
played on a photograph. Based on such resultant model, pupils are tasked to identify
and track back the procedure and repeat it. Their task is to build an other model.
They should express in words the functionality of the already created program, cre-
ate a program based on written instructions(given in words not using icons), and
also modify the already created program. Pupils test and characterise the function-
ality of the presented programs using tilt sensor, and finally gain a real experience
with running three programs in parallel.
Description of Activity: Tasks in the worksheet are divided into three parts. In the
first part pupils have to build a latch from parts at the photo. There are photographs
of the finished model of the latch on the right side Fig. 1. Thereafter pupils have to
build their own windmill based on their imagination. In the second part they have
to work with the program. In the third part, which probably do not succeed all the
pupils have to link the various programming structure in the left column with the
correct description in the right column.

3.8 What Did We Learn?—Activity 8

The goal of this activity is a final verification of programming skills in the Lego
WeDo environment and strengthening of the knowledge gained during previous
lessons. Moreover, it focuses on the explanation of programming concepts, which
have not been properly understood so far, making space for development of creativ-
ity combined with design and construction of own models based on the given criteria
and simple demonstration of parallelism.
How to Teach with LEGO WeDo at Primary School 61

Description of Activity: At the beginning of the lesson teacher distributes the

worksheet with two parts of tasks to each pupil. In the first part pupils have to write
how works the program composed of icons. This second part looks same like the
third part in Activity 7. Follows free creative activity. Pupils should create their
own model to be controlled according to the several criteria described in worksheet.
Pupils need to be lead do not spend the whole time only by building a model, but
focus mainly on the control (programming).

4 Conclusion

In conclusion, we would like to summarize a few recommendations, or research

findings we have gained during verification of our activities in practice. The aim
of our study was to develop activities that would be applicable in ordinary primary
school classes, and not only for gifted children, or for children with increased inter-
est in technology. Although we believe that the activities with robotic kits have their
place especially among these children. As we mentioned at the beginning, all the
activities are designed for approximately one lesson (45 min). Of course it depends
on the potential of students and teachers abilities effectively lead the lesson. This
means constantly be in contact with students and walk from one group to another,
constantly develop the pupils’ productive compulsion to work and make effectively.
From our experiences we know that activities with LEGO constitute a significant
motivation for students and can enjoy working with them for long periods. In addi-
tion, the scientific literature confirms that the implementation of educational robotics
in the primary education with one of the objectives being the development of play-
ful introduction to programming basics has a great educational potential, but has
only a limited use in the real schooling so far. We believe that we have created a
good curriculum for educational robotics for primary school education, which can
be used by teachers to develop various skills and competences in pupils, that is also
an interesting and attractive introduction in educational programming.

References

1. Papert, S.: The Eight Big Ideas of the Constructionist Learning Laboratory. South Portland,
Maine (1999). Unpublished internal document
2. Kafai, Y., Resnick, M.: Constructionism in Practice. Lawrence Erlbaum Asociates, Mahwah,
New Jersey (1996)
3. Gura, M.: Getting Started with LEGO Robotics: A Guide for K-12 Educators. International
Society for Technology in Education, United States of America (2011)
4. Pasch, M.: Teaching as decision making: instructional practices for the successful teacher.
Addison Wesley Publishing Company (1991)
5. Mayerová, K.: Pilot activities: LEGO WeDo at primary school. In: 3rd International Workshop
on Teaching Robotics, pp. 32–39 (2012)
62 K. Mayerové and M. Veselovská

6. NPI.: New national curriculum for primary school: Informatics. National Pedagogical
Institute, Bratislava. http://www.statpedu.sk/sites/default/files/dokumenty/inovovany-statny-
vzdelavaci-program/informatika_pv_2014.pdf, Accessed: 03 Dec 2016
7. Mayerová, K., Veselovská, M.: Robot kits and key competences in primary school. In: Infor-
mation and Communication Technology in Education. University of Ostrava, Ostrava, Peda-
gogical Faculty, pp. 175–183 (2012)
8. Ilieva, V.: ROBOTICS in the Primary School—how to do it? In: International Conference on
Simulation, Modeling and Programming for Autonomous Robots, SIMPAR, Darmstadt, pp.
596–605 (2010)
9. Creswell, John W.: Educational Research: Planning, Conducting, and Evaluating Quantitative.
Upper Saddle River, New Jersey (2002)
10. Kalaš, I.: Pedagogický výskum v informatike a informatizácii (2. časť). DidInfo 2009, pp. 15–
24 (2009)
11. Mayerová, K.: Miesto edukačnej robotiky v informatickej výchove (Educational robotics in
primary education). FMFI UK in Bratislava, Bratislava (2015)
12. Trilling, B., Fadel. Ch.: 21st century skills: Learning for life in our times. John Wiley &
Sons(2009)
13. ATCS.: Assessment and Teaching of 21st Century Skills: Defining 21st century skills.
http://atc21s.org/wp-content/uploads/2011/11/1-Defining-21st-Century-Skills.pdf, Accessed
20 Jan 2015
14. Mayerová, K., Veselovská, M.: How We Did introductory lessons about robot. In: Teaching
Robotics, Teaching with Robotics, Padova, pp. 127–134 (2014)
Using Modern Software and the ICE
Approach When Teaching University
Students Modelling in Robotics

Sven Rönnbäck

Abstract This paper presents a robotic course that was revised with the ICE
(Ideas-, Connections-, Extensions levels) approach of teaching in mind. It includes
practical labs where the students have to derive and implement mathematical mod-
els and verify those on a manipulator arm mounted on a small mobile robot called
KUKA youBot. To support hands-on experience Simulink blocks have been devel-
oped to enable sensor readings and control of the youBot. The students imple-
ment their models using Simulink and run them to control the youBot manipula-
tor. The labs have been used in several course settings with some modifications.
The course also includes a small project where the students use and go beyond the
results achieved in the labs, it relates to the Extensions level in ICE. Even though the
approach is under development it has been tested and successfully used in a course.

1 Introduction

It is important to align teaching activities with learning outcomes, so called Con-

structive Alignment [1], and that the learner get relevant learning activities to the
topic taught.
Our belief is that students can achieve deeper learning in robotics, such as forward
kinematics, inverse kinematics, and dynamics, if they get hands-on experience and
can feel the robot. In our robotics course we want students to achieve knowledge,
develop skills, and gain understanding [2].
The ICE approach [3] came in mind when reconstructing the curriculum for the
course Modeling in Robotics, 7.5 ECTS (2015), previously a bigger course named
Control Methods for Robot Applications, 15.0 ECTS (2014). ICE presents three lev-
els; Ideas-, Connections-, and Extensions level. The different ICE levels do not nec-
essary need to be assigned to different grades but to different parts of a course.
At Ideas level the students learn the theory, vocabulary, and information they can
recall. At Connections level learning happens when students are able to relate their

S. Rönnbäck (✉)
Department of Applied Physics and Electronics, Umeå University, 901 87 Umeå, Sweden
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 63

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_6
64 S. Rönnbäck

learning to what they already know [3]—build connections between the bits and
pieces, and at Extensions level learning happens when students take what they
already know and create something new. The three different levels are often associ-
ated to different course grades, where Ideas level is related to pass grade, and Exten-
sions level to the highest course grade.
We map Ideas level to lectures, recitations, and exercises. In the laboratory work
the students start to put the bits and pieces together, and we consider this to be related
to Connections level since it is where students actually through implementation and
testing gain experience on key concepts taught in lectures and exercises. The labs
are assignments, understanding performances [4], that are assessed through written
reports.
The project part, which is the last part in the course, students choose one project
among proposed ones and applies their understanding in a new setting, understand-
ing performance [4], to stretch knowledge, skills, and understanding.
The project part can be seen as the Extensions level according to ICE; the student
apply and create something new based on the concepts from Ideas and Connections
levels.

2 The Course and Its Hardware

Labs and projects are centered around the manipulator arm on the KUKA youBot
[5]. YouBot is a small mobile platform equipped with a lightweight five degrees of
freedom robotic manipulator, see Fig. 1.

Fig. 1 youBot is a small

versatile mobile robot
platform developed for
education and research. Here
the manipulator arm is posed
in its initial configuration
Using Modern Software and the ICE Approach When Teaching University Students . . . 65

The software used is Matlab with the Simulink support for the Raspberry PI com-
puter. To keep focus on mathematical modeling a Simulink interface for the youBot
was developed. It is based on the youBot API provided with the robot.

2.1 Course Work at ICE Connections Level

The course has several labs, here two labs are presented. In the first lab students
derive forward and inverse kinematics for a three linked robot. The knowledge from
the first lab is later applied when they derive kinematics for the youBot manipulator
arm.

2.1.1 Connections Level—Lab on Kinematics for a Three Linked

Robot:

In this lab students have to assign frames and derive inverse kinematics for a planar
three linked robot see Fig. 2. The robot arm has a roller attached as an end effector.
To solve the task the students need to understand the key concepts of assigning
frames, and kinematic decoupling to solve the inverse kinematics. They also need
to understand the key concept of how to use the Jacobian to calculate the velocity
vector of the end effector. Of course the students also use the Denawit-Hartenberg
convention to assign frames. The implementation in done in Matlab. After correct
implementation the roller attached at the end of the robot follows the circumference
of the oval shaped object, see Fig. 2. What makes this lab especially suitable is that
the planar robot has kinematics that is similar to three links of the youBot manipu-
lator arm, see Fig. 1.

Fig. 2 A three linked planar

robot with a roller represen-
ted as a circle, as an
end-effector. The key
concepts here are to calculate
the forward and inverse
kinematics, to assign frames,
and to calculate the velocity
vector of the end-effector
66 S. Rönnbäck

Fig. 3 The left block computes the inverse kinematics, and the right block computes the forward
kinematics for the youBot. The inverse kinematics block has a boolean output that indicates if a
solution for the desired pose exists

2.1.2 Connections Level—Lab on YouBot Kinematics:

In this lab students assign frames to the youBot manipulator arm according to
Denavit-Hartenberg [6] convention. Then forward kinematics and inverse kinemat-
ics are derived and implemented as Simulink blocks see Fig. 3. The key concepts are
forward kinematics, kinematic decoupling, and inverse kinematics for the youBot
manipulator arm. From a desired pose; [x, y, z]T coordinates and gripper orientation
(elbow twist and pitch angle) students need implement a block that computes the
joint angles for the arm, and one that computes the forward kinematics. After imple-
mentation of inverse kinematics the students can test and validate their solutions
on the real robot. An example of inverse and forward kinematics implementation is
visible in Fig. 3.

2.2 Course Work at Extensions Level—Project Work

In the project students exercise what they learned at Ideas and Connections levels,
they perform their understanding in a new setting. Over the year different projects
have be proposed and implemented by the students. Some projects are listed here:
∙ Sort soft drink cans by weight. (The gripper grabs loops attached to the cans).
∙ youBot control using the Microsoft Kinect sensor as the operator input. The stu-
dents successfully implemented and demonstrated their system which had support
for lateral and transversal movement of the platform, and Cartesian movement of
the gripper inside the robot workspace.
∙ Pick up small wooden cubes and place them on the tray (on the youBot)
Using Modern Software and the ICE Approach When Teaching University Students . . . 67

Fig. 4 The youBot arm has

been programmed to
automatically build a tower
from small wooden cubes
that randomly been placed on
a table in front of it. The web
camera mounted on the robot
arm, attached to a laptop
computer, was used for the
image processing. The look
of the tower was designed in
a GUI interface. A
Raspberry Pi computer and a
wireless router is visible on
the tray of the robot

One recent successful student project was building a tower from small colored
wooden cubes. The GUI (Graphical User Interface) was programmed in Matlab.
A web camera was mounted on the gripper and was used to detect cubes and their
color. Figure 4 shows a picture of the cube picking robot from when the students
demonstrated their project.

2.3 Student Feedback from Course Evaluation

Here follows students feedback from year 2014 and 2015.

Student positive feedback: —“It is very positive that we get to know how the the-
oretical concepts previously learnt work in a real system.” (2015)
—“i liked the amount of hours allocated to practical work and also that we were
constantly supervised.” (2015)
—“Working with the kuka was quite nice and challenging.” (2015) —“The
project was really good and challenging. I learned a lot from it. The labs were
also good, they made us apply the concepts from the lecture in a real robot, what
helped a lot understanding how it works.” (2014)
—“The work with the YouBot was very appreciated and gave us a lot of good
knowledge about how to control a real robot. The project was also a very good
part of the course.” (2014)
Students negative feedback: —“The labs were efficient practice for the theoret-
ical part of the course. But the distribution of the labs was not equal between the
first and second period in the semester.” (2014)
—“I miss applying more the second part of the course in the labs. We did not
really use the control part, so now it feels like it is really abstract and vague the
concepts and applications of it.” (2014)
68 S. Rönnbäck

—“The labs was mostly focused on kinematics and dynamics. The later half of
the lectures was not properly covered by lab work. It would be good to have one
lab with control also.” (2014)

3 Discussion

We have presented a robotics course that was revised with the ICE approach of teach-
ing in mind.
A Simulink block that enables students to work directly on the youBot was pre-
sented. Students have successfully used them in labs and in projects.
The approach by using Simulink for code generation to Raspberry Pi has the ben-
efit that a student can spend more time on the mathematics and modeling. From own
experience I know that equal projects purely implemented in C or C++ directly on
the robot itself would have consumed a lot more time.
What can be noticed in the students negative feedback is that when they did not
use control part (torque control) on the youBot, the student feels it is abstract and
vague. It is just an example that gives support to ICE approach of teaching.

Acknowledgments The author wants to thank the students that were involved in this study. This
work was done during a revision of the Master’s Programme in Robotics and Control at Umeå
University.

References

1. Biggs, J., Tang, C.: Teaching for Quality Learning at University. The Society for Reseach into
Higher Education (2011)
2. Entwistle, N.: Teaching for Understanding at University—Deep Approaches and Distinctive
Ways of Thinking. Palgrave MacMillian (2009)
3. Fostaty Young, S.F.: Teaching, learning, and assessment in higher education: using ICE to
improve student learning. In: Proceedings of the Improving Student Learning Symposium.
Imperial College, London, UK. pp 105–115 (2005)
4. Perkins, D.: Chap. 2 in Teaching for understanding: linking research with practice. What is
Understanding?. Jossey-Bass Publishers (1998)
5. Bischoff, R., Huggenberger, U., Prassler, E.: KUKA youBot- a mobile manipulator for research
and education. In: Proceedings of IEEE International Conference on Robotics and Automation
(May 2011)
6. Siciliano, B., Khatib, O.: Handbook of Robotics. Springer 2008. ISBN: 978-3-540-23957-4
Developing Extended Real and Virtual
Robotics Enhancement Classes with Years
10–13

Peter Samuels and Sheila Poppa

Abstract There is growing evidence of the potential of educational robotics to

enhance science, technology, engineering and mathematics education provided that
they are deployed carefully. This paper describes a developmental research project
between a university and a secondary school in the UK to develop extended
robotics enhancement classes, mainly using LEGO MINDSTORMS robotic kits,
and GeoGebra, which was used to animate virtual robots. Two styles of class were
deployed: student-led project creations and facilitator-led challenges. The peda-
gogical principles underpinning these classes and their design are discussed.
Feedback generally indicated that the classes were successful and appreciated by
the students but they experienced difficulties in incorporating the virtual robotic
element. Lessons learnt from the project, including the development of employa-
bility skills, the potential impact on students with autism, and the effective use of
peer students, are discussed. The possibility of combining the two styles of class
together is proposed.

Keyword Developmental research ⋅

Learning by design STEM education ⋅ ⋅
Robotic kits ⋅
LEGO MINDSTORMS ⋅
Virtual robotics Geogebra ⋅ ⋅
Employability skills

1 Introduction

According to Sanders [1] interest in science, technology, engineering and mathe-

matics (STEM) education increased rapidly in the USA following the publication of
Friedman’s book [2] in 2005. Friedman concluded that China and India were on

P. Samuels (✉)
Centre for Academic Success, Birmingham City University, Birmingham, UK
e-mail: [email protected]
S. Poppa
Lawrence Sheriff School, Rugby, UK
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 69

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_7
70 P. Samuels and S. Poppa

course to overtake the USA in the global economy by surpassing their STEM
education output. The supply of employable STEM graduates has also been an
increasing cause for concern in Europe in recent years [3–5]. The international
comparative Relevance of Science Education research project which investigated
the views of adolescent students [6] found a strong negative correlation (r = –0.85)
between their interest in a future science career and the wealth of their home
country. This indicates that many developed nations, including many in Europe,
face the challenge of meeting the demands for STEM employment. Furthermore,
inquiries into employers’ views indicate that, in addition to subject knowledge,
many STEM graduate roles require softer skills, such as communication, using
initiative, problem solving, teamwork and creativity [7, 8] which are not tradi-
tionally taught in secondary or tertiary education.
In response to these concerns, government programs and institutional research
projects have been initiated on both sides of the compulsory education threshold.
Their strategies have included creating greater awareness of STEM careers through
employer partnerships [7, 9] and seeking to make STEM education more enjoyable
[9]. Others have encouraged the acquisition of softer skills by introducing project
enhancement work into the curriculum [10] but these have mainly taken place in
Higher Education due to the pressures from national and international league tables
on compulsory school education [11]. These encourage schools to emphasize
individual performance and teaching to test rather than promoting divergent
thinking [11], preparing students for the collaborative and unpredictable world of
employment, or developing a love for academic subjects, or a deeper emotional
engagement with them [12].
Some initiatives to make STEM subjects more enjoyable have been criticized for
lacking effectiveness [9]. However, we assert that it is not the subjects that need to
be made more enjoyable, but the use of appropriate challenges in the experience of
engaging with the subject that needs to be encouraged, as the subjects themselves
are potentially intrinsically enjoyable to many students. The first author has
described this approach as “putting the curriculum into the fun rather than the fun
into the curriculum” [13, 14]. In the context of learning technologies, what is
required is an understanding of which technologies are intrinsically engaging and
enjoyable from the perspective of students’ informal use in their free time and how
such technologies could be incorporated into the curriculum [14].
One potential type of learning technology is educational robotics. There is
growing evidence that these have the potential to enhance learning, engagement and
employability skills in STEM subjects provided that they are deployed carefully
[15–17]. There is also evidence of the potential of integrated computer algebra
systems (CAS) with dynamic geometry systems (DGS) to enhance mathematics
education when used thoughtfully [14, 18]. This paper reports on an ongoing
developmental research enhancement project over the last 5 years between a sec-
ondary school and a university in the UK to develop extended robotic enhancement
classes with Years 10–13 students using robotic kits in conjunction with an inte-
grated CAS/DGS.
Developing Extended Real and Virtual Robotics Enhancement … 71

The background to these classes is introduced in Sect. 2. Section 3 provides a

summary of the research and development framework used in the classes. Perfor-
mance and feedback on the classes is reported in Sect. 4. Finally, Sect. 5 provides a
discussion of the outcomes of this developmental research project as a whole and a
possible improvement on the class design.

2 Background

The origins of our collaboration was a teaching idea paper [19], written by the first
author and published in 2010, which suggested using robotic kits along with a
mathematical simulation environment called GeoGebra (http://www.geogebra.org/)
in a context of open-ended project work in order to motivate mathematics learning.
The School has a selective intake and aims to develop well rounded students by
blending traditional style lessons with enhancement activities, both within and
outside class time. Under the UK General Teaching Council’s Teacher Learner
Academy (http://www.gtce.org.uk/tla/) staff were encouraged to undertake projects
to stimulate their learning experiences and that of their pupils, supporting each other
within and beyond their normal settings to enrich their pedagogy, thereby fostering
innovation. In 2010, they were awarded LEGO Innovation Centre status [20],
which enabled them to purchase LEGO MINDSTORMS robotic kits (http://www.
lego.com/en-gb/mindstorms/) for use with older students, and were seeking a way
to use them effectively. A member of staff from the School read the paper and
contacted the first author who was given permission to work with the School.

Table 1 Summary of robotics classes

Year Class type Length No. students Technologies used Evaluation
(days)
2011 Free time, 5 5 LEGO Presentation and
student-led projects, MINDSTORMS, peer review
individual and group GeoGebra and
Bioloid
2012 Free time, 4 11 LEGO Presentation and
student-led projects, MINDSTORMS, peer review
individual and group GeoGebra and
Bioloid
2012 Enhancement class, 3 17 LEGO Performance in
themed challenges, MINDSTORMS challenges and
group and GeoGebra sportsmanship
2014 Enhancement class, 3 7 LEGO Presentation and
student-led projects, MINDSTORMS peer review
group and GeoGebra
2015 Enhancement class, 3 11 LEGO Performance in
themed challenges, MINDSTORMS challenges plus
group and GeoGebra bonus
72 P. Samuels and S. Poppa

A summary of the classes which have been provided so far is shown in Table 1.
The first two classes were slightly longer, operated in student free time and used a
student-led project approach. The three later classes were organized in school
enhancement lesson time; and two of them deployed a themed challenge approach.

3 Research and Development Framework

3.1 Research Methodology

The overall purpose of these classes was to develop educational enhancement

environments using real and virtual robots that are engaging, facilitate the
acquisition of employability skills and motivate further STEM learning. A research
methodology appropriate for a partnership between researchers and teachers at a
university and teachers and students at a school for the development of these classes
was required. The nearest similar research known to the authors into the devel-
opment of classes by a researcher/teacher using a similar technology is that used by
Jaworski to develop classes using GeoGebra to teach algebra concepts [21]. She
reflected upon her practice as a teacher in the implementation of her chosen ped-
agogy and concurrently as a researcher into the effectiveness of the pedagogy itself,
referring to this approach as developmental research [22]. This paradigm was
therefore chosen as it was seen to be more appropriate than analyzing and devel-
oping classes from a single role identity, such as in action research for researchers
or reflection-on-action for practitioners.

3.2 Pedagogical Framework

The pedagogical framework for these classes combined several ideas from the
teaching idea paper by the first author [19] with other principles elaborated in [23].
Fundamental to the former was the combined use of real and virtual robotics to
motivate learning rather than teaching directly. The rationale for experimenting
with this approach was the remarkable success, reported in [18], of the use of
Classpad CAS/DGS calculators to motivate learning in algebra and geometry. As
already explained, GeoGebra can also be used as a CAS/DGS. Using GeoGebra
also provided a way to attempt to motivate mathematics learning using robotics
through animations.
The educational robotics movement can be traced back to Papert’s Mindstorms
book [24] which encouraged social engagement and language development in a
‘math’ world, and which led to the (mainly virtual) turtle graphics movement. The
combination of real with virtual robots is a novel idea, although it was anticipated
by Burdea [25] in a wider context who foresaw one advantage being more effective
Developing Extended Real and Virtual Robotics Enhancement … 73

planning. This is consistent with the use of simulations in other areas of mechanical
engineering, such as computational fluid dynamics in aircraft design. Real robots
are more kinesthetic than virtual ones and encourage greater social identification,
which Catlin and Blamires [26] have called the principle of embodiment. Eisenberg
[27] argued for a greater emphasis on physical robots in mathematics education as
“transitional objects” which bridge the gap between concrete and formal reasoning.
Another important element of the class pedagogy was learning by design [28].
This promotes providing learning environments where students are given the space
to create and develop their own ideas. It contrasts with teacher-led challenges, often
involving constructing and programming a pre-planned robot design to achieve a
pre-planned purpose, and robotic competitions, often involving pre-set challenges
requiring some ingenuity. In a study of 64 engineering undergraduates, Cropley and
Cropley [29] found that students without creativity training were so used to fol-
lowing instructions that they focused on conventional designs in robotics challenges
even when they were marked for creativity. Learning by design was therefore
employed as these classes were aimed at enhancing the curriculum and encouraging
creativity.
The classes also made use of teamwork and peer learning. Teamwork is com-
monly used in robotics competitions, such as the FIRST which employs teams of 20
or more students (http://www.firstinspires.org/robotics/frc/what-is-first-robotics-
competition). However, in their review of computer supported group-based learn-
ing, Strijbos et al. [30] found that teams of two or three were more effective for
performing complex technical tasks due to the amount of effort required to achieve
consensus. Atmatzidou and Demetriadis [31] argue that, “although the [educational
robotics] practitioners have a clear orientation toward collaborative learning
activities, they, nevertheless, lack a more detailed pedagogical perspective of how
to tap the benefits of group-based learning”. By viewing former class members as
an educational resource, peer learning [32] provides such a perspective. Consistent
with [11], emphasis was placed on their experience with former classes rather than
their ages.

3.3 Choice of Technology

The main technologies deployed in the classes were LEGO MINDSTORMS NXT
kits and the GeoGebra software environment. The first two classes also used a more
sophisticated ROBOTIS Bioloid Comprehensive humanoid robotic kit (http://www.
robotis.com/xe/BIOLOID_Comprehensive_en). This kit was inappropriate for the
two challenge-based classes. The choice of appropriate kinds of technology for
these classes is discussed in more detail in [33].
LEGO MINDSTORMS NXT kits comprise of a programmable brick which can
generate sounds, LEGO bricks and other pieces, three different kinds of sensors,
and servo motors [34]. They are used in conjunction with a visual programming
language which controls the robot’s behavior according to a series of instructions or
74 P. Samuels and S. Poppa

events based on inputs received from the sensors. Once a robot has been built and a
program written it can be downloaded onto the brick. The ratio of available robotic
kits to students was quite high but, in order to give each team an equal opportunity,
they were limited to having two kits per team.
GeoGebra is an open source dynamic mathematics software environment. It
comprises of six alternative views covering different aspects of mathematics and
statistics. In particular, it integrates a dynamic geometry system with an algebra
view, enabling the representation of physical objects, such as robots, to be con-
structed and animated, both visually and symbolically. [35] provides the animation
of a LEGO MINDSTORMS robot moving through three points on a plane which
was used as the basis of a minimal instruction activity in the first three classes.

3.4 Class Design

Each class began with a series of briefing sessions on the first day led by the
University partner. These all included an initial challenge to construct and program
the first robot design in the LEGO MINDSTORMS NXT instructions book and a
GeoGebra training session and challenge. In the more recent classes the students
were organized into groups and briefed on the activities for the main period of the
class. Two different styles of class were then adopted:
• Student-led project design: Students were given the freedom to create their own
projects within the parameter of being achievable within the time period
available. In the fourth class the students presented and peer assessed their plans
before they created and programmed their robots. This approach was similar to
that used by another UK university with their first year engineering under-
graduates [10].
• Themed challenges: Students were briefed on a number of specific challenges
around an engaging theme for which they were required to build a robot. The
third class included three challenges with an Olympic theme, coinciding with
the 2012 Olympics. The fifth included a series of challenges with a Rugby
theme, coinciding with the 2012 Rugby World Cup (see Fig. 1).
The middle class period lasted between one and three days and was facilitated by
members of staff from the School and/or peer students who had participated in
previous classes at a lower level. The final day of most of the classes started with a
re-briefing session followed by a final period for robot development. This was
either followed by the presentation of robot designs with a peer review of their
performance in the themed challenges for which they were either scored or ranked.
After this there was an award and certificate giving ceremony. Finally, students
were asked to reflect upon their experiences and make suggestions for improving
the classes in future.
Developing Extended Real and Virtual Robotics Enhancement … 75

Fig. 1 LEGO
MINDSTORMS robot in the
fifth class making a
conversion ‘kick’

The assessment of the robots for the themed challenges included a peer reviewed
sportsmanship score in the third class and a peer student discretionary award for
sophistication in the fifth class. These were included to encourage the students to
look beyond the competitive aspects of the challenges to the wider purpose of the
class.
Certificates were awarded according to the level of participation of the students:
• Level One: Participation in a class as an individual or group member
• Level Two: Facilitation of a class
• Level Three: Design and facilitation of a new class
Students who had been awarded a certificate were encouraged to engage in a
later class at a higher level as a peer student.

4 Performance and Feedback on Classes

4.1 General Findings

Firstly, we note that these classes were held at a male secondary school with a
selective intake. The students were therefore relatively intelligent, well behaved and
76 P. Samuels and S. Poppa

competitive. Furthermore, in common with most UK secondary schools, the stu-

dents were used to following instructions, so many found the idea of student-led
project work, requiring creativity with limited rules, unusual and challenging.
A surprising finding was the students’ general lack of initiative with using the
Internet which appeared to be due to their restricted access within their normal
school environment.
Several planning meetings, sometimes involving peer students as facilitators,
were held before each class took place. Students self-selected to attend classes
which were advertised in the School. There were occasional issues with facilitators
not fully understanding the design of the classes, which were quite different from
normal lessons.
Observations and improvement recommendations were made by the university
researchers, the staff facilitators, and the students as part of their reflection at the end
of each class. The students participated in the training activities and generally
picked up what to do quickly but some had the tendency to go off task quickly if
they became bored. This confirmed the importance of the constructivist principle of
seeing the training as minimal instruction [23].
An initial aim of the classes had been to motivate deeper mathematics learning
by using GeoGebra to create accurate robotic designs requiring use of its algebra
view [19]. However, the earlier GeoGebra training sessions were disappointing as
some students with weaker mathematics backgrounds found the programming too
hard and failed to connect virtual animation with the robotic kits. They also
appeared to have a kinesthetic preference to use physical robot kits. The training
was therefore simplified by only including a straight line movement, using a
rectangle to represent a physical robotic table, and, moving away from a turtle
graphics declarative style program, by adding a feedback event to represent an
ultrasound sensor locating a wall—see Fig. 2. This was an improvement but there
remained the challenge of encouraging students to use GeoGebra later in the
classes. In view of these difficulties, the overall success of the robotic kits, and
consistent with the developmental research paradigm, this aim was widened to
motivating STEM learning in general.
Whilst the challenge-based classes were more competitive, some students lost
interest once they had developed a robot to meet a challenge which initially cap-
tivated their interest. This was particularly evident in the third class when some
students did not attend the full class once they had built a speed robot using gearing.
However, their groups did not perform well on another more technical challenge to
throw a ball. Whilst the students in the student-led classes found the freedom more
daunting, they all remained engaged throughout the class. The possibility of
combining these two approaches is discussed in Sect. 5.
The fifth class was mainly designed and led by two peer students from Year 13
who had participated in a previous class. They only received limited advice during
their facilitation and were able to manage the sessions and keep the other students
Developing Extended Real and Virtual Robotics Enhancement … 77

Fig. 2 Robot table and revised GeoGebra animation representing a robot translation on a table,
beeping when it senses a wall (source https://www.geogebra.org/material/simple/id/2807751)

engaged. Whilst some of the decisions they made about rules and scoring of
challenges were mildly criticized by the other students, this was seen as a positive
learning experience for them to become more proficient in design and facilitation,
which are themselves valuable employability skills.

4.2 Feedback

On the whole the students’ feedback at the end of the classes about their experi-
ences was positive, indicating their affective engagement [11, 12, 36]. Of the 33
written student feedbacks obtained from three of the classes 42 % mentioned
enjoyment or fun whilst none of them made a negative comment about their overall
experience. 12 % also mentioned that they thought the class design idea was good.
Student enthusiasm was also demonstrated by some having to be told to leave at the
end of the day in the school time classes whilst others gave the course leaders gifts.
78 P. Samuels and S. Poppa

In terms of the technical skills learnt, student feedback made predictable refer-
ences to learning to construct (27 %) and program (36 %) robots that achieved the
required goals. The LEGO visual programming language was often criticized,
especially by students who had experience with other programming languages. One
specific technical skill several students reported was learning how to use gearing
(12 %) in order to make robots move faster.
Student feedback made frequent references to acquiring employability skills.
The most common themes identified were teamwork/cooperation (79 %), time
management (36 %) and creativity/problem solving (30 %), indicating that most
students perceived these to be important skills that they had developed, enhancing
what they had learnt from the standard School curriculum. However, there was little
mention of planning/designing (15 %) which may have been due to the immediacy
of the robotic kits encouraging repeated experimentation rather than reflection. This
is discussed further in Sect. 5.
On the negative side, 53 % of students in the third class reported that the
sportsmanship peer evaluation had not worked well. A reason given for this was
that some teams had used tactical scoring. It was therefore decided that the facil-
itators should be made responsible for this in future. This appeared to work well
with the fifth class as the student feedback did not comment on this aspect
negatively.

4.3 Impact of Class on Students with Autism

An unexpected consequence of the first two classes was their positive impact on
some autistic students. Over 10 % of students who had attended a class were
identified as autistic, representing a much higher than average prevalence rate.
However, all the autistic students joined in and performed well in the classes. For
one student in particular the classes had a completely transformative effect to the
amazement of his teachers, one of whom stating that she had “never heard him
laugh before”. He was even willing to be videoed demonstrating his new-found
understanding of the principle of gearing—see Fig. 3. He then progressed to
facilitate a second class he attended. This success is consistent with the aspirations
of [37].
It is believed that Catlin and Blamires’ principle of embodiment [26] is partic-
ularly relevant to autistic students as they appear to be comfortable with relating to
robots as a projection of human relationships but without fear of violating social
rules. Relating to other people in this context also appears to be less threatening to
them, thus increasing their confidence in social engagement. Dautenhahn and
Werry [38] go further in claiming that it may also be the physical movement by the
autistic students themselves within the robotics class which might be therapeutic.
Developing Extended Real and Virtual Robotics Enhancement … 79

Fig. 3 Autistic student demonstrating to the camera the effectiveness of gearing

5 Discussion

This developmental research project has made progress towards its aim of devel-
oping an educational enhancement environment using real and virtual robots that is
engaging and facilitates the acquisition of employability skills. The aim to motivate
further STEM learning affected the design of the classes but was not evaluated.
LEGO MINDSTORMS kits, although their software was not universally liked,
have the potential to be used effectively in such classes. The use of a three level
certification structure, employing peer students from previous classes as facilitators
and class designers has demonstrated the value of seeing former students as an
educational resource. The classes were also unexpectedly successful with autistic
students, with the first two having a profound effect one student in particular.
The virtual robotic element with GeoGebra has yet to be proven, although the
GeoGebra training activities have been improved. The class designers still believe
that GeoGebra should be retained in order to encourage awareness and use of the
employability skill of planning. Based on feedback from students in the fifth class, a
possible way forward was identified. As the challenge-based activities are easier to
follow but may not be as engaging it is proposed that the middle class period should
be split into two halves, the first half being a series of challenges and the second
half a student-led project. In order to encourage planning, as in the fourth class,
students will be required to submit a plan of their project ideas for peer review
before they develop and present the project itself. They will not be forced to use
GeoGebra but it will be provided as an option (along with alternatives, such as
animations in Microsoft PowerPoint). This new design will be investigated in the
next class. The use of GeoGebra and alternatives to represent other robotic sensor
events will also be explored.
The success of the project’s pedagogy has also impacted positively upon the
approach of the class designers in their other teaching enhancement activities,
causing them to trust students more to develop their own ideas and providing them
80 P. Samuels and S. Poppa

with less direct guidance. We believe this style of robotics class could be employed
in similar contexts to encourage engagement with STEM education and the
development of employability skills.

References

1. Sanders, M.: STEM, STEM Education, STEMmania. The Tech. Teach. 68(4), 20–26 (2009)
2. Friedman, T.: The World is flat: a brief history of the twenty-first century. Farrar, Straus and
Giroux, New York (2005)
3. Roberts, G.G.: SET for Success: The Supply of People with Science, Technology, Engineering
and Mathematics Skills: The Report of Sir Gareth Roberts’ Review. HM Treasury, London
(2002)
4. Becker, F.S.: Why don’t young people want to become engineers? Rational reasons for
disappointing decisions. Eur. J. Eng. Educ. 35(4), 349–366 (2010)
5. Henriksen, E.K., Dillon, J., Ryder, J. (eds.): Understanding Student Participation and Choice
in Science and Technology Education. Springer, Dordrecht (2015)
6. Sjøberg S., Schreiner C.: The ROSE Project: An Overview and Key Findings. Technical
report, University of Oslo (2010)
7. Mann A, Oldknow A.: School-industry STEM links in the UK: a report commissioned by
Futurelab. Education and Employers (2012)
8. Science, Technology, Engineering, and Mathematics Network: Top 10 Employability Skills.
http://www.exeter.ac.uk/ambassadors/HESTEM/resources/General/STEMNET%
20Employability%20skills%20guide.pdf
9. Kudenko, I., Gras-Velázquez, À.: The Future of European STEM Workforce: What Secondary
School Pupils of Europe Think About STEM Industry and Careers. In: Papadouris, N.,
Hadjigeorgiou, A., Constantinou, C. (eds.) Insights from Research in Science Teaching and
Learning. Contributions from Science Education Research, vol. 2, pp. 223–236. Springer,
Berlin (2016)
10. Adams, J., Kaczmarczyk, S., Picton, P., Demian, P.: Problem solving and creativity in
engineering: conclusions of a three year project involving reusable learning objects and robots.
Eng. Educ. 5(2), 4–17 (2010)
11. Robinson, K.: RSA animate—changing education paradigms (2008). http://www.youtube.
com/watch?v=zDZFcDGpL4U
12. Fredricks, J.A., Blumenfeld, P.C., Paris, A.H.: School engagement: potential of the concept,
state of the evidence. Rev. Educ. Res. 74(1), 59–109 (2004)
13. Samuels, P.C., Maitland, K.: Redefining maths learning technologies: putting the curriculum
into the fun. In: 1st HEA Annual Conference on Aiming for Excellence in STEM Learning
and Teaching. Higher Education Academy, London (2012)
14. Haapasalo, L., Samuels, P.C.: Five recommendations for mathematical learning technologies
from the learner’s perspective. Submitted to Ed. Tech. Res. & Dev
15. Melchior, A., Cohen, F., Cutter, T., Leavitt, T.: More than Robots: An Evaluation of the
FIRST Robotics Competition Participant and Institutional Impacts. Heller School for Social
Policy and Management, Brandeis University, Waltham, MA (2005)
16. Benitti, F.B.V.: Exploring the educational potential of robotics in schools: a systematic review.
Comput. Educ. 58(3), 978–988 (2012)
17. Kandlhofer, M., Steinbauer, G.: Evaluating the impact of educational robotics on pupils’
technical—and social-skills and science related attitudes. Rob. Auton. Syst. 75, 679–685
(2016)
Developing Extended Real and Virtual Robotics Enhancement … 81

18. Eronen, L., Haapasalo, L.: Making mathematics through progressive technology. In: Sriraman,
B., Bergsten, C., Goodchild, S., Palsdottir, G., Dahl, B., Haapasalo, L. (eds.) The First
Sourcebook on Nordic Research in Mathematics Education, pp. 701–710. Information Age,
Charlotte, NC (2010)
19. Samuels, P.C.: Motivating mathematics learning through an integrated technology enhanced
learning environment. Int. J. Tech. Math. Educ. 17(4), 197–203 (2010)
20. Lawrence Sheriff School: The Griffin Teaching School Alliance. http://www.
lawrencesheriffschool.net/downloads-all/category/21-national-teaching-school?download=
893:the-griffin-alliance-portfolio-oct-2015
21. Jaworski, B.: Challenge and support in Undergraduate Mathematics for Engineers in a
GeoGebra Medium. MSOR Connect. 10(1), 10–14 (2010)
22. Jaworski, B.: Developmental research in mathematics teaching and learning: developing
learning communities based on inquiry and design. In: Liljedahl, P. (ed.) Proceedings of the
2006 Annual Meeting of the Canadian Mathematics Education Study Group, pp. 3–16.
University of Calgary (2006)
23. Haapasalo, L., Samuels, P.C.: Responding to the challenges of instrumental orchestration
through physical and virtual robotics. Comput. Educ. 57(2), 1484–1492 (2011)
24. Papert, S.: Mindstorms: Children, Computers and Powerful Ideas. Basic Books, New York
(1980)
25. Burdea, G.C.: Invited review: the synergy between virtual reality and robotics. IEEE Trans.
Rob. Autom. 15(3), 400–410 (1999)
26. Catlin, D., Blamires, M.: The Principles of Educational Robotic Applications (ERA): a
framework for understanding and developing educational robots and their activities. In:
Clayson, J.E., Kalas̆, I. (eds.) Proceedings for Constructionism 2010: the 12th EuroLogo
Conference, Paris (2010)
27. Eisenberg, M.: Mindstuff: educational technology beyond the computer. Convergence 9(2),
29–53 (2003)
28. Kolodner, J.L., Crismond, D., Gray, J., Holbrook, J., Puntambekar, S.: Learning by design
from theory to practice. Proc. Int. Conf. Learn. Sci. 98, 16–22 (1998)
29. Cropley, D.H., Cropley, A.J.: Fostering creativity in engineering undergraduates. High Abil.
Stud. 11(2), 207–219 (2000)
30. Strijbos, J.W., Martens, R.L., Jochems, W.M.: Designing for interaction: six steps to designing
computer-supported group-based learning. Comput. Educ. 42(4), 403–424 (2004)
31. Atmatzidou, S., Demetriadis, S.: Evaluating the role of collaboration scripts as group guiding
tools in activities of educational robotics: conclusions from three case studies. In: IEEE 12th
International Conference Advanced Learning Technologies (ICALT), pp. 298–302. IEEE
(2012)
32. Topping, K.J.: Trends in peer learning. Educ. Psych. 25(6), 631–645 (2005)
33. Samuels, P.C., Haapasalo, L.: Real and virtual robotics in mathematics education at the
school-university transition. Int. J. Math. Educ. Sci. Tech. 43(3), 285–301 (2012)
34. Rinderknecht, M.: Tutorial for Programming the LEGO® MINDSTORMS™ NXT. http://
www.legoengineering.com/wp-content/uploads/2013/06/download-tutorial-pdf-2.4MB.pdf
35. Samuels, P.C.: Animation of a Robot Moving through Three Points. http://www.geogebra.org/
material/simple/id/2807809
36. Jones, A., Issroff, K.: Learning technologies: affective and social issues in computer-supported
collaborative learning. Comput. Educ. 44(4), 395–408 (2005)
37. Costa, S., Resende, J., Soares, F.O., Ferreira, M.J., Santos, C.P., Moreira, F.: Applications of
simple robots to encourage social receptiveness of adolescents with autism. In: 31st IEEE
Engineering in Medicine and Biology Society Conference, pp. 5072–5075. IEEE, Minneapolis
(2009)
38. Dautenhahn, K., Werry, I.: Towards Interactive Robots in Autism Therapy: Background,
Motivation and Challenges. Pragmat. Cognitive 12(1), 1–35 (2004)
Project Oriented Approach in Educational
Robotics: From Robotic Competition
to Practical Appliance

Anton Yudin, Maxim Kolesnikov, Andrey Vlasov and Maria Salmina

Abstract The paper shows a way of effective organization of education and student
motivation in personal digital fabrication environment. Student activity within the
approach is divided in two tracks. The paper concentrates on the second, real-world
project oriented track with a series of diagrams and a case study.

Keywords Educational robotics ⋅ Project-oriented education ⋅ Mobile robotics ⋅

Mechatronics ⋅ Engineering ⋅ Human capability development

1 Introduction

The current state of education in the world confronts school, industry and science
with an urgent task to find effective ways of knowledge transfer to future genera-
tions taking into account the real constraints of our time. Urgent need of changes
in education worldwide can and should be considered along with the global crisis
in other fields of human activity: economic, political, environmental, etc. Recently
emerged movement of digital fabrication laboratories spread around the world [1]
could be part of the answer we are looking for. Technological advances allowed us
to make another evolutionary step in industrial production which now becomes per-
sonal again [2]. Remembering the vast trade shop experience of the past it is natural
to adapt the best traditions of manufacturing organization. Of course this also leads
to the change in technical education.
Affordable portable electronics which appeared several decades ago allowed
enthusiasts to work on numerous technical projects. This led to many breakthroughs
in device design. One of those resulted in affordable and hackable 3D-printer con-
cept widely used in production labs today. This is an example of how small groups of

A. Yudin (✉) ⋅ M. Kolesnikov ⋅ A. Vlasov

Bauman Moscow State Technical University, Moscow, Russia
e-mail: [email protected]
URL: http://www.bearobot.org
M. Salmina
Lomonosov Moscow State University, Moscow, Russia

© Springer International Publishing Switzerland 2017 83

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_8
84 A. Yudin et al.

interested people can address problems big and small more efficiently and granularly
and push technology forward. Such additive technology machinery uses the Planet’s
resources much more effectively than subtractive ones.
Though computers are widely used today their use most of the time is too far from
what they are capable of or what they were meant to provide to human beings and
children in particular [3]. This power to amplify human’s mental abilities is still to
be rediscovered for education [4] and for everyday life. Today when computers come
closer to people in personal digital fabrication this step becomes obvious and brings
attention of many young engineers to the problem of productive software for makers
versus systems capable of reproducing old media in a new fashion.
Another way to improve the current situation with education could be in address-
ing real multidisciplinary team projects. Due to project learning technologies and
visual methods of system design it is possible to effectively integrate real engineer-
ing tasks in educational programs [5, 6].
The paper shows a way of effective organization of education and student motiva-
tion in personal digital fabrication laboratory environment. The educational process
being reviewed considers continuous progressive consistent approach complement-
ing classic elementary school, secondary school and university, uniting and extend-
ing knowledge obtained within these common means. Results of such an educational
process are presented with a case study.

2 Project Oriented Approach to Education

The authors present supplementary education approach and emphasize the possibil-
ity of pairing knowledge obtained in ordinary nowadays’ schools with practice of
actual real-world device design.
As shown in Fig. 1 the main track to gain a competence level for a newcomer is
through competition projects. Mobile robot competition is a tool for basic project-
based engineering education. Tasks for students are drawn from the competition’s
rules. The choice of a mobile robot as an object for study is due to possible project
difficulty versatility.
Competition projects’ track could be described as intensive training for intellec-
tual sport’s competition event. Appropriate age for a student to start a normal study
process for this track is 10 years old. For expected results the minimum work pace is
2 study sessions per week (each session is 2 h long).
Junior student start age is around 8 years old. Juniors would normally double the
time needed to take the competition projects’ track before they are ready for individ-
ual work resulting in 2–4 years of such studies.
After a year or sometimes two a student is able to form his own project ideas and
be involved in parallel individual project’s track. This leads to more working hours
and higher overall competence achieved. The projects results often carry economical
value but still show room for education continuation within the digital fabrication
laboratory environment.
Project Oriented Approach in Educational Robotics: From Robotic Competition . . . 85

Fig. 1 Project oriented approach to education

Individual projects’ track could be described as real-world project training with

all the limitations, deadlines and project management seen in industry’s everyday
work. The minimum work pace is 2 study sessions per week (each session is 2 h
long). These sessions are independent of other tracks’ sessions meaning their time
adds up to the competition track study time.
It is important to mention that both tracks form enough knowledge for a student to
start off his career even without finishing the expected 4 years of studies. As shown
in Fig. 1 after approximately 2 years he is able to work with machinery as a service
man or consult customers at an electronic components’ store. After 3 years he is able
to organize simple work-flows and work with simple technical designs. After 4 years
of practice he is a qualified engineer able to solve complicated technical cases and
even lead such projects if he was hard-working enough in studies.
It is important to mention that by “a qualified engineer” authors mean practical
aspects of the profession. The student is able to understand how to solve complicated
complex technical tasks, see the direction to take to actually succeed in development
but of course he still lacks university knowledge of in-depth subjects to do that sci-
entifically correct and effective. Simply put, the student by himself is able to choose
the right university and the technical area for his further studies and is ready to enter
the chosen institution at the moment of successful graduation from the discussed
course.
The competition projects’ track was described previously by the authors [7–9].
There is no need to repeat that work as the interested reader could always address
the mentioned work.
Figure 2 shows a typical individual project’s evolution in the described context
and study subjects required or involved in the process. Projects in this track are based
on robotics which is multidisciplinary in nature. But since the developed projects’
86 A. Yudin et al.

Fig. 2 Typical individual project’s evolution

ideas are proposed by students and are inspired by the real world problems technical
disciplines are usually expanded with other sciences as well.
In this case we have an idea of making an exoskeleton which comes from stu-
dents. It has a lot of questions at first which might not be clear to those who propose.
On the next step it is discussed with a teacher and the core idea changes to “robot
arm” as it is a more adequate proposition in the given technological environment,
resource constraints and deadlines. Next step of decomposition gives each projects’
team member an individual task to perform in the given time. Some of those tasks
are design oriented, engineering ones and some are scientific. Finally at some prede-
fined time parts of the projects are assembled together to form a product. The cycle
could continue from decomposition stage to find better solutions or to contribute to
a bigger project goal.
In the presented particular case the time span for the project is 3 mon. In the end
of the project students could present a working gripper with several human-machine
interfaces. While the usual actuator worked fine (servo) the scientific proposition for
a new type of magnetic fluid based actuator was not ready and needed more time
to “translate” the “invented” physics principle to an engineered device. This could
become a goal for the next cycle of the project.

3 Approach Results and Practical Appliances

A real-world project developed by those who successfully finished the proposed

mobile robot competition training track will be presented below. The idea behind
this case study is to show how technical systems developed in a competition based
educational project can be used in further student work (which could be inside the
individual projects’ track or even outside the educational environment). This way
of thinking about their studies allows students to form united teams with a techno-
logical platform able to cope with many real-life tasks for starting their after school
career or business.
Project Oriented Approach in Educational Robotics: From Robotic Competition . . . 87

Fig. 3 Applying competition track’s results to real-world problems

Figure 3 shows possible cases divided into categories like “Simplification”,

“Modification” and “Use as is”.
“Simplification” category gathers projects which exploit mobile competition
robot’s systems developed by a team in the competition project’s track. Here the
robot’s tasks are practical but simple or carry additional educational value. Such
projects play an important role in forming the right educational environment when
students teach each other.
“Modification” category gathers projects which bring evolution to mobile com-
petition robot’s systems developed by a team in the competition project’s track. Such
projects allow getting a new perspective on how the robot’s systems could be trans-
formed in something useful but different from the logical point of view. An example
of such a case could be a CNC machine, using same motors and control electron-
ics, but requiring different mechanics and programming techniques. Such projects
usually add something new to the team’s unique technological platform.
“Use as is” category gathers projects which use more or less same systems from
robots developed by a team in the competition project’s track. Such projects show
how “powerful” the team’s technological platform is in real-life applications. These
projects of course could involve slight modifications but generally they are relatively
quick to develop compared to starting from scratch. Such a project we’ll discuss in
detail below.
Figure 4 shows typical mobile autonomous robots that participated to an interna-
tional competition.
The figure is flagged with sensor positions. Both robots use ultrasonic range find-
ers in their obstacle avoidance systems to track mobile opponent’s position and other
static field objects. This is a common competitions task that is solved in a similar way
by many participating teams.
Getting to know the sensor part of robotic systems in the competition track which
like in the presented case are ultrasonic based or any other part as well could lead to
88 A. Yudin et al.

Fig. 4 Typical mobile autonomous robots designed in a competition track

many real-world applications using the same physics. An example of such an appli-
cation is presented below as a case study.

4 Case Study: Waste Sensor

Garbage handling and collection is a big problem in most of the cities on our
planet. When it seems that everything around us has become digital - garbage is still
neglected on the global scale for some reason. Though existing electronics and sen-
sors show a very good potential to help in solving the problem of effective garbage
collection and saving valuable resources while optimizing the collection routes for
example. One of the typical pictures depicting the problem is shown in Fig. 5.
Building an autonomous range finder sensor with a wireless connection to Inter-
net, durable enough to work for several years without maintenance and cheap in pro-
duction could help in this case (Fig. 6). Such a sensor is placed in every garbage bin
forming a network of wireless devices. Each sensor sends its status (full or empty)
to the central server where all the routes for garbage collection are computed auto-
matically.
There are 2 design variants for the sensor unit: for mounting it on the garbage bin
lid and for mounting it on its wall.
This project was based on prior competition robot’s building experience. The
ultrasonic sensor was chosen for the range finder role. This system naturally comes
out of the robots presented previously in Fig. 4.
Project Oriented Approach in Educational Robotics: From Robotic Competition . . . 89

Fig. 5 Overfilled garbage bins in a street of Paris

Fig. 6 Autonomous garbage level sensor unit—2 design variants

The sensor unit consists of two main parts: a body containing all control electron-
ics with a battery unit (on the Fig. 7 to the right) and a base used to firmly attach the
sensor to the garbage bin (on the Fig. 7 to the left).
90 A. Yudin et al.

Fig. 7 Sensor unit composition

Fig. 8 Sensor unit detailed composition

Dimensions of the unit are determined by the size of the internal components: the
PCB with control electronics and communication antenna’s height (Fig. 8). Ultra-
sonic range-finder is placed with force fitting in the top part of the body in a special
deepening. The base has 3 holes for bolt fixing it to the bin. For better hermetic seal-
ing between the body and the base a rubber ring can be placed as well as a rubber
pad between the base and the garbage bin.
The sensor body’s form has no corners or protruding parts and it was chosen
based on heavy usage condition in a garbage bin (the sensor must not be damaged
or knocked down, and must not prevent garbage from its normal movement during
filling/emptying the bin). The PCB is fixed to the body with 2 screws.
The base allows to firmly attach the entire sensor to the bin and at the same time
serves as a lid covering the hole in the body and hermetically sealing it.
Mounting of the sensor body to its base occurs by means of a threaded connection.
A special locking screw (Fig. 9) is introduced in the bottom of the body in order
to prevent undesirable unwinding of the sensor’s parts (accidental or intentional).
Project Oriented Approach in Educational Robotics: From Robotic Competition . . . 91

Fig. 9 Locking screw in the

sensor body to prevent
unwinding

The sensor unit fabrication is a complex process. In the presented case it was
simplified to match those main machines which are in possession in a digital fabri-
cation laboratory of the development team: 3D-printer, laser cutter, precise milling
machine. Finally the unit body and base are meant to be produced with moulding
and casting.
3D-printing requires a computer modelling step. The sensor base and body are
developed within this stage. It is important to consider technological constraints of
both 3D-printing hardware and later moulding and casting process.
Moulding and casting could be an expensive operation and that is why it is impor-
tant to separate it from the better form search which could be done easily on a 3D-
printer. In the presented case there were 4 stages of the sensor unit’s outlook evolu-
tion (see Fig. 10).
While 3D-printing is a convenient means of prototyping it is important to consider
the time needed to produce the print. For the sensor body in this project it takes about
21 h (Fig. 11), the base—14 h. This makes the printing to be on the critical path of
the project realization.
Electronics is prototyped on a precise milling machine called modella. The process
of electronics development is carried out in parallel to other processes of the sensor’s
fabrication. In this case electronics design also went through a 4-stage evolution (3
first stages are shown in Fig. 12). The final stage result is meant to be produced in
the classic way with a factory order.
The development process took about 3 mon to finish the prototype for the cus-
tomer. Most of the work was done in the field of 3D-printing (modelling and fabri-
cation) and in the field of electronics design to meet the requirements. Programming
was one of the steps which happened easier than others as it was very similar to prior
robotic applications built by the team.
While most of the work (like the electronics design and programming) was made
by senior students and graduates of the discussed course, the project was successfully
92 A. Yudin et al.

Fig. 10 The sensor unit outlook evolution

Fig. 11 3D-printing parameters and technological restrictions

used in the educational process for all groups of students in the developing educa-
tional lab. A lot of original ideas like the “rotational” design of the case and ability
to measure with ultrasonic sensor were at first proposed and tested by 2nd and 3rd
year students and later adapted to the project’s needs by the core developing team of
seniors.
Project Oriented Approach in Educational Robotics: From Robotic Competition . . . 93

Fig. 12 Electronics design evolution

5 Conclusion

The paper summarizes and structures information on project oriented approach to

technical education which authors obtained in more than a ten-year period of skilled
engineer preparation within robotic competition teams. Some of the recent ideas and
constructs are included in the work to make another step towards the description of
a working approach to education.
Student activity within the approach is divided in two tracks, the initial motiva-
tion and main educational track is based on robotic competitions and thus is called
the competition projects’ track. The complementary competence developing track
is based on real-world application projects with economical value and possibility to
work with real customers. It is thus called the individual projects’ track.
Most of the paper is describing the individual projects’ track. In the competition
projects’ track the time is measured in years because each educational cycle to pre-
pare a robot for the competition takes 1 school year. In the individual projects’ track
the time is measured in periods of 3 mon which is a comfortable time scale for a
customer to wait for a desired prototype of his or her system.
Approach results and a typical system’s demonstration that could be easily derived
from the experience of building a competition robot are presented with a case study
of building a waste sensor for a commercial customer. Some of the other possible
uses of such a unique team’s technological platform are presented with a category
chart.
The paper fills another part of methodology description the authors started several
years ago and could be interesting to those seeking practical techniques of technical
education and engineer preparation in the emerging new world of widely spread
personal digital fabrication.
94 A. Yudin et al.

References

1. Gershenfeld, N.: How to make almost anything, the digital fabrication revolution, For-
eign Aff. 91(6) 2012. http://www.foreignaffairs.com/articles/138154/neil-gershenfeld/how-to-
make-almost-anything
2. Gershenfeld, N.: The third digital revolution, presented at the Solid 2014 Conference, San Fran-
cisco, CA, USA, May 2122, 2014. http://solidcon.com/solid2014/public/schedule/detail/35425
3. Kay, A.: The Real Computer Revolution Hasnt Happened Yet, Viewpoints Research Institute,
1209 Grand Central Avenue, Glendale, CA 91201, 2007. http://www.vpri.org/pdf/m2007007a_
revolution.pdf
4. Kay, A.: Thoughts About Teaching Science and Mathematics To Young Children, Viewpoints
Research Institute, 1209 Grand Central Avenue, Glendale, CA 91201, 2007. http://www.vpri.
org/pdf/m2007003a_thoughts.pdf
5. Shakhnov, V., Vlasov, A., Zinchenko, L., Rezchikova, E.: Visual learning environment in elec-
tronic engineering education. In: 2013 International Conference on Interactive Collaborative
Learning (ICL), Proceedings (2013)
6. Okunev, Y., Dovbysh, S., Lokshin, B., Salmina, M., Formalskii, A.: Programme scientifique
et educatif pour des eleves et des instituteurs de mecanique, de mecatronique et de robotique.
In: 9 Colloque Francophone de Robotique Pedagogique, ser. Pre-Actes, pp.141–143. La Ferte-
Bernard, France, 14–16 May 2007
7. Yudin, A., Sukhotskiy, D.: Startup robotics course for elementary school. In: Research and Edu-
cation in Robotics—EUROBOT 2010, ser. Communications in Computer and Information Sci-
ence, vol.156. Rapperswil-Jona, pp.141–148. Springer Berlin Heidelberg, Switzerland, 27–30
May 2010
8. Yudin, A., Sukhotskiy, D., Salmina, M.: Practical mechatronics: training for mobile robot
competition. In: Dessimoz J.D., Balogh R., Obdzralek D. (eds) “Robotics in Education, RiE
2015”, Proceedings of the 6th International Conference on Robotics in Education, RiE 2015,
HESSO.HEIG-VD, Yverdon-les-Bains, Switzerland, 20-23 May, 2015. Roboptics Editions,
pp.94–99. Cheseaux-Noreaz, Switzerland (2016). ISBN 978-2-9700629-5-0
9. Salmina, M., Kuznetsov, V., Poduraev, Y., Yudin, A., Vlasov, A., Sukhotskiy, V., Tsibulin, Y.:
Continuous engineering education based on mechatronics and digital fabrication. In: Dessimoz
J.D., Balogh R., Obdzralek D. (eds) “Robotics in Education, RiE 2015”, Proceedings of the 6th
International Conference on Robotics in Education, RiE 2015, HESSO.HEIG-VD, Yverdon-
les-Bains, Switzerland, 20–23 May, 2015. Roboptics Editions, pp.56–57. Cheseaux-Noreaz,
Switzerland (2016). ISBN 978-2-9700629-5-0
ER4STEM Educational Robotics
for Science, Technology, Engineering
and Mathematics

Lara Lammer, Wilfried Lepuschitz, Chronis Kynigos,

Angele Giuliano and Carina Girvan

Abstract Robotics is a popular vehicle to introduce young people to science,

technology, engineering and mathematics (STEM) with various approaches
worldwide that use robotics to teach or entertain or both, accompanied by various
tools and repositories. However, the stakeholders involved have different goals and
methods, thus difficulties in finding common ground. E.g. the focus in most cases is
on increasing interest in STEM, but research methods are unspecified or vague; or
despite the vastness of offerings, teachers are reluctant to incorporate activities in
the classroom. In this paper, we introduce the Educational Robotics for STEM
(ER4STEM) project that will realize a creative and critical use of educational
robotics to maintain children’s curiosity in the world. An open and conceptual
framework will bring three main stakeholders of educational robotics—teachers,
educational researchers and organizations offering educational robotics—together
through a user- and activity centered repository.

Keywords Educational robotics ⋅ Framework ⋅ Teachers ⋅ Repository

L. Lammer (✉)
ACIN Institute of Automation and Control, Vienna University of Technology, Vienna,
Austria
e-mail: [email protected]
W. Lepuschitz
PRIA Practical Robotics Institute, Vienna, Austria
e-mail: [email protected]
C. Kynigos
Educational Technology Lab, University of Athens, Athens, Greece
e-mail: [email protected]
A. Giuliano
AcrossLimits Limited, Hamrun, Malta
e-mail: [email protected]
C. Girvan
Cardiff University, Cardiff, Wales, UK
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 95

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_9
96 L. Lammer et al.

1 Introduction

Robotics is a popular vehicle to introduce young people to science, technology,

engineering and mathematics (STEM) with various approaches worldwide. The
most common aims of these activities are either the teaching of robotics and STEM
subjects or invoking young people’s interest in STEM fields and careers [1].
Consequently, various stakeholders are involved; e.g. teachers looking for inter-
esting ways to teach their subjects, researchers developing pedagogically informed
activities to foster certain skills, robotics enthusiasts offering fun and play, or
educational technology developers commercializing different tools.
In the current situation in Europe, the difficulty presents itself not so much in the
quality of the different offerings but rather on, firstly, evaluating their impact, and
secondly, bringing different stakeholders together for synergies. As a consequence,
the main stakeholder—the young people—is lost out of sight; e.g. many activities
engage already interested young learners but fail to attract others, technology has
become more important than the learning activity or educational robotics schedules
are fragmented and touch young learners only once if at all (e.g. [2, 3]). There is a
pressing need to exploit the multidisciplinary potential of robotics with a well
structured interdisciplinary approach to foster children’s curiosity, leading them to
entrepreneurial, industrial and research careers in STEM fields.

2 Educational Robotics

Robotics is an excellent tool for teaching science and technology [4], therefore
many educational robotics activities focus on STEM [1]. One of the most successful
and long lasting initiatives to promote STEM in Europe, especially focusing on
girls, is the Roberta Initiative [5] that uses special gender-appropriate teaching and
learning materials. Besides these, the long-term success of the initiative may be
attributed to the certification process of “Roberta teachers” and the repository with
access to a wide variety of materials. The main goal of the project is to engage and
motivate girls and boys to take a sustained long-term interest in STEM.
There are two other interesting former European projects. One is TERECOP
(Teacher Education on Robotics-Enhanced Constructivist Pedagogical Methods)
with the aim to develop a framework for teacher education courses in order to
enable teachers to implement the robotics-enhanced constructivist learning in
school classrooms, and report experiences from the implementation of this frame-
work [6]. This framework can be very helpful in designing activity plans and new
curriculums that enhance STEM and Educational Robotics education. The other is
Centrobot [7], with the aim to stimulate the interest of young people in technology
and research. The Centrobot project organized international competitions in the
Vienna-Bratislava region, two scientific conferences as well as three summer
schools and an exchange program within the educational sector. In that way
ER4STEM Educational Robotics for Science, Technology … 97

students engaged in robotics, learned about STEM and got involved in activities
with students from other countries.
Robotics competitions (e.g. First League or RoboCup Junior) are one of the most
widespread events regarding educational robotics activities. They offer a structure
for problem solving in group settings by encouraging focused hands-on problem
solving, team work, and innovation [8]. They are also social events where young
people from different countries meet each other. However, the focus is mostly on
technical problem solving, thus, certain types of students and teachers are attracted
to participate. Robotics workshops are one of the other most widespread activities,
maybe more diverse, addressing young people according to age, either in or outside
school settings. Some activities follow a curriculum; others are rather summer camp
activities. The main purpose is either the teaching of robotics or STEM subjects and
invoking young people’s interest in STEM fields and careers, which is followed and
documented more or less systematically depending on the organizer of the activity.
The above mentioned projects and initiatives in Europe involve different
stakeholders and address different aspects in educational robotics (Fig. 1). Young
people and teachers are often addressed as if each was a homogenous group, which
they are not. E.g. not all young people may be competitive or have talents in STEM
fields; or not all teachers may be talented tinkerers willing to read through tutorials
in repositories of technology tools. Organizers of educational robotics activities are
also a very heterogenous group with different backgrounds and motivations, e.g.
some may use robots as a motivational factor to entertain children, and the teaching
or invoking interest in STEM may be a by-product. On the other hand, educational
researchers may rather be interested in developing pedagogically informed activities
and measuring the impact in an empirical way, e.g. improvement of skills or change
of science related attitudes [9]. A closer look at the educational robotics landscape
reveals that the different stakeholder needs and requirements should be addressed in
a more structured way, and brought together under one framework to leverage
synergies.

Fig. 1 Educational robotics

stakeholders

Teachers

EducaƟonal
RoboƟcs for
Young
EducaƟonal People Organizers
Researchers of ER
acƟviƟes
98 L. Lammer et al.

3 The ER4STEM Project

The ER4STEM project sets out to create a continuous STEM schedule by lever-
aging on different already existing European approaches of innovative science
education methods and measures based on robotics within one open operational and
conceptual framework. Students aged 7−18 as well as their educators will be
offered different perspectives and approaches to find their interests and strengths to
pursue STEM education or careers through robotics and semi-autonomous smart
devices. At the same time students will learn about technology (e.g. circuits), about
a domain (e.g. math, physics, biology, psychology) and acquire skills (e.g. col-
laborating). New methods will be developed to achieve an integrated and consistent
concept that picks children up at different age levels starting with primary school to
accompany them until graduation from secondary school.
ER4STEM Framework
The ER4STEM framework builds the base of all activities and innovations offered.
It will create processes, tools and artefacts that allow the use of robots in learning
spaces and will be the catalyst to improve young people’s learning experience
through the use of robotics in formal and informal spaces. Different perspectives
inform the framework (Fig. 2): Workshops and curricula, conferences and com-
petitions, pedagogical design and innovations, and educational technologies and
repositories. The whole concept will undergo a rigorous evaluation.
Workshops and Curricula
Each partner develops specific content and organizes workshops each year that
provide multiple-entry points and facilitate a continuous STEM schedule. The
workshops cover three different age groups (7−10, 11−14 and 15−18) and will be

Fig. 2 ER4STEM structure

Workshops
and Curricula

Conferences
and
CompeƟƟons

EducaƟonal Framework
Technologies
and and
Repositories EvaluaƟon

Pedagogical
Design and
InnovaƟons
ER4STEM Educational Robotics for Science, Technology … 99

offered as units in schools or informally to be attended by young people during their

school-free time. More than 4000 students will be reached by the workshops. The
evaluation of these activities will flow into the framework and inform the design of
the generic process for development of workshops and curricula that empower
children.

Conferences and Competitions
Conferences designed with and for young people provide an opportunity to present
theoretical work and demonstrate artefacts to exchange ideas with each other and
meet robotics researchers and companies. The competition aspect completes these
activities by not only focusing on the best and quickest robots built by the student
teams, but also on the provided documentation and the quality of technical solu-
tions. The evaluation of the ECER1 conference will flow into the framework and
inform the design of conference and competition development.
Pedagogical Design and Innovations
There is a need to describe educational robotics activities analytically to become
more explicit and elaborate about pedagogical design and also have activities that
can be shared and compared. Parameters and criteria need to be defined to be able to
identify best practices and innovations in pedagogical activities. Tools to help
become more explicit and elaborate about pedagogical design will be developed
and tested.
Educational Robotics Technologies and Repositories
A user-centred repository will reflect the framework and focus on mapping the
processes needed by its stakeholders. The aim of bringing “added value” for the
users will help avoiding the common trap of “another great resource that is left
unutilized” and ensure sustainability after the end of the project. The repository will
be populated with a variety of open content made of different types of learning
objects and materials, links to outside resources, and educational tools built during
the project (Hedgehog and Slurtles prototypes). In order to do this, the seven
principal purposes of Open Educational Ressource (OER) repositories [10]
including quality, flexibility, ease of use and collaboration will be followed.
Evaluation
Both quantitative and qualitative data are collected across studies in educational
robotics. While qualitative data is collected in large scale studies (e.g. [11]), these
studies often lack rigorous and systematic analysis of the data and can become
anecdotal. There is little triangulation between data sources to increase reliability
and validity of claims made. In smaller scale studies there is evidence of qualitative
data analysis addressing this issue, e.g. [12]. The challenges of ER4STEM evalu-
ation will be to undertake high-quality analysis of qualitative data at scale to
evaluate the impact of the framework tools and activities on young people.

1
http://pria.at/en/ecer.
100 L. Lammer et al.

4 Conclusion

This paper introduced ER4STEM, a project that aims to realize a creative and
critical use of educational robotics to maintain children’s curiosity in the world
leading them to entrepreneurial, industrial and research careers in STEM fields, by
exploiting the multidisciplinary potential of robotics with a well structured inter-
disciplinary approach. The project will provide an open and conceptual framework
with processes, tools, and artefacts for educational robotics stakeholders to find
common grounds and ease collaboration between each other. The underlying
principles will be mapped in a user- and activity-centered repository that follows the
seven principals of Open Educational Resources [10].
The main stakeholders of the framework and repository have been identified
already: teachers, educational researchers and organizations offering educational
robotics activities. They are involved in the design process from the beginning to
the end. Their needs and requirements will build the use cases on which the
framework and repository will be designed and tested. The workshops and con-
ferences have also started in many project consortium countries. They have been
described with pedagogically informed activity plans and their evaluation results
will also inform the framework and repository development.

Acknowledgments The project has received funding from the European Union’s Horizon 2020
research and innovation program under grant agreement No. 665972. We would like to thank our
partners in ER4STEM project European Software Institute CEE and Certicon and all other con-
tributing members of the ER4STEM team.

References

1. Baretto, F., Benitti, V.: Exploring the educational potential of robotics in schools: a systematic
review. Comput. Educ. 58(3), 978–988 (2012)
2. Alimisis, D.: Educational robotics: open questions and new challenges. Themes Sci. Technol.
Educ. 6(1), 63–71 (2013)
3. Bredenfeld, A.H., Steinbauer, G.: Robotics in education initiatives in Europe—status,
shortcomings and open questions. In: Teaching Robotics-Teaching with Robotics—SIMPAR
Workshop, Darmstadt (2010)
4. Mataric, M.J.: Robotics education for all ages. In: Proceedings of AAAI Spring Symposium
on Accessible, Hands-on AI and Robotics Education, Palo Alto (2004)
5. Bredenfeld, A.H., Leimbach, T.: The roberta initiative. In: Proceedings of SIMPAR 2010
Workshops International Conference on Simulation, Modeling and Programming for
Autonomous Robots, Darmstadt (2010)
6. Alimisis, D.: Robotic technologies as vehicles of new ways of thinking, about constructivist
teaching and learning: the TERECoP project. IEEE Robot. Autom. Mag. 16(3), 21–23 (2009)
7. Balogh, R., Dabrowski, A., Hammerl, W., Hofmann, E., Petrovič, P., Rajníček, J.: Centrobot
Portal for Robotics Educational Course Material (2011)
8. Feil-Seifer, D., Mataric, M. J.: Defining socially assistive robotics. In: International
Conference on Rehabilitation Robotics (2005)
ER4STEM Educational Robotics for Science, Technology … 101

9. Kandlhofer, M., Steinbauer, G.: Evaluating the impact of educational robotics on pupils’
technical- and social-skills and science related attitudes. Robot. Auton. Syst. 75 Part B,
679–685 (2016)
10. Robinson, M.: Robotics-driven activities: Can they improve middle school science learning?
Bull. Sci. Technol. Soc. 25(1), 73–84 (2005)
11. Somyürek, S.: An effective educational tool: construction kits for fun and meaningful learning.
Int. J. Technol. Des. Educ. 25(1), 25–41 (2015)
12. Jung, I., Sasaki, T., Latchem, C.: A framework for assessing firness for purpose in open
educational resources. Int. J. Educ. Technol. Higher Educ. 13, 3 (2016)
 Part III
Design and Analysis of Learning
Environments
The Educational Robotics Landscape
Exploring Common Ground and Contact
Points

Lara Lammer, Markus Vincze, Martin Kandlhofer and Gerald Steinbauer

Abstract In the last decades, educational robotics has gained increased attention
evoking a need to discuss and document different approaches and lessons learned.
In this article, we report our findings made during the “Educational Robotics Café”,
a workshop format where experts engage in an open discussion about opportuni-
ties and challenges of the educational robotics landscape as well as advantages and
shortcomings of various approaches. Interestingly, participants working on different
educational robotics topics with different methods realized that all seemed to have
similar problems and experiences. They could define areas of common ground,
yet had difficulties in finding contact points between their educational robotics
approaches to compare them. Known categorizations seemed not to fit or to be too
high level. Based on these findings, we finish our article by suggesting a “tagging”
approach to enable better communication between experts from different domains
like education or robotics.

Keywords Educational robotics ⋅ Teacher ⋅ School ⋅ Society

1 Introduction

In the last decades educational robotics has gained increased importance and atten-
tion worldwide [1]. Many different educational robotics approaches and frameworks
exist [2], yet, teachers and educators face the problem of keeping an overview and

L. Lammer (✉) ⋅ M. Vincze

Automation and Control Institute, Vienna University of Technology, Vienna, Austria
e-mail: [email protected]
M. Vincze
e-mail: [email protected]
M. Kandlhofer ⋅ G. Steinbauer
Institute for Software Technology, Graz University of Technology, Graz, Austria
e-mail: [email protected]
G. Steinbauer
e-mail: [email protected]
© Springer International Publishing Switzerland 2017 105
M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_10
106 L. Lammer et al.

finding the right ones for their needs. Consequently, it is important to discuss and
document concepts as well as lessons learned and failure stories. In order to stim-
ulate a vital discussion process among different researchers, teachers and experts
in the field of educational robotics, we conceptualized and conducted a workshop1
guided by a simple question: “Which approaches do actually work when teaching
children robotics?”

2 Educational Robotics Café

2.1 Method

To provide structure and guide discussions, we chose the combination of two method-
ologies. One is the World Café, a methodology of collaborative inquiry and learn-
ing with the aim to create knowledge. The café metaphor is used to bring the focus
to a space where dialogue, reflections and shared meaning—or “conversations that
matter”—happen [3]. The other is SWOT analysis, a strategic planning tool to evalu-
ate strengths and weaknesses, opportunities and threats. The resulting SWOT matrix
contrasts the results of the internal analysis (strengths and weakness) and the exter-
nal analysis (opportunities and threats) to define strategic fields of action [4]. The
goal was to encourage a dialogue between different experts to create the educa-
tional robotics “landscape”, first by mapping the environmental factors that either
pose opportunities or threats, then by placing different approaches and methods on
it, highlighting advantages and shortcomings. For instance when teachers’ fear of
complex technology poses a “threat” to educational robotics, it can be addressed for
example with plug and play technology, which can be simple in use (strength) but
expensive (weakness).

2.2 Procedure

Participants and Setting: In sum 13 experts from different fields concerning robot-
ics (researchers, teachers and students) participated in the workshop. Participants
were divided on four tables, where we tried to create a relaxed atmosphere with dif-
ferent tools to write and doodle. Coffee and sweets were in another room but could
be taken to the tables. The workshop lasted three hours plus half an hour break.
Pre-phase: Before the cafe discussions started, four researchers presented their
approaches to the audience that were reflecting different angles of the same goal:
“teaching children with robots”. While the approach A [5] concentrated on teaching

1
At the 6th International Conference on Robotics in Education (May 2015, Yverdon-les-Bains,
Switzerland).
The Educational Robotics Landscape Exploring Common Ground and Contact Points 107

natural science to primary school children in a holistic way by using robotics as a tool,
the approach B [6] brought together children from secondary school with children
from kindergarten and their grandparents to let them discover science via robotics,
computer science and artificial intelligence. Approach C [7] was very specific about
teaching children (as young as in elementary school) programming with a specific
robotics kit. Approach D [8] set secondary school children aged 11–13 into center
(with technology in the background) to let them discover their interests and strengths
while tackling real-world problems with the help of robotics.
Main-Phase: There were four rounds of questions and a final wrap-up round.
First round—Educational Robotics Landscape 1: Warming up where participants
meet and start identifying the landscape (defined in Sect. 2.1) with opportunities and
threats.

∙ What single memorable incident convinced you to teach robotics or why are you
teaching robotics to children?
∙ What is the potential of educational robotics to impact our society?
∙ What opportunities and threats can be identified?

Second round—Educational Robotics Landscape 2: Participants continue dis-

cussing the landscape.

∙ What are your specific challenges?

∙ What works really well?
∙ What depends on the environment?
∙ What factors influence your concept?

Third round—Educational Robotics Approaches 1: Participants analyse the four

presented approaches (see Sect. 2.2).
∙ Place the concepts A, B, C and D on the ER landscape
∙ What do method A, B, C or D have in common?
∙ In which aspects do method A, B, C or D complement each other?

At this point, we offered a categorization (a given set of categories preselected

by the authors) to the participants, so they could compare different approaches in
a structured manner. This did not work very well. Participants found the categories
too high level or generic, or they found more than one category fitting their approach
and were not sure which one to choose. Interestingly, they could also not come up
with contact points between the approaches themselves, besides age and school level.
The same problems occurred in the fourth round but it did not hinder the participants
from diving into deep discussions about educational robotics and its problems.
Fourth round—Educational Robotics Approaches 2: Participants continue the
analysis by adding other approaches

∙ Place your concepts on the ER landscape

∙ What does your method and A, B, C or D have in common?
∙ In which aspects does your method and A, B, C or D complement each other?
108 L. Lammer et al.

2.3 Analysis

At each of the four world café tables the table host noted findings and summaries of
the discussion rounds on a poster. Participants were also encouraged to take notes
using post-its and paper notepads. The final poster presentations as well as the con-
cluding feedback round with all workshop participants were audio recorded. Follow-
ing the workshop all written documents (posters, post-its, notes) were collected and
audio recordings were transcribed. The data was analyzed qualitatively [9] by struc-
turing outcomes of the workshop in various rounds. Four categories emerged from
this analysis (see Sect. 3).

3 Discussion

3.1 Robots

Robots are powerful tools to motivate children synthesizing knowledge, making it

tangible and meaningful. For instance when children code a robot, there is an imme-
diate physical result of the abstract task: the robot performs along the children’s
instructions. Additionally, teaching subjects are not limited to coding; with robots
children can learn about almost anything by trying out, making mistakes, exploring
and experiencing. However, the term educational robotics comprises robotic kits as
well as robots themselves. In some cases, robots have the role of a teacher or tutor,
in other cases, the robot or robotics kit is the tool to teach a concept or subject. It is
important to distinguish between these roles and areas of application, as well as the
complexity. The discussion among experts reveals that robotics kits and black box
approaches should be provided to younger children, older ages can handle robotics
platforms with more mechatronics (“white-box”) and dive deeper into detail.

3.2 Teachers and Schools

Teachers are the key for the deployment of educational robotics in classrooms.
Their involvement and attitude are critical; without them, educational robotics can-
not be introduced in classrooms. Some teachers think that children are not able to
learn about science and technology, especially on the kindergarten and elementary
level, others are afraid of technology, considering it too complex. Teachers are over-
whelmed with the vast materials offered, at the same time they lack proper train-
ing. In this matter, the expertise of educational robotics experts and researchers is
needed. Finally, environmental factors influence the spreading of educational robot-
ics in schools, e.g. money, resources, availability of materials, time constraints and
group size play a role. Robots serve different goals and are applied in varied ways
The Educational Robotics Landscape Exploring Common Ground and Contact Points 109

according to the school level. In pre-school and elementary, children use a lot of
imagination, thus robots have to be contextualized (e.g. using story telling). In junior
high school, students use robots as tools to learn different concepts; in senior high
school, the focus is building (mechanics and electronics) and coding “real” robots.
The university level continues with a more detailed and specialized understanding
of the basic robot building concepts. No matter which level, “learning by doing”
should be a central aspect. This can be enhanced by specifically designed curricula
for all school levels. Given the knowledge and cognitive capabilities of the students,
curricula are more needed at the elementary and junior high level to perform class-
room activities or workshops with educational robotics. For older ages the “learning
by doing” should be concentrated on a specific problem or task. Older students can
also be motivated by competitions and after-school activities. However, there should
be complementary motivating options, since competitions can also be demotivating
also or not appeal to every student.

3.3 Society, Politics and Media

Society, politics and media influence the use of educational robotics as a learning
tool. Robots are neutral; experts need to design activities in such a way that children
are expressing their thoughts, their ideas, their knowledge and their interests gender
independently. Children’s expectations are formed by society and media. It then can
be frustrating to work on a simple prototype that “only” drives around. This can be
countered by lab tours in universities where children see robots in demonstration to
understand what they can accomplish when they choose this field and stay on the
subject.

3.4 Method “Educational Robotics Café”

The Educational Robotics Café was a novel and fruitful way to hear and discuss in
an open atmosphere different approaches, opinions, experiences and perspectives
among the workshop participants from different countries. The experts rated the
method as helpful and the discussions as very interesting. They were amazed to find
out that, in principle, approaches from other colleagues focus on similar questions
and problems, although results and the way of doing things differ. They all agreed
that strong motivators and proofs were needed to convince important stakeholders
like teachers, school administrators, policy-makers, etc. The participants underlined
the importance of “making a case” together, and to foster cooperation among experts
and educators in educational robotics to transfer this knowledge to the stakehold-
ers. However, the difficulty presented itself in finding common elements that made
sharing or comparing possible. The approaches presented in the pre-phase were
not described in that way. It would have helped, had they been broken down into
110 L. Lammer et al.

Table 1 Overview of exemplary tags

Overall concept Strategy Setting Target group Structure
#top-down #classroom #kindergarten #exploration
#bottom-up #after-school #elementary #activity
#K12
Educational goals Focus Teaching goals Theory/Method
#STEM #robotics #inquiry-based
#literature #collaboration #project-based
skills
Materials Principle Software Coding concepts
#white-box #text-based #high-level
#black-box #visual interface #low-level
Evaluation General Methodology Focus Target group
#evaluated #quantitative #technical skills #student
#not evaluated #qualitative #interests #teacher

elements that many approaches or activities have in common. Finally, as authors, we

would like to underline the importance of the questions in the method Sect. 2.2. They
influence the direction of the conversations and the overall output. Also, we believe
that the SWOT analysis did not work well because participants could not find com-
mon points of the approaches to compare them. Then, we were amazed to see what
started as formal talks of scholars defending their own concepts to become an honest
discussion and sharing of experiences.

4 Conclusions

We introduced the “Educational Robotics Café”, a workshop format where experts

engage in an open discussion on the educational robotics landscape. One major find-
ing the experts from different domains and countries realized is that, in principle,
all have similar goals and experiences although results and the way of doing things
differ. As it unfolds, the educational robotics landscape becomes abundant, covering
different pedagogical methods, settings, age groups, tools, and activities. The work-
shop showed that there is a great need to discuss and share experience, yet known
categorizations to compare approaches may not break differences down into the nec-
essary micro level. The difficulty rather presents itself in finding common ground or
common elements that make sharing or comparing possible. We close our paper with
an outlook on how a comparison of different approaches can be made possible by a
“tagging” approach to enable better communication between experts from different
domains like education or robotics.
The Educational Robotics Landscape Exploring Common Ground and Contact Points 111

5 Outlook: Tagging Approach

Tags can cover key aspects on a micro level and have the advantage of serving as
attributes, thus more than one tag can be assigned to one approach and new tags
can be added if none seems to fit properly. Table 1 shows exemplary tags divided
into four groups. We are aware that this selection does not cover all key aspects
in educational robotics. Rather, we intend to stimulate discussion and foster a vital
knowledge exchange between researchers and practitioners in the field of educational
robotics worldwide.

Acknowledgments Thanks to the participants of the workshop at the RIE 2015 for their contri-
bution. This research has partly received funding from the FWF Science Communication Project
WKP42, “Schraege Roboter” (“crazy robots”) as well as from Land Steiermark (“Wissenschaft und
Forschung”).

References

1. IEEE Trans. Educ. 56(1) (2013). Special Issue on Robotics in Education

2. Catlin, D., Blamires, M.: Principles of Educational Robotic Application (ERA). In: Construc-
tionism Conference, Paris (2010)
3. Brown, J.: The world cafe: living knowledge through conversations that matter. PhD thesis. The
Fielding Institute (2001)
4. Pahl, N., Richter, A.: SWOT Analysis. Methodology and a Practical Approach. GRIN Verlagr,
Idea (2007)
5. Cacco, L., Morro, M.: The hedgehog: how to integrate robotics in natural science scenarios.
In: Workshop position paper at International Conference on Robotics in Education (RiE 2015),
Yverdon-les-Bains, May 2015
6. Kandlhofer, M., Steinbauer, G., Hirschmugl-Gaisch, S., Eck, J.: Children discover science:
robotics, informatics and artificial intelligence in kindergarten and school. In: Workshop posi-
tion paper at International Conference on Robotics in Education (RiE 2015), Yverdon-les-Bains,
May 2015
7. Koppensteiner, G., Vittori, L., Miller, P., Goodgame, S.: Teaching programming on the elemen-
tary level. In: Workshop position paper at International Conference on Robotics in Education
(RiE 2015), Yverdon-les-Bains, May 2015
8. Lammer, L., Vincze, M.: Crazy robots—Introducing children with different backgrounds and
interests to robotics. In: Workshop position paper at International Conference on Robotics in
Education (RiE 2015), Yverdon-les-Bains, May 2015
9. Diekmann, A.: Empirische Sozialforschung—Grundlagen, Methoden. Rowohlt, Anwendungen
(2007)
A Workshop to Promote Arduino-Based
Robots as Wide Spectrum Learning
Support Tools

Francesca Agatolio and Michele Moro

Abstract This paper discusses the design of a workshop in Educational Robotics

based on Arduino offered to fifty different classes in Italian schools, from primary
school to high school. We provided different examples with two robotic platforms.
We present some observations about the current conception of Educational Robotics
among teachers and students and about the related problems in its introduction in
schools. The paper includes a description of the proposed activities and a preliminary
evaluation of the results.

Keywords Educational robotics · Arduino · Learning support tool · Conception of

robotics

1 Introduction

This paper presents a workshop in educational robotics that is taking place this year
in Italy starting in February 2016 and ending in May 2016. The workshop is the
result of a collaboration between the Department of Information Engineering of the
University of Padova and the Italian association Gruppo Pleiadi1 whose mission is
the scientific dissemination. The reasons behind conception and design of this project
are:
1. a strong belief in the potential of educational robotics as a learning tool for
the improvement of psycho-pedagogical skills and as an innovative support for
traditional theoretical lessons [1–4];

1 http://gruppopleiadi.it/.

F. Agatolio · M. Moro (B)

Dipartimento di Ingegneria dell’Informazione, University of Padova,
Via Gradenigo 6/A-B, Padova 35131, Italy
e-mail: [email protected]
F. Agatolio
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 113

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_11
114 F. Agatolio and M. Moro

2. the awareness that, despite this potential, educational robotics still plays a mar-
ginal and sporadic role in schools [1];
3. the possibility to exploit some current favorable conditions: relatively low-cost
robots on the market and suitable for educational purposes; the previous experi-
ence of the involved researchers; a relevant demand of teaching/learning innova-
tion coming from the school system and families in Italy.
Nonetheless some limitations still remain and we decided to focus especially on the
two of them that we considered the most relevant:
• the lack of resources to buy the right quantity of robotic kits for the entire classes
[5];
• the reticence of many teachers, often due to the conception that robotics activities
are difficult for students but also too complex for their personal competences
[6, 7].
Keeping these points in mind, we started the project Officina Robotica2 willing and
hoping to spread a new, different conception of educational robotics.
The workshop that we will describe in this paper is only the first step of the project
and consists in two lessons of two hours each, dedicated especially to automation
aspects; the planned recipients of the lessons are fifty classes of different grades and
ages, with an average of 25 students for each class.
The main purpose of these lessons is to engage both students and teachers increas-
ing their curiosity about robotics, and to convey the positive conception of robots as
a powerful and amusing learning tool. In particular, we want to convince teachers
about these facts:
• robotics can be addressed with different levels of difficulty, from simple to very
complex;
• the presence of an artifact during the learning process increases the student engage-
ment and encourages active learning (a concept consistent with constructivism and
constructionism theories [3, 8]);
• during robotic activities a teacher has a different role compared to traditional
lessons: he/she is personally stimulated to find the solution/solutions collaborating
with the students and not imposing ready-made solutions;
• it is possible to have an affordable but fully operational laboratory with robots,
adaptable both to a computer-based lab and to the usual classroom;
• using robotics, it is very simple and natural to develop links with curriculum
subjects (educational robotics does not mean to study robots in technical details).
Regarding the students, the goals of the project are:
• to convey the concepts of sensor, actuator and microcontroller, and how they
interact with each other;
• to make them glimpse some practical applications of the theoretical knowledge;

2 http://www.officinarobotica.it/.
A Workshop to Promote Arduino-Based Robots … 115

• to inform the students of the existence of affordable robotic tools that can be further
explored out of school.
As better explained below, for the purposes of this workshop many reasons convinced
us to adopt Arduino-based robots. In spite of our previous experiences, the main
factor not to use the more common LEGO Mindstorms robot for the workshop was
the limited allocated time and the cost.

2 Structure of the Workshop

The workshop activity originates from the project Officina Robotica and is being
developed thanks to a collaboration between the Department of Information Engi-
neering of UNIPD and Gruppo Pleiadi.
The purpose of the activities is to introduce among teachers and students a new
and stronger perception of the wide possibilities educational robotics can convey in
school along with the idea that robotics is a tool accessible and ideal for all ages;
the project is therefore addressed to all school levels, from primary school to high
school.
Keeping this objective in mind, we presented the workshop to fifty classes among
thirty different schools, located in different cities of north (78 %) and central (22 %)
Italy, namely:
• twenty primary school classes (1st–5th grades);
• twenty junior high school classes (6–8th grades);
• ten high school classes (9–12th grades).
The project involved among 1250 students and 40 teachers. We selected the
schools according to the order in which they have requested to participate. They
were both public schools and private schools. Regarding high schools, almost all
the teachers who asked to participate belonged to technical schools. The age of the
students was highly variable, especially for primary school (40 % of the students
was 6–8 years old). Regarding junior high school and high school, among 90 % of
the students were 12–15 years old. Almost always the students who took part in the
workshop were not voluntary but they belonged to a class chosen by the teacher.
The workshop is organized into two lessons of two hours each, with a break of a
month between the first and the second. The first lessons took place from the second
half of February 2016 to the second half of March 2016. The second lessons were
scheduled in April 2016. For almost all the students who take part to these activities,
this is the first experience with robotics and also mostly for programming.
116 F. Agatolio and M. Moro

3 The Instrument

3.1 The Tool: Why Arduino

Arduino-based or compliant robots were our choice due to some important positive
aspects:
• relative easiness of construction;
• low-cost components, giving the students the possibility to replicate experiences
at home;
• a variety of usable programming environment (beyond the standard Arduino IDE,3
Ardublock,4 Scratch for Arduino,5 Snap for Arduino,6 Mblock7 ).
Arduino is specifically linked to the Maker philosophy: it is in fact especially suited
to the improvement of manual skills and active experimentation. The goal is to make
students reflect on the strong relation between the theoretical contents they just
learned and a direct real experience, in order to lead them to think about the world
in an informed and scientific way, as opposed to a “magical one” [9]. The rich set of
sensors and actuators compatible with Arduino is well appropriate to this purpose.
The cost-effective quality of Arduino was particularly appreciated in our case in
order to provide more kits during the lessons, one for every two/three students. It
is also a good premise to convince schools and single students to invest in robotics
and thus to promote its wider diffusion. Our experience has already showed also its
flexibility: Arduino is easily adaptable to different levels of competence and school
stages; it can be used both to introduce the fundamentals of robotics/electronics and
to develop more complex projects.
Concerning junior high schools and high schools, we opted for the usual Arduino
hardware (Arduino UNO, with a starter kit included); the Arduino electronic board
has to be assembled on a mobile platform, a choice we made with the hope to increase
the students involvement by showing them a structure closer to what we observed
their expectations about a robot were.
For the application of our workshop in primary schools we chose Mbot 2.4G
version, an Arduino-compatible platform which presents some further interesting
characteristic:
• the kit includes a set of sensors and actuators, some on board, some as additional
extensions on mini-cards that makes the interaction between the robot hardware
and the ambient clear, immediate, and partly referable to interactions common in
human beings;

3 https://www.arduino.cc/.
4 http://blog.ardublock.com/.
5 http://s4a.cat/.
6 http://s4a.cat/snap/.
7 http://www.mblock.cc/.
A Workshop to Promote Arduino-Based Robots … 117

• among the other solutions, it can be controlled through a customized version of

Mblock, a graphical environment based on Scratch.
Moreover it can be operated using an infrared remote controller and this makes
also suitable for little children (aged 6–7).

3.2 Programming Languages

One of the main issues in the realization of the workshop was the choice of the
programming language and how much time to allocate to the programming part. In
Italian schools basics of programming are taught only in some high schools, and we
are conscious that it is very difficult to introduce programming in just two lessons;
on the other hand, we firmly believe that robotics without any programming tasting
loses its deep meaning, so we decided to introduce a dedicated section in order to
show the fundamental role of programming in the robots’ behavior.
Eventually we chose the following block-based programming environments for
the different school levels:
• Primary school: Mblock (that it is based on Scratch)
• Junior high school: Scratch for Arduino (S4A)
• High school: ArduBlock
Considering that most students do not have any previous knowledge about pro-
gramming, we chose to mediate heavily this part.
In junior high school and in high school we use an introductory activity (the smart
lamp) to give a direct and simple demonstration. Students are first asked to replicate,
on their pc, a project with an already tested piece of code for a first familiariza-
tion. In the following activities we gave them driving sheet with a support for the
programming.
In primary school we make the choice to avoid so young students were challenged
by programming technicalities in order to focus on the robot behavior and on com-
mand rich semantics. Therefore the experience is first devoted to design robot actions
in terms of simple sequences of commands: this phase is supported by some work-
sheets illustrating in simple form the library of selected commands. These sequences
are preliminarily tested using a body syntonic approach (the teacher or one classmate
“executes” commands like a robot). Then the teacher codes the sequence using the
graphical environment and the students observe if the robot precisely reproduces
what is expected.
118 F. Agatolio and M. Moro

4 Activities

The collaboration with Vivigas&Power,8 which is supporting the Officina Robotica

project, asked to orientate the activities towards the energy theme. We considered
Arduino well adaptable to this theme.
The competences that we wanted to develop through the workshop are:
• understanding the structure of a robot and the relation between sensors, actuators
and the microcontroller;
• understanding the role of the human pilot in controlling the robot through the
programming language;
• understanding the fundamentals of building electronic circuits.
For the realization of the workshop we had to keep in mind that:
• many of the students do not have any knowledge in Arduino and robotics in general;
• the selected classes presented a highly variability in age and probably also in level.
Therefore, for each of the three school levels, we chose to design a large set of
activities different in complexity (both with regards the automation aspect and the
programming aspect). In this way, we can easily calibrate any lesson depending on
the level and the response of the different classes.
Anyway the first activities have the aim to make the students feel comfortable
using robots, so these activities are focused on the manual building of very simple
electronic circuits and do not involve any actual programming of a robot.

4.1 Primary School Activity Examples: “Macchina

Scribacchina” and an Example of Activity with MBot

For primary school, the first part of the workshop is strongly dedicated to improve
manual skills. For this we chose to start assembling some simple circuits using
LEDs, DC motors and battery. Then children could use this knowledge to build
the artifact “Macchina Scribacchina” (literally Scribbling Machine) [10] (Fig. 1), a
moving machine obtained by assembling a plastic glass with a DC motor and four
markers. The vibration of the motor is transferred to the glass and to the markers so
that they leave some circular traces on the path.
After this activity, we introduced the robot MBot. The first activity was to analyze
the behavior of “Nocturnal MBot”. In this case, students didn’t have to program the
robot but only to observe the result of the program. “Nocturnal Mbot” moves only
if the light intensity goes down a certain threshold. So the first impression is of a
magical artifact that dances and blinks only when it is dark and we usually cannot
see it [11]. In this way we want to introduce children to robotics starting from a
conception of a robot close to a toy. After this first impact we go to unravel the

8 http://www.vivigas.it/.
A Workshop to Promote Arduino-Based Robots … 119

Fig. 1 “Macchina
scribacchina”

Fig. 2 Smart lamp

reasons behind the robot behavior introducing a scientific explanation. The children
are introduced to the sensor and actuator concepts and their role in designing a robot
control program.

4.2 Junior High School Activity Example: The Smart Lamp

This is the first activity actually introducing to programming language. Keeping

in mind the saving energy theme, we create a lamp that lights up only if the light
intensity is low enough for the environment to be considered darkness (Fig. 2).
To realize this object, a simple LED and a photoresistor are sufficient; the aim of
the activity is to introduce the concept of analog input and the necessity to understand
sensor’s values and how to translate electric signal into something which is under-
standable to humans. The students meet for the first time some typical structures of
a programming language, i.e. creation and use of variables, the if/else command,
etc. The students are also encouraged to establish autonomously the threshold that
allows them to distinguish darkness from light. Moreover, they are induced to reflect
on the difference between the last activity, where the LED was not controlled by
anyone, and this activity, where the lightning of the LED is programmed. We want
120 F. Agatolio and M. Moro

Fig. 3 Playing with the light

to make explicitly clear that the behavior of any robot strongly depends on our given
instructions.
In junior high school, after the first demonstrative phase (the students have to
copy already working code), in the second part of the activity we want the pc to play
a musical “instrument” using the variations of the light sensor mounted on the robot
and using a little of mathematics (Fig. 3). In this activity we use the same circuit of
the “smart lamp” and students can concentrate only on the programming part. For
this exercise we give them a worksheet for the programming part, so they can reflect
on it. The reaction to the outcome is very enthusiastic: the students are rewarded
and they can see the potentialities of the robot. In particular, the introduction of
sound elements is very engaging for students who usually present serious learning
problems, giving them a chance of actively participate (e.g. choosing the musical
instrument, proposing tunes).

4.3 High School Activity Example: Ecological Wake up

and Sunflower Robot

In high school, after the introduction to programming through the “smart lamp”
activity (see Sect. 4.2), we proposed the “ecological wake up” activity. The students
used a photoresistor and a buzzer to create an alarm clock that starts playing when
there is a sufficient quantity of light. We provided them a driving sheet for the
programming and they had to complete autonomously the code. We noticed that
most of them found hard to distinguish between the setup phase and the loop phase
(Fig. 4).
The activity “sunflower robot” will be carried out in the second part of the work-
shop. We create with Arduino a robot that follows a light source. This project is
strongly related to the eco-save question and can be used to create a solar panel
that, during the day, moves following the sun. To implement the project, we use two
photoresistor oriented so that they form an angle of 90◦ . When the light source is
closer to one of the two sensors, the robot turns in order to stay in line with the source
A Workshop to Promote Arduino-Based Robots … 121

Fig. 4 An example of common error in programming

[12]. This project is strictly connected to practical needs and allows an immediate
understanding about the possibilities when using robotics and theoretical knowledge
in a daily context. This can be useful to improve the engagement of students and to
justify the effort of the activity.

5 Evaluation of the Workshop: Preliminary Considerations

5.1 The Evaluation Tool

The evaluation of the workshop will be done at the end of the second lesson, through
a short questionnaire that will be issued to the involved teachers. The questionnaire
aims at evaluating essentially three points:
• the conception about the potential of educational robotics as a learning support
tool;
• the intention to introduce an educational robotics curriculum;
• the level of satisfaction about the workshop.
122 F. Agatolio and M. Moro

The respondents are asked to answer both to closed questions (yes/not) and open
questions (“Do you think that educational robotics can be a useful support learn-
ing tool? Why?”, “Do you think about introducing some robotics activities in your
lessons? Why?”).
Regarding the students, we considered meaningless to evaluate any improvement
in learning after only two lessons. Therefore, we preferred to give them the opportu-
nity to elaborate and to reflect about the recent knowledge through a contest, that is
to design a robot with the purpose to help the ambient, using strongly the concepts
of sensors, actuators and energy in general. In junior high school and high school,
students can present an original project using an hardware like, for example, Arduino
and LEGO Mindstorms, or a report of an activity of the workshop. Instead, in primary
school, students can simply describe and draw the imagined robot. A contest is not
an evaluation tool but, using a contest, we would like:
• to have a general feedback about the understanding of the conveyed concepts;
• to induce students and teachers to carry on a part of the workshop autonomously,
without the external intervention.
To encourage the schools participations, we reserved a prize for the best project (a
kit of five Arduino for secondary schools and a kit of five MBot for primary schools).

5.2 The Role of the (Curricular) Teacher

One of the greatest problems we met introducing robotics during this workshop was
the rapid demoralization of students when they face a failure (especially at junior high
school and high school). This is probably due to the fact that many of the arguments
are introduced ex novo, but it could also be closely related to the different learning
approach of robotics compared to traditional lessons. In particular, at the beginning
students show a strong resistance to the trial-and-error approach that is underlying to
educational robotics: they seem to be afraid of the possibility to search for a solution
by themselves (they often ask the teacher for support even before making any attempt
to solve the task) and most of them don’t seem interested in spontaneously exploring
new solutions, unless the teacher compels them to do it. This is particularly evident
in female students who sometimes appear less engaged in the subject.
In view of the above, we consider the role adopted by the curricular teacher
fundamental for the positive involvement of the students [13, 14]. The teacher in
such a constructivist learning approach like the one described above does not work
as a teacher—authority that transfers ready knowledge to students—but rather acts
as an organizer, coordinator and facilitator of learning for students. He/she gives the
guidelines of the activity, provides to students with many materials for thoughts and
observes their learning process. His/her presence should be very discrete and his help
should be offered only when necessary. He/she should allow students to work with
creativity, imagination and independence [15]. Acting like this, the teacher should:
A Workshop to Promote Arduino-Based Robots … 123

• encourage the students to adopt an active role in the learning process, based on
direct experience without fear of making errors;
• help to build their self-confidence, trying to make them feel that his/her help is less
crucial and the stepwise refinement of solutions through a trial-and-error process
is the main guideline.
Finally, in order to increase their involvement, the teacher has to mediate the meaning
of the experience underlining both the links with reality and the connection with the
theoretical knowledge learned at school; in this way he/she realizes a mutual con-
tribution amongst robotics and theoretical lessons aimed at increasing the students’
interest.

5.3 Preliminary Results

Until now about ten classes have completed both the two lessons, but we can anyway
make some preliminary observations about the achieved goals.
In general, our perception of the project’s outcome is positive. Due to the diversity
of the involved schools and classes (grade, number of students in each class, partic-
ipation of the teacher), we can observe a different feedback regarding the robotics
conception that we try to convey. Of course in classes with a large number of students
(more than 25) it can be hard to take on such a workshop: numerous requests of help
and support can concurrently come from dfferent groups and this would suggest the
presence of more tutors. The risk of that is to observe a certain degree of discour-
agement. Even if it is not surprising, we registered the more enthusiastic feedback
in primary school (especially in students), while junior high school confirmed as
belonging to the most critical range of ages (in particular it resulted more difficult to
design activities with the right level of complexity using Arduino).
Regarding students, in general we noticed a sincere enthusiasm, despite some
difficulties (see Sect. 5.2); moreover some of them seemed to be interested in the
possibility to carry on autonomously the activity out of the school. Regarding teach-
ers, we perceived an high appreciation as well. This confirmed by the partial data
obtained by the questionnaire. On the basis of the opened questions and other general
observations, it results that the good level of satisfaction is due especially to:
• the engagement of their students in using robots;
• the awareness that educational robotics takes advantage of diverse skills, which
don’t usually come up in traditional education. In particular, they pointed out that
in many cases, among the students that better succeeded in the activity, there were
surprisingly students with an otherwise low school performance.
However, despite this positive reaction, several teachers revealed they are not inten-
tioned to introduce educational robotics in the curriculum because they are not con-
fident of their competences (Fig. 5). Also among the teachers that said yes to this
question, most of them declared to not feel able to do it without an external support.
124 F. Agatolio and M. Moro

Fig. 5 Partial results from the questionnaire

6 Conclusions

At the moment of writing, all the fifty selected classes have already taken part in the
first meeting of the workshop and about ten classes have taken part also in the second
meeting. All the schools involved in the workshop asked voluntarily to participate,
often due to a specific interest of teachers qualified in scientific subjects. In general,
our perception of the project’s outcome is positive: after an initial resistance, most
students felt rewarded by the results and seemed impatient to continue the experience
with a next lesson. The feedback of the involved teachers was positive, as well, and, in
particular, their conception about the potential of the educational robotics as learning
support tool seemed to be increased. Despite this, most of them not yet feel able to
carry out autonomously educational robotics activities.
A more complete evaluation of the workshop will be done after the end.

Acknowledgments We thank the power and gas society Vivigas&Power, that supports the Officina
Robotica project.
This work was also partly supported by the project: ERASM: Educational robotics as a validated
mindtool: methodology, platforms, and an experimental protocol, code: CPDA145094, funded by
the University of Padova.

References

1. Alimisis, D.: Educational robotics: open questions and new challenges. Themes Sci. Technol.
Educ. 6(1), 63–71 (2013)
2. Sullivan, F.: Robotics and science literacy: thinking skills, science process skills and systems
understanding. J. Res. Sci. Teach. 45(3), 373–394 (2008)
3. Mikropoulos, T., Bellou, I.: Educational robotics as mindtools. Themes Sci. Technol. Educ.
6(1), 5–14 (2013)
4. Eguchi, A.: Educational robotics for promoting 21st century skills. J. Autom., Mob. Robot.
Intell. Syst. 8(1) (2014)
A Workshop to Promote Arduino-Based Robots … 125

5. Avvisati, F., Hennessy, S., Kozma, R.B., Vincent-Lancrin, S.: Review of the Italian strategy for
digital schools. OECD Education, Working Papers 90, OECD Publishing (2013)
6. Tondeur, J., Braak, J., Sang, G., Voogt, J., Fisser, P., Ottenbreit-Leftwich, A.: Preparing pre-
service teachers to integrate technology in education: a synthesis of qualitative evidence. Com-
put. Educ. 59, 134–144 (2012)
7. Di Battista, S., Menegatti, E., Moro, M., Pivetti, M.: Introducing educational robotics through a
short lab in the training of future support teachers. In: 6th International Conference on Robotics
in Education, RiE 2015, Yverdon-les-Bains, Switzerland (2015)
8. Alimisis, D., Arlegui, J., Fava, N., Frangou, S., Ionita, S., Menegatti, E., Monfalcon, S., Moro,
M., Papanikolaou, K., Pina, A.: Introducing robotics to teachers and schools: experiences from
the TERECoP project. In: Clayson, J., Kalas, I. (eds.) Proceedings for Constructionism, pp.
16–20. Paris, France (2010)
9. Roy, D., Gerber, G., Magnenat, S., Riedo, F., Chevalier, M., et al.: IniRobot: a pedagogical kit to
initiate children to concepts of robotics and computer science. In: 6th International Conference
on Robotics in Education, RiE 2015, Yverdon-les-Bains, Switzerland (2015)
10. The Tinkering Studio. http://tinkering.exploratorium.edu/
11. Dickel, M., https://www.youtube.com/user/MechDickel
12. Alfieri, M., http://www.mauroalfieri.it/
13. Agyei, D., Voogt, J.D.: Exploring the potential of the will, skill, tool model in Ghana: predicting
prospective and practicing teachers use of technology. Comput. Educ. 56, 91–100 (2011)
14. Ertmer, P., Ottenbreit-Leftwich, A., Sadik, O., Sendurur, E., Sendurur, P.: Teacher beliefs and
technology integration practices: a critical relationship. Comput. Educ. 59, 423–435 (2012)
15. Alimisis, D.: Exploring paths to integrate robotics in science and technology education: from
teacher training courses to school classes. IJREA 2(2), 16–23 (2012)
Robotics in School Chemistry
Laboratories

Igor M. Verner and Leonid B. Revzin

Abstract This paper proposes an approach to integrated learning of robotics and

chemistry implemented in a laboratory course, in which high school students
majoring in mechanical engineering are involved in the development of laboratory
robotic devices and use them for chemical experiments. We consider learning
activities at different stages of the course and characterize students’ engagement
along the stages.

Keywords Educational robotics ⋅ Chemistry ⋅ Experiential learning ⋅ Labo-

ratory automation ⋅
High school

1 Introduction

Modern chemical research and development laboratories widely exploit automatic

devices, hence many universities teach students to use automation technology, and
some already stopped teaching traditional manual methods [1]. In line with this
progress, high schools are making efforts to modernize their science laboratories
and transform them into technology enhanced laboratory learning environments [2,
3]. Educators believe that new automation technologies can make laboratory
experimentation more efficient and accessible, providing significant educational
advantages, especially in facilitating constructivist inquiry-based learning [4] and in
fostering students’ higher order thinking skills [5].
Currently, many school laboratories are equipped with computerized systems
which provide automation of experimental data collection and analysis by means of
sensors and data loggers [5]. It should, however, be noted, that while these com-
puterized systems provide automated data management, all manipulation operations
required for chemical experiments are still performed manually. Budget constraints

I.M. Verner (✉) ⋅ L.B. Revzin

Faculty of Education in Science and Technology,
Technion – Israel Institute of Technology, Haifa, Israel
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 127

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_12
128 I.M. Verner and L.B. Revzin

limit opportunities for schools to purchase commercial automation systems. The

lack of automation of manual operations causes difficulties in organizing laboratory
experimentation that answers the safety, accuracy and timing requirements.
Therefore, it became necessary to find affordable solutions for the automation of
manipulations in school chemistry laboratories.
Our research proposes and investigates an approach in which students experi-
entially learn chemistry, using data management systems and self-constructed
computer-controlled devices for automation of basic manual operations. We
developed a learning environment which includes data loggers and sensors, robot
construction kits, and automation devices constructed by the authors and students.
We conducted studies of learning in the proposed environment for two categories of
high school students: 11−12th graders studying chemistry at advanced level, and 10
−12th graders majoring in mechanical technology. We assert that through practice
in construction and application of automation devices students can acquire
knowledge and skills in both chemistry and technology.

2 Conceptual Framework

The use of robotic learning environments in school science laboratories has rapidly
expanded in recent years. This trend has motivated research in educational robotics
and in the broader context of learning in technology enhanced learning environ-
ments (TELE). Educational processes in TELE are mainly in the form of con-
structivist inquiries that foster understanding concepts of science and technology,
and the development of research skills [4]. In such environments the students
construct knowledge in science and technology through practice in creating and
operating technological tools. Researchers point to the potential of TELE to foster
interest in learning [6] and present evidence on the increase of students’ motivation
[7]. Kim et al. [4] note that in technology enhanced learning, student motivation
should come together with teacher’s scaffolding. Linn [8] points that technology
enhanced learning environments can provide guidance to students.
Dori and Kaberman [9], based on their research findings, claim that TELE can be
especially effective in supporting learning of low achieving students. According to
Girasoli and Hannafin [10], scaffolded practice in TELE promotes the development
of student’s self-efficacy.
New science curricula emphasize the need for personalized learning adapted to
thinking styles and learning strategies of each learner. Educators argue that through
the application of advanced technologies learning can be customized even for large
classes [11].
Most studies in which learning in school science laboratories is enhanced by
computer-controlled devices, consider situations when the devices are purchased
from manufacturers. Little research is about courses in which students under teachers’
guidance construct such devises and use them for science experimentation.
Robotics in School Chemistry Laboratories 129

The Technion Center for Robotics and Digital Technology Education has per-
formed a series of studies of integrated learning of robotics and physics [12, 13],
biology [14], and engineering [15]. These studies have been conducted in school
and Technion robotics laboratories. In the study presented in this paper we consider
a case of learning in school robotized chemistry laboratories. Some robotic devices
developed for automation of manual operations in chemistry experiments and
preliminary analysis of their educational effectiveness were presented in [16, 17]. In
this paper we focus on results of our educational study of students’ perception of
the automated laboratory environment and engagement in experiential learning.

3 Robotized Laboratory Environment

We enriched the standard school chemistry laboratory by sensors and data loggers,
robot construction kits, and devices for automation of basic manual chemical
operations. We presented in [16, 17] the first devices constructed by the authors and
the ways they were improved, based on students’ feedback. Based on this experi-
ence we engaged students in construction of new devices, some of which are
presented below.
The turntable (Fig. 1a) carries out rotations on given angles, the sensors
holder (Fig. 1b) puts the electrode into the solution and puts it up after the

Fig. 1 a. Computer controlled turntable; b. Computer controlled sensors holder;

c. Robot-dispenser; d. Robot-titrato; e. Automatic titration system
130 I.M. Verner and L.B. Revzin

experiment, and the peristaltic pump delivers a solution at the given computer
controlled rate.
Both computer controlled devices were constructed using LEGO NXT, each of
them by two 10th grade students majoring in technology. Figure 1d presents a
robot-dispenser for automatic aliquoting of chemical solutions. This robot was
constructed by other two tenth graders. The dispenser includes a turntable, a sensors
holder, and a peristaltic pump, all controlled by one LEGO NXT controller. Figure 1e
shows a robot for simultaneous automatic titration of a number of probes. It was
constructed by two 12 graders majoring in chemistry and computer science. In
addition to a turntable, a holder, and a pump, it also includes an automatic
sensor-washing device, and the Vernie pH sensor. The robot system is controlled by
two NXT controllers connected with each other via the Bluetooth communication.

4 The Study

The goal of our educational study was to investigate learning through practice in
constructing robotic devices and using them for studying high school chemistry
through laboratory inquiry. We were interested to explore our approach for teaching
chemistry at both advanced and basic study levels.
Two categories of participants of the research were: students from two com-
prehensive high schools studying advanced chemistry (172 students of grades
11−12, age 17−18), and students majoring in mechanics technology from a
vocational HS (12 male students studied our course in grades 11−12) and from a
comprehensive HS (14 male and 11 female students of grade 10, age 16). The
students in the former category were high achievers, while in the latter one, were
related to the group of youth at risk.
Our results related to the first category of students were presented in [16, 17]. In
this paper we will consider learning chemistry by technology students and focus on
the following research question: What are specific features of engagement of the
technology students in the course and what factors of engagement were dominant at
different stages of the course?
To answer the research question, we used questionnaires, observations and
interviews to follow up after changes in behavior of the technology students during
the course and their perceptions of the automated chemistry laboratory.

5 Learning Process

The learning process in the course given to two groups of technology students
consisted of the same eight stages described below.
Stage 1: Frontal teaching. During the first three weeks we tried this method, as it
is conventionally used in chemistry education. The method was found unsuitable
Robotics in School Chemistry Laboratories 131

for this category of learners, characterized by limited background, learning and

communication skills, and low motivation for learning science.
Stage 2: Practice in construction of automation devices. We put the study of
chemistry concepts aside and offered the students a constructive assignment: to
develop a LEGO-based automation device. Each student chose a device from the
given list and constructed it under the teacher’s guidance (Fig. 2a). Through this
practice the students learned about gears, mechanisms, and motors.
Stage 3: Device calibration experiments. The teacher posed to the students
questions on practical usability and accuracy of the constructed devices, and he
suggested to answer the questions experientially by calibrating the developed devi-
ces. He taught the students to use Fourier sensors and the data logger, as well as
laboratory devices, in order to do calibration experiments in the automated mode.
Here, for the first time, the students made a scientific experiment. On the
need-to-know basis they got familiar with physics concepts (velocity, force, pressure,
temperature, mass, volume, density) and chemistry concepts (atom, molecule, state of
matter, gaze laws). They learned to analyze experimental data by means of Excel.
For example two students made an experiment aimed to calibrate the computer
controlled peristaltic pump that they developed. They measured the mass of water
automatically pumped during certain time intervals, using the OhausB100 digital
balances (Fig. 2b). From the data analysis the students found that the solution
feed-rate for the given motor power is constant (see Fig. 3a). They also determined
the feed-rate dependency on the motor power, shown in Fig. 3b. During these
experiments the students had to help each other and this way developed partner-
ships that continued at the next stages.
Stage 4: Industrial tour. Each of the two student groups visited a chemical plant
and then participated in the discussion of the observed technological processes and
the underlying chemical phenomena. The teacher demonstrated simple experiments
related to the phenomena and initiated a discussion in which the students developed
ideas on how their devices can be used to perform the experiments in the automated
mode. At this stage the students acquired concepts of chemical reaction and its
equation, process control, and heat exchanger.

Fig. 2 Peristaltic pump models: a. Construction; b. Calibration experiment

132 I.M. Verner and L.B. Revzin

(a) (b)
water feed-rate, g/sec
mass, g 0.1
100

0.05
50 y = 0.081x + 63.416
y = 0.0012x - 0.0332
time, sec motor power, %
0 0
0 100 200 300 400 40 60 80 100

Fig. 3 Solution feed-rate: a. Determination; b. Dependency on the motor power

Stage 5: Concept development and experimental design. The teacher demon-

strated a series of chemical processes and posed the project assignment: to imple-
ment them using the automation devices developed by the students. The students
performed the projects in the groups of two. Each group was guided to formulate the
inquiry question, design an experiment, and specified needed materials.
Stage 6: Creation of an automated experiment setup. The students presented in
class the devices that they developed at the previous stages and discussed using
them in their projects. Based on these devices, each group designed, constructed,
and calibrated a modular robotic system for the assigned chemical process. The
implemented processes included titration, sedimentation, and neutralization of
dangerous materials.
Stage 7: Chemical inquiry. Each group of the students applied the developed
system to make a series of experiments in which they collected and analyzed data.
At this stage the students learned to program LEGO controllers, process data by
means of the MultiLab software. The students practically used mathematical con-
cepts of linear function, graph, slope, as well as chemistry concepts, such as sat-
urated and diluted solutions, acid-base indicator, chemical reaction speed, and pH.
Stage 8: Project report and presentation. The students analyzed results of the
experiments, and summarized the findings in the project reports. At this stage they
conceptualized the studied science and technology concepts. In order to get credit
for the project in the school matriculation certificate the students had to present it
for external examination of the Ministry of Education. The two groups of students
presented their reports as school graduation projects in mechatronics. One of the
groups also passed the matriculation exam in basic chemistry. Majority of the
students successfully passed the exams.

6 Learning Engagement

Students’ engagement is an important characteristic of learning and a predictor of

its outcomes. It characterizes student’s attention, interest and effort in the work of
learning [18]. Engagement is expressed through learning behavior, emotional
Robotics in School Chemistry Laboratories 133

Table 1 Structures of learning engagement [18, 19]

Engagement Description
structure
Get The Job Done The student’s motivating desire is to satisfy a sense of obligation to
complete an assigned task, to correctly follow the given instructions
Look How Smart I The motivating desire is to impress classmates and achieve positive
Am self-regard by demonstrating the ability, knowledge, and intelligence
Check This Out The motivating desire is to get a payoff or benefit
I’m Really Into This Here the desire emerges when the student is excited by the learning
activity
Don’t Disrespect The desire emerges when meeting a perceived threat to the student’s
Me dignity, status, or self-respect
Let Me Teach You The student desires to help to a classmate who faces difficulties in
performing learning tasks
Pseudo-Engagement The desire is to look good to the teacher or to peers by seeming to be
engaged

involvement, and intellectual effort. It can change during learning practice and
depends on its outcomes. Goldin et al. [18] proposed a theory of engagement
structures for systematic observation of typical student behaviors in class. They
presented a list of 12 engagement structures revealed through observation in
mathematics lessons. Verner [19] proposed to use the theory of engagement
structures in the studies of experiential learning in robotic environments. In this
paper we use these structures to follow-up the progress of student engagement in
our course. Among the twelve structures proposed by Goldin, seven were observed
in our case study. They are listed and described based on [18, 19] in Table 1.
From observations throughout the course and by applying the inductive method
[19] we found three additional new engagement structures:
I’ll Do Something Else The student’s motivating desire is to substitute a given
assignment by a personally meaningful one, without compromising the difficulty.
Don’t Want To Learn It Students’ reluctance to perform learning assignments
imposed against their will.
It’s Interesting To Discuss The students find the learning experience interesting
and desire to discuss it.
In order to follow-up the progress of learning engagement along the learning
process stages discussed in Sect. 5, we analyzed indications of presence of the ten
engagement structures. Our results are presented in Table 2.
The table indicates a significant behavioral change in the students. At the first
stage, when the course was delivered through conventional frontal teaching, the
students actually disengaged from the learning and even resisted it. At the second
and third stages, when the teaching method changed, the students had become
engaged in construction and calibration of automation devices and got a desire “to
do the job well” and even implement their own ideas. The industrial tour (Stage 4)
prompted the students to discuss with peers the seen chemical processes and this
134 I.M. Verner and L.B. Revzin

Table 2 Progress of learning engagement during the course

Stage

5.Concept development and

6. Creation ofan automated

2. Practice in construction
of automation devices

8. Project report and

3.Device calibration

experimental design

7. Chemical inquiry
1. Frontal teaching

experiment setup
4. Industrial tour

presentation
Experiments
Engagement structures
Pseudo-Engagement
Don’t Disrespect Me
Don't Want to Learn It
Get The Job Done
I'll Do Something Else
Check This Out
It's Interesting To Discuss
I’m Really Into This
Look How Smart I Am
Let Me Teach You

way contributed to building a learning community. The chemistry inquiry activities

at Stages 5−7 deeply engaged and excited the students. We note the multifaceted
impact of creation of the automated experiment setups (Stage 6) that prompted
students’ to demonstrate their mastery and help peers. At the final stage of the
course (Stage 8) the students showed persistence in preparing reports and presen-
tations satisfying the criteria of the Ministry of Education. Results of the external
examination of the students’ projects support the notable presence of the engage-
ment structure “Get The Job Done”. 80 % of the school students from the first
group and 91 % from the second group successfully completed their graduation
projects in mechatronics. 40 % of the first group students by their initiative took the
matriculation exam in basic chemistry and successfully passed it.

7 Conclusion

We developed, implemented, and evaluated an approach to integrated learning of

science and technology, in which high school students perform chemical experi-
ments in a laboratory enhanced by self-made robotic devices for automation of
manual operations. Our experience shows the considerable potential of this
approach for students studying advanced chemistry (presented in [16, 17]), and
Robotics in School Chemistry Laboratories 135

technology students studying a basic chemistry course. The reported case study
indicated that for technology students, to whom the conventional frontal teaching
was ineffective, the proposed approach, based on the development of robotic
devices and using them for chemical experiments, was engaging and contributed to
significant outcomes in learning chemistry. It should be noted, that in the recent
survey of chemistry education in technology enhanced laboratory environments our
work was the only one that studied the use of robotics for automation of manual
operations [20]. Based on positive results of the study, we have started experiments
of applying the proposed approach to teaching basic physics in a laboratory auto-
mated environment to 10th graders (age 16) majoring in mechanics.

References

1. Sterling, J.D.: Laboratory automation curriculum at Keck Graduate Institute. J. Assoc. Lab.
Autom. 9(5), 331–335 (2004)
2. Wang, F., Hannafin, M.J.: Design-based research and technology-enhanced learning
environments. Educ. Technol. Res. Dev. 53(4), 5–23 (2005)
3. Wu, H.K., Huang, Y.L.: Ninth-grade student engagement in teacher-centered and
student-centered technology-enhanced learning environments. Sci. Educ. 91(5), 727–749
(2007)
4. Kim, M.C., Hannafin, M.J., Bryan, L.A.: Technology-enhanced inquiry tools in science
education: An emerging pedagogical framework for classroom practice. Sci. Educ. 91,
1010–1030 (2007)
5. Barnea, N., Dori, Y.J., Hofstein, A.: Development and implementation of inquiry-based and
computerized-based laboratories: reforming high school chemistry in Israel. Chem. Educ. Res.
Pract. 11, 218–228 (2010)
6. Ainley, M., Hidi, S., Berndorff, D.: Interest, learning, and the psychological processes that
mediate their relationship. J. Educ. Psychol. 94, 545–561 (2002)
7. Mistler-Jackson, M., Songer, N.B.: Student motivation and Internet technology: Are students
empowered to learn science? J. Res. Sci. Teach. 37(5), 459–479 (2000)
8. Linn, M.: Technology and science education: starting points, research programs, and trends.
Int. J. Sci. Educ. 25(6), 727–758 (2003)
9. Dori, Y.J., Kaberman, Z.: Assessing high school chemistry students’ modeling sub-skills in a
computerized molecular modeling learning environment. Instr. Sci. Int. J. Learn. Sci. 40(1),
69–91 (2012)
10. Girasoli, A.J., Hannafin, R.D.: Using asynchronous AV communication tools to increase
academic self-efficacy. Comput. Educ. 51(4), 1676–1682 (2008)
11. Corcoran, T., Silander, M.: Instruction in high schools: the evidence and the challenge. Future
Child. 19(1), 157–183 (2009)
12. Verner, I., Ushin, I., Korchnoy, E.: Learning physical fields through operating robot
movements: a case study. In: Jamshidi et al. (eds.) Robotics, Manufacturing, Automation and
Control, vol. 14. TSI Press, Albuquerque, NM, pp. 383–388 (2002)
13. Korchnoy, E., Verner, I.: Characteristics of learning computer-controlled mechanisms by
teachers and students in a common laboratory environment. Int. J. Technol. Des. Educ. 20(2),
217–237 (2008)
14. Cuperman, D., Verner, I.: Learning through creating robotic models of biological systems. Int.
J. Technol. Des. Educ. 23(4), 849–866 (2013)
15. Verner, I., Hershko, E.: School graduation project in robot design: a case study of team
learning experiences and outcomes. J. Technol. Educ. 14(2), 40–55 (2003)
136 I.M. Verner and L.B. Revzin

16. Verner, I.M., Revzin, L.B.: Characteristics and educational advantages of laboratory
automation in high school chemistry. Special focus paper. Int. J. Online Eng. 7(S1), 44−49
(2011)
17. Verner, I.M., Revzin, L.B.: Automation of manual operation in a high school chemistry
laboratory: characteristics and students’ perceptions. The Chem. Educ. 15, 141–145 (2010)
18. Goldin, G.A., Epstein, Y.M., Schorr, R.Y., Warner, L.B.: Beliefs and engagement structures:
Behind the affective dimension of mathematical learning. ZDM Math. Educ. 43, 547–560
(2011)
19. Verner, I.: Characteristics of student engagement in robotics. In: Omar, K., et al. (eds.) FIRA
2013, CCIS 376, pp. 181–194. Springer, Heidelberg (2013)
20. Tortosa, M.: The use of microcomputer based laboratories in chemistry secondary education:
present state of the art and ideas for research-based practice. Chem. Educ. Res. Pract. 13,
161–171 (2012)
Breeding Robots to Learn How to Rule
Complex Systems

Franco Rubinacci, Michela Ponticorvo, Onofrio Gigliotta

and Orazio Miglino

Abstract Educational robotics has been extensively used to teach hard skills such
as computer science, computational thinking and coding because traditional robotics
is the outcome of analysis, design and programming. Other approaches to robotics,
namely evolutionary robotics, open the way to reflection on emergence,
self-organization, dynamical systems. As these issues are relevant in present days
society, we propose a robotic laboratory where children are trained to rule com-
plex systems. In particular, the integrated hardware/software system BrainFarm, that
allows to evolve and train virtual robots and then test them in physical environments,
is employed to train these skills and a successful experience in informal context is
described.

Keywords Dynamical and complex systems ⋅ Robotics lab ⋅ Evolutionary robotics

1 Introduction

Cultivating, breeding, training and teaching are activities that men employ to rule
and shape nature. Farmers, breeders, teachers and parents have always used their
ability to observe, follow and address the development and evolutionary pathways
of living beings, be it, plants, animals or children. Even if it is clear that a botanist
and a teacher have different expertise and adopt different techniques, they share the
ability to establish a connection and a dialogue between two autonomous entities.
They follow organisms’ history and intervene to correct, inhibit, facilitate behaviours
or morphologies. In this process the man does not control nature completely: every
organism has a certain amount of autonomy and therefore it can not be controlled,
but rather looked after.
On the opposite side, technology has developed effective techniques to build
machine which are completely different from the ones we have just cited and which

F. Rubinacci ⋅ M. Ponticorvo (✉) ⋅ O. Gigliotta ⋅ O. Miglino

Department of Humanistic Studies, University of Naples “Federico II”, Naples, Italy
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 137

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_13
138 F. Rubinacci et al.

are adopted by breeders. Machines are the final outcome of analysis, designing and
programming: no space for autonomy is left to machines, no unforeseen is allowed.
For many years breeders and engineers have had nothing in common. This is no
longer true as there are nowadays organisms that are designed and machines that are
bred. Biotechnologies allow to design living organisms to a certain extent, whereas
new approaches to technology have opened the way to emergence, self-organization,
dynamical systems. It can be extremely difficult to design a priori an effective solu-
tion to some problems which are unpredictable and require adaptation, change, flexi-
bility. In these cases it is necessary to provide machines with adaptation mechanisms.
In robotics many approaches have been proposed in this vein, such as adaptive
robotics [1], biorobotics [2], and evolutionary robotics [3, 4]. On the educational
side, robots have been extensively and successfully used to teach hard skills, related
to STEM [5]: in this case robots are designed, programmed and implemented. Robots
are machines which undergo an engineering process.
But, if the robot is seen as an organism that must adapt to the environment it oper-
ates in relying on its motor apparatus and the stimuli coming from the external and
internal environment, it is possible to consider it as a complex system to be ruled.
The present society requires children, the future citizens, to deal with self-organizing
entities, with dynamical systems, with emergence. In other words, together with hard
skills related to precise expertise, everyone will have to master transversal compe-
tencies, soft skills, including complexity management.
To help achieving this goal, breeding robot can be the main activity of a robotic
laboratory to train to manage complex systems. In what follows the robotic sys-
tem BrainFarm is described together with a successful experience in informal, non-
curricular educational context.

2 A Robotic System to Breed Robots: BrainFarm

BrainFarm and its prequels BreedBot and BestBot [6, 7] are integrated software/
hardware platforms that allow players, even without any programming or computer
skill, to breed, within customizable virtual worlds, artificial organisms that can be
downloaded onto real robots (Fig. 1).
The breeding is implemented through a user-guided genetic algorithm [8], based
on Interaction Evolutionary Design [9]. Evolutionary robotics aims at developing
robots taking inspiration from evolution in biology. Small populations of robots
undergo an evolutionary process to adapt to a particular environment and accom-
plish a task. The robots are reinforced with reinforcement learning algorithms [10]
which model learning theories developed in psychological research. Interaction Evo-
lutionary Design allows the user to realize objects continuously interacting with a
software that proposes many variants of the object itself.
Therefore the software side of BrainFarm shows users a population of nine
wheeled robots. Every robot is provided with some infrared sensors, to detect nearby
obstacles, and two motors that control wheels speed. A differential drive system
Breeding Robots to Learn How to Rule Complex Systems 139

Fig. 1 The BrainFarm interface with custom neural networks

allows robots to steer in any directions. Each robot is controlled by a simple feedfor-
ward neural network, representing an artificial nervous system [11]. The input layer
consists of infrared sensors and two motor context units that are simply two relay
units of the previous motor activation (all inputs are normalized between 0, 1). The
input neurons are connected to the output layer made up of two motor neurons that
control the right and left motor of the robot. Robots speed is updated according to
motor neurons activation. The neural network parameters are in turn encoded in a
genetic string that will undergo an evolutionary process [12] guided by either the
users (artificial selection) or the machine (automatic selection). In this latter case the
player can anyway manipulate the relevant evolutionary variables.
The player can affect directly the evolutionary pathway acting as a breeder that
selects the preferred agents or observe the possible evolutionary outcomes depend-
ing on different conditions. In this latter case, by manipulating directly the para-
meters related to the genetic algorithm the player can understand the underpinning
dynamics and experience different evolutionary pathways in a controlled environ-
ment. Moreover BrainFarm allows to design the robots brain architecture. Users can
use a simple feedforward network or more complex architectures to control robots
behavior, as shown in Fig. 1. This allows to introduce users to brain structure and
dynamics, through artificial neural networks models.
140 F. Rubinacci et al.

These platforms have been used to teach evolutionary biology [13], that is to say
a hard skills, but they are fit to soft skills training too. In particular BrainFarm can
be a gym where children can receive a training to complex systems management. An
experience about this approach is introduced in next section.

3 The BrainFarm Lab with Children

In this section we describe a successful laboratory experience that was held in

Naples, at Citta’ della Scienza. This is a cultural initiative to promote and popu-
larize scientific knowledge. Citta’ della Scienza has a multifunctional structure with
an interactive scientific museum and a training centre. In this latter centre many ini-
tiatives allow children to take part to laboratories in an informal context.

3.1 Partecipants

The BrainFarm lab lasted from January 2016 to February 2016 and involved 10
classes of about 25 student accompanied by 2 teachers.The BrainFarm lab was
attended by 245 children with an average age of 11 years. 20 teachers took part to
the lab together with students.

3.2 Lab Scheduling

The BrainFarm lab was arranged as depicted in Fig. 2: first of all a pre-lab ques-
tionnaire was administered both to teachers and students to assess their expecta-
tions about the laboratory. In particular teachers and students expectations about the
chance to learn complex systems management were investigated. The lab itself lasts
75 min: after a theoretical introduction about robotics, algorithms and interaction
design, children can build a simple rover bot according to a suggested morphology.
Then the task to be accomplished is introduced. It is a navigation task that foresees
a starting point, some areas to be reached and the target reaching. The environment
is built in the physical world and then reproduced in the software.
The group is divided into 4 sub-groups that must interact with the BrainFarm
software to select and evolve the best robots according to the Interactive Evolutionary
Design introduced above. In other words they select the best robots on the screen and
allow them to reproduce. This way a new generation of robots, similar to the selected
ones but with some differences, can be observed and undergo the selection process.
The robot morphology does not change along the process, but the parameters of the
artificial neural networks are modified.
Breeding Robots to Learn How to Rule Complex Systems 141

Fig. 2 The BrainFarm lab

schedule

The students then test their selected robots by transferring the simulated control
system into the physical robot in the physical environment. They act as breeders that
allow to reproduce only the best exemplars, according to the proposed solutions.
At the end a post-lab questionnaire is administered to teachers and students to
verify if the lab was effective, in their opinion, to train about complex system man-
aging.

3.3 Comments on BrainFarm Lab

In this section we will report some preliminary results about the lab experience at
Citta’ della Scienza. Teachers and students liked the lab very much, 84 % of stu-
dents rated the lab as very interesting on a Likert scale from ‘not interesting at all’
to ‘very interesting’. The great majority of students appreciated the chance to work
142 F. Rubinacci et al.

with robots and breed them (82 %). Students do confirm tat the observing and ruling
the evolutionary process increased their interest in science, technology, designing
and in biology, especially related to evolution, development and adaptation. In fact
on a Likert scale from 0 ‘I do not agree at all’ to 5 ‘I completely agree’, students
rated, on average, their increase of interest in cited discipline as follows: science
4.2; technology 4.5; designing 4; biology 4.8. They also confirmed that an appealing
aspect was related to the unforeseeable behaviours displayed by robots in interaction
with the environment 75 %. They had to carefully observe single robots behaviours
to choose the best exemplars and this stimulated their ability to address solutions
without designing them. Teachers answered questions about soft skills, and they con-
firmed that communication between students was stimulated (90 %), problem solv-
ing and group working were favoured (83 %), during the lab experience, in respect
to routine activities.
The BrainFarm lab allowed students to increase the ability to communicate well,
to develop good relations with others and to find shared solutions. Moreover students
learned that it was possible to make mistakes and that learning from mistakes is
good for everyone. The user acts a breeder, thus learning in a safe context to build an
expertise in selecting solution and using this knowledge to rule a complex system.

References

1. Ziemke, T.: The construction of reality in the robot: constructivist perspectives on situated
artificial intelligence and adaptive robotics. Found. Sci. 6(1–3), 163–233 (2001)
2. Webb, B., Consilvio, T.: Biorobotics. Mit Press (2001)
3. Nolfi, S., Floreano, D.: Evolutionary Robotics: The Biology, Intelligence, and Technology of
Self-organizing Machines. MIT Press (2000)
4. Ponticorvo, M., Walker, R., Miglino, O.: Evolutionary robotics as a tool to investigate spatial
cognition in artificial and natural systems. Artif. Cogn. Syst. 210–237 (2007)
5. Eguchi, A.: RoboCupJunior for promoting STEM education, 21st century skills, and techno-
logical advancement through robotics competition. Robot. Auton. Syst. 75, 692–699 (2016)
6. Gigliotta, O., Miglino, O., Schembri, M., Di Ferdinando, A.: Building Up Serious Games with
an Artificial Life Approach: Two Case Studies. In Evolution, Complexity and Artificial Life,
pp. 149–158. Springer, Berlin (2014)
7. Miglino, O., Gigliotta, O., Ponticorvo, M., Nolfi, S.: Breedbot: an evolutionary robotics appli-
cation in digital content. Electron. Libr. 26(3), 363–373 (2008)
8. Oladiti, B.T.: User-Guided Evolutionary Algorithms (1997)
9. French, M.: The Interplay of Evolution and Insight in Design. Evolutionary Design by Com-
puters, 77 (1999)
10. Kaelbling, L.P., Littman, M.L., Moore, A.W.: Reinforcement learning: a survey. J. Artif. Intell.
Res. 237–285 (1996)
11. Patterson, D.W.: Artificial Neural Networks: Theory and Applications. Prentice Hall PTR
(1998)
12. Goldberg, D.E., Holland, J.H.: Genetic algorithms and machine learning. Mach. Learn. 3(2),
95–99 (1988)
13. Miglino, O., Rubinacci, F., Pagliarini, L., Lund, H.H.: Using artificial life to teach evolutionary
biology. Cogn. Process. 5, 123–129 (2004)
A Thousand Robots for Each Student: Using
Cloud Robot Simulations to Teach Robotics

Ricardo Tellez

Abstract One of the main problems when teaching robotics is the lack of robots for
all the students in the class due to its costs and difficulty to maintain. In this paper,
we analyze the advantages and drawbacks of using simulations for teaching robot-
ics instead of real robots, as a solution to that problem. We describe a simulation
tool based in the cloud that allows the simulation of complex robots off the shelf by
using only a web browser as the base system for teaching robotics. Finally we pro-
vide the description of a teaching protocol for minimizing costs and troubles while
maximizing students experience, which makes a combined use of simulations and
real robots.

Keywords Teaching robotics ⋅ Robot simulations ⋅ Web application

1 Introduction

Robotics is a fashion subject, that is in an exponential growth all over the world. New
companies about robotics are created every week and the forecast indicates that it is
going to increase even more [1]. Hence, there is an increasing need of more prepared
engineers in this subject.
Teaching robotics requires to teach in the most optimal way in order to transfer a
knowledge that advances quickly, and to engage students into it. Fortunately, at this
moment in time, we have access to a large base of commercial robots, schematics of
open source robots [2], and free software [3]. All of that makes learning robotics eas-
ier than ever. However, the own existence of such large number of robotics options,
makes very difficult to teach it, because making the learning experience work for all
the students of the class can be, first, very expensive, and second, very complex.

R. Tellez (✉)
The Construct Sim, San Francisco, USA
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 143

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_14
144 R. Tellez

We identify the main problems of teaching robotics as the following:

1. Robots and components for robots are expensive. Hence it can be very expensive
for the school to provide those robots or parts to each of the students.
2. Even if the school has access to the robots or parts, it takes a long time to correctly
make work a given robotic setup. Robotics is a difficult subject, and requires the
proper functioning of a large amount of different parts. If the student has to build
all those parts and make them work it is going to take time. Furthermore, the
number of problems exponentially increase with the number of students in the
class.
3. Finally, setting up a robotics development environment requires a huge effort
for the students. Teachers can design the best experiments with robotics ever,
but making those run in the students own environments (or even in the environ-
ments the school provides) is very tricky. Students have access to different types
of computers and operating systems. In the same line, schools have different com-
puter configurations configurations, old computers, etc. Hence, even if the teacher
achieves to make the exercises work in her system, making the exercises work for
all the students is a different and more complex task.

In order to solve the first and second problems we propose the use of robot sim-
ulators for teaching robotics. In order to solve the third problem we propose the use
of web based simulations in the cloud.

2 Using Simulations for Teaching Robotics

The use of robot simulations for teaching means that, instead of using a real robot to
make students learn about robotics, the teacher uses a computer program (the simu-
lator) that shows the robots on the computer screen as if it were the real robot. The
simulated robot is called a model of the robot. There are many different simulators
available, each one with its own characteristics and working flow. Depending on the
simulator used and of the model quality, the model can look and behave more or
less close to the behavior of the real robot (see Fig. 1 with real Nao robot and Nao
simulation model for Webots simulator).
In order to start simulating a robot, the simulator program must be installed in the
computer of the user previously to start working with it. It usually requires specific
setups or configurations (for example, a given operating system in the computer,
some specific libraries installed, etc.) prior to the installation. Those requirements
make the installation process a little complex in most of the cases.
Once a simulator has been installed in a computer, the teacher must create the
model of the robot that she wants to use in her classes. Another option is to get the
model from an already created source. There are different repositories on internet
providing models already made for very famous robots, even if in most of the cases,
make those models work on the students computers represents a challenge in itself.
A Thousand Robots for Each Student . . . 145

Fig. 1 The real Nao robot and its simulation model of Webots simulator

Users always have the option of building their own models for existing robots, or for
other robots that they invent.
Once the robot model is available, it is required to create a simulated environment
for it. For example, the Nao robot of Fig. 1, is in a main room environment. The
whole thing including the robot model and the environment is called a simulation.
The simulation, is the program is executed inside the simulator and describes what
must be shown on the screen by the simulator (that is, the robot in its environment).
The robot model must have a programming interface that allows users to com-
mand the simulated robot, reading its sensors and sending commands to its actua-
tors. For example, in the case of Nao robot, the programming interface allows users
to capture data from the (simulated) camera that the robot has in its head, or send
commands to the motors of the joints and make the robot walk.
Once a model and its environment are created inside the simulator, teacher and
students can program it as if it were the real robot using the programming interface.
When the simulator executes the created program, the resulting behavior of the robot
is shown on the screen.
Latest simulators tend to present to the user the same kind of programming inter-
face that the real robot has. This fact represents a huge improvement from old robotics
systems, since a program created for the simulated robot can be used later in the real
robot without modification, and make the real robot behave (more or less) the same
as the simulated robot.
Finally, to say that there exist a lot of different simulators available, each one with
its own specifics. Some simulators are mainly focused on industrial robots, others
on service robots. Some are based on standards and some have their own protocol.
Deciding which one to use may be key for the teaching at hands. Table 1 contains a
table of the most used simulators for robotics.
146 R. Tellez

Table 1 List of most popular simulators, their intended scope and their availability for different
operating systems
Simulators Operating system Scope
Gazebo Linux, MacOS Service and industrial robots
Webots Windows, Linux, MacOS Service and industrial robots
V-rep Windows, Linux, MacOS Service and industrial robots
UsarSim Windows Service and industrial robots
Morse Linux Service robots
RobotStudio Windows Industrial robots
Virtual robotics toolkit Windows, MacOS LEGO and VEX robots
Visual components Windows Industrial robots
Robot virtual worlds Windows LEGO, VEX and TETRIX
robots

2.1 Advantages of Using Simulators

Using a simulation in front of a real robot has several advantages:

1. High reduction of the cost. Real robots tend to costs between hundreds and thou-
sands of euros. Simulations instead can start from zero to a hundred, depending
on the requirements and quality of the simulator.
2. Each student has access to her own (simulated) robot. Having a real robot for
each student is impossible even in the best schools, because of the prices. The
result is that the students must group in order to access the real robot and test
their algorithms. This results in a very poor learning experience for the student.
Instead, when using a simulator, each student can have to the simulated robot by
using only a computer.
3. Students cannot break the robot by doing errors. If a real robot is used, a simple
mistake in the code can make the robot crash or fall, breaking its parts. With
simulations, student’s mistakes cause no harm.
4. Access to the best robots of the world (see Fig. 2). The best robots of the world,
the ones being used now in the best research labs of the world, have a simulation
model available. By using a simulation, students can access those robots and learn
from the results of the best labs of the world.
5. Access to many different environments for the robot. Even if the school man-
ages to have a real robot for each student, it is very unlikely that will be able to
provide interesting environments for those robots (basically because you have to
build those environments). If you use a simulator, you can place the robot in any
environment you can think of. For instance it is possible to have a robot on space,
deep water or nuclear reactor, and test the students programs for the robots in
those environments.
A Thousand Robots for Each Student . . . 147

Fig. 2 A simulation of Baxter (from Rethink robotics), Reem-C (from Pal robotics), and BB-8
(from Disney) robots, some of the most advanced robots in the world

2.2 Drawbacks of Using Simulators

Using a simulator for teaching, has a series of drawbacks in front of real robots:
1. No learning about real hardware. By concentrating the student on the simulation,
she learns all the inner workings about simulations, a thing that is very different
from the inner works of real robots. Students don’t get this special learning that
is obtained by working with real stuff, and all the problems that are associated to
real robots.
2. Learning with simulators is mostly about software, with almost no learning about
mechanics nor electronics. 90 % of simulators are concentrated on learning how
to program a robot to do something. There is no learning about how to build the
mechanical parts of the robots or the electronic systems inside the robot that run
the control software.
3. Difference between the behavior of real and simulated robot. Even for the best
simulators, the models of the robots are not perfect. Hence, the behavior of the
simulated robot varies from the behavior of the real robot. How this variation is
important or not depends on the level of programming one wants the students to
concentrate at. If the level of programming is at that of the controller, then there
is usually a large difference between real and simulation. If instead the focus at
the level of functionality, the behaviors are very similar to real robots.
4. Simulation are less engaging for students. Making a simulated robot grasp an
orange from a table is not as exciting as making the real robot do it. Hence, simu-
lations do not have the same appealing as real robots. However, this engagement
highly depends on the implication of the teacher and her ability to create really
engaging simulation examples or exercises.
5. Many of current robot simulators are tricky to install, set up and have them work-
ing in different equipments. When using a simulator, there is a previous work
that needs to be done by the teacher, which requires installing the simulator in
the computers that are going to be used, if they belong to the school. If students’
148 R. Tellez

computers are going to be used, all kinds of errors and problems appear just trying
to install, for which the teacher must provide some kind of support.
6. Creating the robot model at present is not a simple task and requires a lot of
previous work to create it. This situation is at present quickly improving, because
there is an emerging market for web pages that provide already made complex
robot simulations to be used off the shelf [4, 5].
7. If several simulators are needed (because of different levels of proficiency of the
students, or because teaching different functionalities are required) then the set-
ting up work multiplies and becomes more complex. Of course, teachers must
provide the simulation for each of the setups which is not straight.

3 Using Web Based Simulations for Teaching

Given some of the drawbacks presented in the previous section, we suggest the use
of web based simulators for teaching instead of desktop based simulators. The differ-
ence between desktop and web simulators is that the former requires to be installed
in a specific computer (with all the drawbacks indicated above), and the later requires
no installation at all. While desktop simulators require the user to install the simu-
lator in the user desktop or laptop computer, the web simulator is just a web page
that the user can access with her account using only a web browser. Users can use
any browser, which means that they can simulate from any computer at any location,
including from home, school of internet cafe.

3.1 Advantages of Web Based Simulators

These are the advantages of using a web based simulator:

1. Do not need installation nor maintenance. Installation and maintenance of the
simulator is performed by the web portal that provides the simulator. Hence, nei-
ther the school nor the students have to deal with this unrelated task. This is spe-
cially convenient for some schools where the policy of installation of software is
very strict and requires a long bureaucracy process.
2. Teachers can be sure that their exercises are going to work as they have designed
them. Since all students and teachers are using the exact same simulation system,
the results that are going to be obtained are exactly the same. Teacher can prepare
exercises and show steps to follow that, if correctly performed, will produce the
expected results in the simulated robot.
3. Students cannot break the simulation program by making errors or miss use of
the simulation. If students make mistakes and something goes wrong crashing
the simulation, they just can relaunch the web simulation again and start again
from the initial conditions. Performing that error in desktop computers is one of
the errors that consumes most of the time when teaching robotics, since students
A Thousand Robots for Each Student . . . 149

perform errors that destroy the simulation system in their computers. This may
require even to reinstall the whole operating system. With web simulators stu-
dents cannot break any system, what makes students more prone to experiment
and try new programming ideas.
4. Different simulators available which allows entrance at different level of com-
plexity. A web simulation system can provide different simulators at the same
portal, each one concentrated in teaching a specific subject (for instance, simula-
tors concentrated in robotics arms, or simulators for humanoids or simulators for
evolutionary robotics), or different levels of proficiency (for high school students,
for university students, etc.).
5. Students can use any type of computer. In a class, there are all kinds of students,
each one using their preferred operating system. As can be seen in Table 1, not
all the simulators can work in all the operating systems. With a web simulator,
this problem fades away since the only required thing is a web browser, which is
available for all types of operating systems.
6. Students and teachers can work from anywhere. Since the only requirement is
to have access to internet, teachers and students can actually be anywhere while
doing their simulations. This opens a huge door for remote teachings.
7. Students can cooperate with their mates while working on the simulation. Web
simulation is by nature a collaborative thing, at the contrary of desktop simula-
tions. Hence, differently from desktop simulators, web simulators allow the work
of different people at the same time on the same simulation. This is very useful
when students need to cooperate to program the different parts of a robot, or when
the teacher needs to see where the student is stuck and help her get the solution.

3.2 Drawbacks

As any system, web simulations have also drawbacks:

1. Users need a fast enough internet connection. Web simulations are run through an
internet connection, and since the simulations tend to be complex, it is required
to have an internet connection with fast speed.
2. By being on the cloud, web simulation inherits all the security problems that any
cloud system has. It is very difficult to ensure that the data stored in the cloud will
not be accessed by unauthorized people.

4 A Case Study: The Construct Web Simulation System

As a solution for web simulation we propose The Construct. The Construct is a web
platform in the cloud that provides a large list of simulators ready to be used by
means of a web browser. It is a web based simulator with all the advantages (and
drawbacks) explained in the previous section.
150 R. Tellez

Fig. 3 A view of the simulators available at The Construct

The Construct allows to execute simulations off the shelf by using WebGL visu-
alization. Users do not have to install anything, not even in their browsers, since
all modern browsers already support WebGL based code. When the user access the
platform using her preferred web browser, she has access to a list of different pre-
configured robot simulators (Fig. 3).
Once the user access the portal, she can select which simulator to start simulating
with. Selection of the simulator depends on the type of work she is going to do, and
the level of complexity of the simulations. For instance, for beginners, it is easier to
start using Webots simulator [6] because it is more simple in terms of interface and
tools. However, if you want to learn about the latest robotics technologies like ROS
[3] you may need to use the Gazebo simulator [7]. Finally, if you want to develop
robotics with deep learning you should use the DRC+CUDA simulator [8].
The simulators available in The Construct are in fact existing desktop simulators.
This implies that simulations created with the desktop versions of the simulator are
compatible with the web version, and vice versa. Hence, teachers and students can
upload to The Construct their already existing simulations and run them in the web.
The opposite is also possible.
The Construct only needs a WebGL enabled browser. Most modern browsers are.
Officially supported are Safari, Chrome, Firefox. This means that it can be used with
any device that has one of those browsers, including Linux, Windows, MacOS, or
even tablets and smartphones.
Apart from showing the simulator on the web browser, The Construct incorpo-
rates a series of very useful features for teaching robotics, that make the system go
far beyond its desktop versions.
A Thousand Robots for Each Student . . . 151

4.1 Sending Simulation Files to Students

When a teacher has created a simulation for their students, she can directly share the
files that compose the simulation with all her students with a single click, without
having to send files through email, or storing the files in some place of the internet.
With a single click, the file selected is sent to a list of students provided by the
teacher. From that moment, the students of the list can run the exact simulation that
the teacher created, each student in her own environment free to experiment without
interfering with the other students simulations. The opposite is also true: the students
can send their simulations to the teacher from within The Construct and be sure that
their simulations will work for the teacher as they expected.

4.2 Sharing Simulation with Teacher

If a student has a problem and does not know how to continue, she can request help
from the teacher, by sharing her running simulation with the teacher. When a running
simulation is shared, both the student and the teacher are seeing the same running
simulation. This implies that the teacher can see what is failing in the simulation and
programs of the student, and help her understand the error. The teacher has complete
access to the simulation of the student and can modify it or suggest new ways of
continuing.
This same mechanism of sharing can be used between students to collaborate
between them in the programming of a simulated robot. In the same sense a Google
Docs allows different people to write over the same document, sharing a simula-
tion allows different people to work over the same simulation. It is just a matter of
selecting the option of sharing and specifying the list of users that the owner of the
simulation wants to share with.
After the collaboration is finished, the owner of the simulation has the right at
any time to close the sharing session and forbid the access to her simulation.

4.3 Integrated Development Environment

Neither students nor teachers need extra programs than the web browser to simulate
the robots and create the control programs for the simulated robot. A full environment
is provided including IDE, web console and also a Python web environment. Hence,
the full learning can be done using only a web browser.
152 R. Tellez

Fig. 4 Simulation contest of Nao robot running the 10 m

4.4 Engage Students Through Simulation Contests

One way to encourage students and robotics research is by doing contests where
the participants have to compete against each other. For example one of the most
important robotics simulation competition was the Virtual Robotics Challenge [9]
where the simulation was web based and involved the best robotics teams of the
world to control a human size humanoid robot on a nuclear disaster area.
This kind of competition can be easily done using web simulations at The Con-
struct, like for example the Robto Race to Hawaii contest based on Nao robot race,
where students around the world should make a Nao robot walk for 10 meters as fast
as possible (see Fig. 4).

5 Proposing an Optimized Robotics Teaching Environment

One of the main criticisms to using simulations to teach robotics instead of real robots
is that students get a distorted vision of the field and separate themselves from the
real robots, by concentrating too much on the simulations instead of real robots.
One way to solve this problem without incurring into too much costs is to combine
simulations with real robots.
Modern simulators work based on a hardware abstraction layer (HAL) [10]. A
HAL is a framework for a robot in which the commands sent to the motors or sen-
sors of the robot do not differentiate between the real robot or the simulated robot.
Usually, the HAL is provided by the company that builds the robot. Sometimes, the
robotics community creates this HAL for interesting robots that do not come with it.
Other times, when a robot does not provide such HAL, the simulator may provide
instead a cross compilation facility [11]. This is the case for example of some of the
robots provided by Webots simulator. In the case, for instance of Aibo robot, it does
not have such abstraction layer due to its low CPU computer and being an old robot.
A Thousand Robots for Each Student . . . 153

Webots provides a mechanism that allows the translation of the code you use in the
simulation of Aibo robot to code to be executed in the real robot. The result is that,
the same things that the robot does in the simulation with the user program will be
done by the real robot when using the cross-compilated code [12].
Having this framework available, either through HAL or cross-compilation, a new
teaching environment that optimises costs and maximizes students experience can
be devised. We call this environment the Optimized Robotics Teaching Environment
(ORTE). It works as follows:

1. The teacher selects the real robot to use and buys it. Only one or two units of the
real robot may be necessary (depending on the amount of students per class and
the resources available).
2. The teacher creates a simulation of the robot which mimics the way the real robot
will be used for the class. It is very likely that a simulation for that robot and
environment already exists in the net, hence it is suggested to perform a search
for it. Getting an already made model of the robot can save many days of work.
3. The teacher distributes the simulation among students and specifies what it is
required to do (for instance, make the robot recognize an orange, make it walk
up to the table, etc.).
4. Students create their control programs for the simulated robot in order to accom-
plish the assigned task.
5. Once a student has achieved a program in the simulation that performs the task,
she can transfer her program to the real robot making use of the HAL or the cross-
compilation, and check if it works the same way in the real robot. Chances are
not. Hence the student returns back to simulation to analyze why not and devise
a new strategy.
6. Keep repeating fourth and fifth points until fifth is accomplished.
One example of successful use of the ORTE is the MIT course MIT 2.166 of
2016 on autonomous vehicles (class named Duckietown [13]), where students have
to build their real robots using arduino boards and use the simulation of those robots
in The Construct to build and test control programs (Figs. 5 and 6 show the real and
simulated Duckiebot model and real and simulated environment).

6 Conclusion

Simulations are a powerful tool for teaching robotics, since allow the creation of
cheap robotics environments, and access to the latest (simulated) robotics technolo-
gies at a very low cost.
Simulations can be simplified for teachers and students by making use of web
based simulators. Web based simulators universalize the access to such software
because allow anybody use them with any device with a fraction of the cost of the
desktop simulator.
154 R. Tellez

Fig. 5 Real Duckiebot and its simulation model

Fig. 6 Real Duckietown environment and simulated environment using The Construct

Finally, simulations can be integrated in a complete learning system that includes

real robots in the path, minimizing costs and maximizing usage and students expe-
rience (ORTE method).
A Thousand Robots for Each Student . . . 155

Acknowledgments We would like to thank all the Duckietown team for making it possible and
specially Andrea Censi for making us part of the project.

References

1. Silicon Valley Robotics: Service robotics case studies in silicon valley. Tech Report (2015)
2. Poppy humanoid robot. http://www.poppy-project.org
3. Robotics operating system. http://www.ros.org
4. Simulating Husky robot: the easy way. http://www.theconstructsim.com/simulating-husky-
robot-the-easy-way
5. Simulating Tiago robot: the easy way. http://www.theconstructsim.com/simulating-tiago-
robot-the-easy-way
6. Michel, O.: Webots: professional mobile robot simulation. J. Adv. Robot. Syst. (2004)
7. Koenig, N., Hsu, J.: The many faces of simulation: Use cases for a general purpose simulator.
IEEE International Conference on Robotics and Automation (2013)
8. Hsu, J.M., Peters, S.C.: Extending open dynamics engine for the DARPA virtual robotics chal-
lenge. In: 4th International Conference, SIMPAR (2014)
9. Agero, C.E., Koenig, N., Chen, I., Boyer, H., Peters, S., Hsu, J., Gerkey, B., Paepcke, S., Rivero,
J.L., Manzo, J., Krotkov, E., Pratt, G.: Inside the virtual robotics challenge: simulating real-time
robotic disaster response. IEEE Trans. Autom. Sci. Eng. 12(2) (2015)
10. Jorg, S., Tully, J., Albu-Scheffer, A.: The hardware abstraction layer—supporting control
design by tackling the complexity of humanoid robot hardware. IEEE International Confer-
ence on Robotics and Automation (2014)
11. Michel, O., Rohrer, F.: The Rats Life Benchmark: Competing Cognitive Robots RSS (2008)
12. Holh, L., Tllez, R., Michel, O., Ijspeert, A.: Aibo and Webots: simulation, wireless remote
control and controller transfer. Robot. Auton. Syst. 54(6), 472–485 (2006)
13. Duckietown MIT 2.16. http://duckietown.mit.edu/
 Part IV
Technologies for Educational Robotics
Networking Extension Module
for Yrobot—A Modular Educational
Robotic Platform

Michal Hodoň, Juraj Miček and Michal Kochláň

Abstract The contribution describes a modular educational robotic platform

Yrobot and its networking extension module. This module is designed for wireless
communication among multiple Yrobot platforms as well as for interfacing with
smart devices and computers. The networking module enables to verify various
wireless communication standards and technologies in ISM band. In the text, the
motivation for developing of such extension module is described as well as circuit
solution implementation and firmware attributes for the module.

Keywords Networking extensions ⋅ Yrobot ⋅ Educational robotic platform

1 Introduction

Mobile robotics systems for educational purposes come still more and more to the
foreground especially in terms of modern teaching. This progress goes
hand-in-hand with OpenHW systems advancements, which enable a wide com-
munity of people to enter the world of computer engineering. In [1], the authors
proposed lab work for learning fault detection and diagnosis, training the skills
important for engineering education in mechatronics. An approach in [2] is based
on the pushing of students to design and test their own original circuits and software
code by modifying, extending or expanding the sample circuits and example code
described in the lecture notes, in order to keep students highly curious, motivated
and engaged in self-regulated learning. The work in [3] presents the design of an
open educational low-cost modular and extendable mobile robot based on Android

M. Hodoň (✉) ⋅ J. Miček ⋅ M. Kochláň

University of Žilina, Univerzitná 8215/1, 010 26 Žilina, Slovakia
e-mail: [email protected]
J. Miček
e-mail: [email protected]
M. Kochláň
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 159

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_15
160 M. Hodoň et al.

and Arduino to be used as an educational tool in labs and classrooms. Other

low-cost but open-source robotic platform AERobot usable for introductory pro-
gramming and robotics teaching was described in [4]. Further interesting robotic
platforms as well as they use in education were described for example in [5–9].
Beside the tangible platforms, robotic simulation environments took place in an
educational process as well, e.g. [10, 11].
Yrobot high-school education kit was for the first time introduced at 2014
RAAD conference. From that time, already after 18 months, the robot was suc-
cessfully integrated into the educational process in more than 20 high schools in
Slovakia. During 2015, we decided to extend the kit with additional application
modules to target an actual problematic of wireless communication and derived
areas, such as Internet of Things; Collaborative Signal Processing; Distributed
Signal Processing; etc. For that reason, Yrobot modules “Y-WiFi”, “Y-BlueTooth”
and “Y-ZigBee” were introduced. These modules can be easily installed within the
platform, where provided functionalities, which can be easily translated into edu-
cational process. Especially in the popular fields such network administrator or
network specialist.
To understand the concept of Yrobot, we have to mention an effort of Volk-
swagen Foundation, which was supporting the project continuously from 2013. The
idea behind was to develop an educational kit that can be used for the purposes of
IT education among the Slovak high schools. Original intent of authors was, by
using of a simple technical device, to increase the motivation of
secondary-school-students in the study of technical fields. Among the others,
especially to arise their interests in information technology. The concepts of the
system, its features and functions, as well as initial results from deployment in
teaching were presented in [12, 13].
As supporting activities, different workshops, where particular teachers got in
first touch with the platform, were realized. Furthermore, two regional competitions
“YrobotCup - lets program the robot” were organized for the students of Žilina and
Bratislava region. In these contents, the students could compare their knowledge
with other competitors what could motivate them in other work with Yrobot. To
facilitate the work with Yrobot, the textbook where basic principles together with
programming examples were described by the usage of simple, friendly and
easy-to-understand way, was published [14].
As a reminder to be mentioned [12, 13]—the robotic platform has the nature of
open source hardware with presumes on the development of interesting applications
(hardware and software modules) directly by the secondary school students. In the
figure below (Fig. 1), the platform with textbook is shown.
To invent an interesting and non-traditional application with the design of
appropriate HW and SW means is a very complex task. The successful solution
implies a good knowledge of several disciplines (sensory, electronics, mechatron-
ics, communications technology, programming, etc.) as well as the ability to adopt
the knowledge from different disciplines to synthesize and thus achieve the desired
result.
Networking Extension Module for Yrobot—A Modular Educational … 161

Fig. 1 Yrobot with textbook (Slovak Release)

Another result of such complex challenges treatment is the fact that solving them
asks for the creation of interdisciplinary teams. This represents a natural way to
support and develop the ability of students to work in teams:
• to respect the opinion of colleagues;
• to find a way and courage how to defend the ideas;
• to develop interesting approaches of colleagues
• to find effective solutions; etc.
Relatively high expectations of the Yrobot-project-team were put in the ability of
students to process and develop inventing tasks by themselves. In this area, the
great inventiveness of students was expected. We assumed that we meet original
ideas and interesting solutions that would not be burdened with conventional
approaches. However, during the first year of Yrobot usage, we saw relatively small
activities of students to develop new extension modules and related new applica-
tions (e.g. [15]).
This can be on the one hand caused by generally poor knowledge of students of
technological processes necessary for the HW production. On the other hand,
financial demands play also an important role. For these reasons, we have decided
to support the development of new applications by using of special extension
modules that extend application possibilities of the basic kit.
When designing such modules, it is necessary to respect existing communication
capabilities of the Yrobot base (UART/SPI) as well as mechanical and space
restrictions. From the viewpoint of price minimizing, it is necessary to minimize
PCB dimensions and implement a kind of simple circuit solutions while keeping an
adequate reliability and flexibility.
162 M. Hodoň et al.

Table 1 Communication technologies

Technology (2.4 GHz) Standard Module Max bit rate
WiFi 802.11 b/g/n WizFi250 65 Mbps
Zigbee 802.15.4 JN5168M0 1 Mbps
MRF24J40 250 kbs
Bluetooth v.2.0 EDR 802.15.1 BTM182 3 Mbps

2 RF Communication Modules

According to the evaluation of study programs of the Slovak high schools, taking
into account current trends in IT development, the field of wireless communication
was chosen as one of the popular application areas for Yrobot expansion. It can be
assumed that additional wireless connectivity added to mobile robots extends sig-
nificantly the potential application cases and scenarios.
The Yrobot was originally developed as an autonomous Yrobot device capable
to solve simple tasks by reading the status info of installed sensors (e.g. moving
across the line, avoiding obstacles, discovering the space, etc.).
Implementation of wireless communication allows transformation of Yrobot
from single and autonomous functioning to robust multi robotic system able to
solve the robust challenges and to bring the complex solutions. For an effective
operation of the system it is possible to use various communication technologies,
protocols and different network topologies. In our approach, we decided to
implement three separate communication modules operating in the 2.4 GHz ISM
band. The chosen standards can be seen in the Table 1.
All developed modules, with dimensions of 42.5 × 60 mm, can be connected
via connectors on the motherboard to the Yrobot MCU. Through connector JP2 is
assured power feeding of the module and the serial asynchronous communication
with Yrobot MCU. Synchronous communication interface SPI employs JP9 con-
nector. The module contains circuitry to secure the power supply, to convert the
signal logic levels, and to select the communication line between UART and SPI.
Module states can be controlled through signalization LEDs. The buttons on the
module allows basic control or setting up of the operation mode.

3 Communication Module Y-WiFi

As a first module from the mentioned RF Communication Modules was realized

communication module Y-WiFi allowing to manage and monitor the performance
of Yrobot systems by using of WiFi network (IEEE 802.11x). The module should
primarily secure the wireless communication between mobile Yrobots.
As the basic component of this module was chosen communication module
WizFi250. WizFi250 comprises an integrated LNA as well as integrated antenna
Networking Extension Module for Yrobot—A Modular Educational … 163

3V3 5V
Yrobot
JP9
WizFi250 TXS
3V3 SCK
LED1
SPI
WiFi_LED MOSI
4xCL
MISO
LED2 JMP3
MODE
1
VIN
SS/ 2
Yrobot
TL1 3 JP2
FUNC/

UART
JMP2 1xCL
RXD TX
TL2 1 2
RESET/ TXD RX
BOOT 3 4

VA 3V3
JMP1 LED5
VIN
1 2 3
1 2
JMP4
VUSB

JMP5 5V
VIN
FT232RL 1
USB RX
2 STAB2
VUSB TX
DM
DP
3V3
LED3 3V3
VA
CBUS0
STAB1
LED4
CBUS1

Fig. 2 Block schematics of Y-WiFi module

with connector for an external antenna connection. It can serve the secure com-
munication protocols WEP, WPA, WPA2-PSK. Its control can be realized through
a suitably chosen set of AT commands.
Module Y-WiFi integrates all other necessary components making it easy to
interconnect the module with Yrobot or with personal computer via USB port. The
conversion between UART and USB was assured through FT323RL circuit. Block
diagram of the Y-WiFi module is shown in the Fig. 2. Yrobot with attached
Y-WiFi module is shown in the Fig. 3.

3.1 Operation Modes of Y-WiFi Module

Communication module Y-WiFi can work in the two basic configurations/modes

(mode Yrobot and mode PC) and in one auxiliary configuration in which it is not
used WizFi250 module but Yrobot can communicate through USB port with master
system.
1. Yrobot Mode—module Y-WiFi is connected via JP9 and JP2 to the Yrobot
mother board. In this mode, the Y-WiFi module enables Yrobot to communicate
164 M. Hodoň et al.

through wireless WiFi network. In the basic mode it is possible for the com-
munication between the Yrobot MCU and Y-WiFi module to use SPI or UART
interface.
(a) In case that SPI communication has to be used, it is possible to apply TX
signal as the device slave select signal (SS). Then it is necessary to inter-
connect pins 2–3 on the jumper JMP3, while on the jumper JMP2 it is
necessary to disconnect the pins 1−2.
(b) In case an asynchronous serial communication UART has to be used, it is
necessary to interconnect pins 1–2 and 3−4 on the jumper JMP2 and on the
jumper JMP3 disconnect the pins 1–2.
2. PC Mode—module Y-WiFi is through a USB cable connected to a host system
(PC, tablet, etc.). Personal computer identifies the connection of a new device
and automatically creates a virtual COM port. Then it is possible to commu-
nicate with the module via an application that enables serial communication
(e.g. Hyperterminal, …). The module is powered from the USB port. In this
case, it is necessary:
(a) to disconnect the pins 2−3 on the jumper JMP3;
(b) to interconnect the pins 1/JMP5 with 1/JMP2;
(c) to interconnect the pins 2/JMP5 with 2/JMP2;
(d) to interconnect the pins 1−2 on the jumper JMP4.
After the setting up of all module jumpers (JMPx), it is possible—via the master
system—to communicate with the module and to control its activity by using of AT
commands. Eventually it is also possible to create and verify additional WiFi-based
communication applications. The additional modules (SW or HW), which can be
connected through PC virtual COM port, allow the development of other network
applications—from simple peer-to-peer connections, through private WiFi net-
works up to the complex client/server applications.
In this mode, it is also able to update the firmware of WizFi250 module. When
updating, it is necessary to interconnect pins 1−2 on the jumper JMP1. When these
pins are connected, signal BOOT gets to the LOW state. This causes WizFi250
system transition into the BOOTING mode after the system restart (Tl2 button).
3. Auxiliary Mode—Y-WiFi module configured in this mode utilizes the circuitry
that secures conversion UART/USB (FT232). According to this, it is able for the
Yrobot MCU to communicate with the superior master system (PC, tablet, etc.).
This makes it possible to control Yrobot platform through the commands
entered from the connected computer. In this mode it is necessary to intercon-
nect pin 1 on the jumper JMP5 together with the pin 2 on the jumper JMP2as
well as the pin 2 on the jumper JMP5 with the pin 4 on the jumper JMP2. It is
Networking Extension Module for Yrobot—A Modular Educational … 165

also recommended, in auxiliary mode, to power the Y-WiFi module from the
Yrobot motherboard. Therefore, it is necessary to interconnect the pins 2−3 on
the jumper JMP4.

4 RF Communication Modules Y-Bluetooth and Y-ZigBee

Communication modules Y-Bluetooth (Y-BT) andY-ZigBee have the similar

characteristics as the previous module Y-WiFi. When Bluetooth connectivity has to
be implemented, BT modules BTM11x or BTM18x could be used instead of
WizFi250. When the ZigBee communication standard is selected, modules
MRF24J40x or JN5168x could be used.
The selected Bluetooth module has to be assembled on the Y-WiFi PCB through
the component side. Communication with Yrobot MCU can be realized through
UART or SPI interface. The module can be controlled by a set of AT commands as
well. Y-Bluetooth supports PC and Yrobot Modes.
The same assembly procedure as for the Y-BT is valid for the Y-ZigBee too.If
the user chooses the Y-ZigBee solution, the ZigBee module should be placed from
the component side of PCB. From the supported Zigbee modules,
(MRF24J40MA/MB, or JN5168), only the JN5168 supports UART/SPI commu-
nication, while MRF24J40 can communicate with Yrobot MCU only via SPI.

Fig. 3 Yrobot with Y-WiFi module

166 M. Hodoň et al.

5 Conclusion

An extension of the Yrobot kit by the set of network modules significantly expands
the variety of applications that can be implemented under it. The kit is since its
inception conceived and designed for the needs of teaching of subjects in IT. In
addition to this primary function, the kit serves also the popularization function.
The kit should be used for an encouragement of the students for the study of the
technical subjects/fields.
In the near future, the focus will be put on the development of interesting and
original applications designed according to experiences with the communication
modules usage. The delivery of supplementary textbook is in this case more than
necessary. In the textbook, the basic capabilities of individual RF network tech-
nologies supplied with the simple examples, that will illustrate the benefits and
limitations of wireless communication, will be described. Hopefully, other inter-
esting applications, which could motivate the students to the own further devel-
opment, will be part of the textbook too.
Further steps are, beside the textbook development, oriented in the development
of additional modules in the field of RF communication. At the present time, RFID,
NFC, Z-Wave modules, together with the chosen proprietary communication sys-
tems in the free ISM bands (e.g. RFM70), are being developed. It is expected that
these extensions will expand the current status of the kit with other interesting ICT
applications.

References

1. Gomez-de-gabriel, J.M, Mandow, A.., Fernandez-lozano, J., Garcia-cerezo, A.: Mobile robot
lab project to introduce engineering students to fault diagnosis in mechatronic systems. IEEE
Trans. Educ. 58(3), 187–193 (2015). doi:10.1109/TE.2014.2358551
2. Cubero, S.N.: A fun and effective self-learning approach to teaching microcontrollers and
mobile robotics. Int. J. Electr. Eng. Educ. 52(4), 298–319 (2015). doi:10.1177/
0020720915585798
3. Lopez-rodrigues, F.M., Cuesta, F.: Andruino-A1: low-cost educational mobile robot based on
Android and Arduino. J. Intell. Robot. Syst. 81(1), 63–76 (2016). Special Issue: SI, doi:10.
1007/s10846-015-0227-x
4. Rubenstein, M., Cimino, B., Nagpal, R., Werfel, J.: AERobot: an affordable
one-robot-per-student system for early robotics education. In: IEEE International
Conference on Robotics and Automation (ICRA 2015), 26–30 May 2015, pp. 6107−6113,
Seattle, WA, USA. ISSN:1050-4729, doi:10.1109/ICRA.2015.7140056
5. Suzuki, K., Kanoh, M.: Effectiveness of a robot for supporting expression education. In:
Conference on Technologies and Applications of Artificial Intelligence (TAAI 2015), 20–22
Nov 2015, pp. 498−501. doi:10.1109/TAAI.2015.7407119
6. Zhi, D., Wu, Z. Li, W., Zhao, J., Mao, X., Li, M., Ma, M.: Education-oriented portable
brain-controlled robot system. In: IEEE International Conference on Robotics and
Biomimetics (2015 ROBIO), 6–9 Dec 2015, pp. 1018−1023, Zhuhai. doi:10.1109/ROBIO.
2015.7418905
Networking Extension Module for Yrobot—A Modular Educational … 167

7. Fabregas, E., Farias, G., Dormido-canto, S, Guinaldo, M, Sanchez, J., Bencomo, S.D.:
Platform for teaching mobile robotics. J. Intell. Robot. Syst. 81(1), 131–143 (2016). doi:10.
1007/s10846-015-0229-8
8. Binugroho, E.H., Pratama, D., Rizqy Syahputra, A.Z., Pramadihanto, D.: Control for
balancing line follower robot using discrete cascaded PID algorithm on ADROIT V1
education robot. In: International Electronics Symposium IES 2015, 29–30 Sept 2015, pp. 245
−250, ISBN: 978-1-4673-9344-7, Surabaya, Indonesia. doi:10.1109/ELECSYM.2015.
7380849
9. Kettler, A., Szymanski, M., Liedke, J., Wörn, H.: Introducing Wanda—A new robot for
research, education, and arts. In: International Conference on Intelligent Robots and Systems
(IROS 2010), IEEE, 18–22 Oct 2010, pp. 4181−4186, ISSN:2153-0858, Print ISBN
978-1-4244-6674-0, Taipei. doi:10.1109/IROS.2010.5649564
10. Torres-torriti, M.: Arredondo, T, Castillo-pizarro, P: Survey and comparative study of free
simulation software for mobile robots. Robotica J. 34(4), 791–822 (2016)
11. Banduka, M.L.: Robotics first-a mobile environment for robotics education. Int. J. Eng. Educ.
32(2), 818–829. Part A
12. Kochláň, M., Hodoň, M.: Open hardware modular educational robotic platform—Yrobot. In:
RAAD 2014, 23rd International Conference on Robotics in Alpe-Adria-Danube Region,
Slovakia (2014). ISBN 978–80-227-4219-1
13. Miček J., Karpiš O.: Audio communication subsystem of multi-robotic system YROBOT. In:
RAAD 2014, 23rd International Conference on Robotics in Alpe-Adria-Danube Region,
Slovakia (2014). ISBN 978-80-227-4219-1
14. Miček, J. et al.: Sprievodca po svete Yrobota, Žilinskáuniverzita (2015). ISBN
978-80-554-1120-0
15. Miček J., Karpiš, O., Kochláň M.: Audio-communication subsystem module for Yrobot—a
modular educational robotic platform. In: EDERC 2014, 6th European Embedded Design in
Education and Research, Italy (2014). ISBN 978-1-4799-6842-8
Aeris—Robots Laboratory with Dynamic
Environment

Michal Chovanec, Lukáš Čechovič and Lukáš Mandák

Abstract Is it possible to create an ant robot which can leave “scents” behind?
Is it possible to create robotic football on the same platform? What about mouse
maze, line follower or sumo robot? And what about the possibility to have dynam-
ically scalable environment which can interact with robots? Everything mentioned
and more can be done on the same platform called Aeris. In this paper, robot plat-
form called Aeris is presented. Department of Technical Cybernetics has long years
experience with implementing different robots into education. The logic step was
trying to integrate existing robots experiment scenarios into one platform but the
result went over borders of common “robot” discipline. Universal and interactive
robot “playground” concept is presented, with potential to be easily usable on teach-
ing purposes from simplest robotics to technical cybernetics, embedded systems or
artificial intelligence. This platform has a potential to be also powerful equipment
for researchers, for example in dynamic learning systems, swarm systems or other
learning algorithms. The actual state of Aeris is presented with an overview of future
work.

Keywords Robots in education ⋅ Robot laboratory ⋅ Dynamic map on LCD ⋅ Robot

simulations

1 Introduction

It is said that if we want to see best of ourself we should let ourself play.

1.1 History of Robots on DTC

Department of Technical Cybernetics implements robots into education since 2002

when AT90S1200 [1] microcontroller was released. Our first two wheel drive robot

M. Chovanec ⋅ L. Čechovič (✉) ⋅ L. Mandák

Faculty of Management Science and Informatics, Department of Technical
Cybernetics, University of Žilina, Univerzitná 8215/1, 010 26 Žilina, Slovakia
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 169

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_16
170 M. Chovanec et al.

with stepper motors and ultrasound sensors was built on this device. As this was
introduced into education, the interest from the students side was enormous. Next
step was to bring some challenges into the process. ISTROBOT [2] competition on
Slovak Technical University became an arena for the students to measure technical
and software parts of their robots. Every year since then, students were guided to
build their own or use existing wheel drive robots with various microcontrollers and
sensors prepared specially for competition categories as line follower, mouse maze
or sumo [2].
These activities were focused on University students and their study process.
Figure 1 shows presentation of robot called George.
Then there come interest from high schools for education platform and as there
was importance for better preparation for potential students, platform YROBOT was
created in 2012 [5, 6]. Parallel with developing YROBOT, faculty became the host of
First Lego League [4] regional competition from 2012. This competition is primarily
focused on primary and secondary school students.
Since then DTC helps to bring the whole scale of students at different school
levels to robotic education.
In late 2014 research need brings idea of an ant robot which can leave mark—scent
behind, as a part of an ant colony optimization problem. Standard robot approach
with the static track was not sufficient so it become clear that there is a need to bring
a kind of interaction upgrade to the existing approach. This effort resulted into the
system called Aeris.

Fig. 1 Presentation of robot called George on European Researchers’ Night in Bratislava, Slovakia,
2011. It was the predecessor of YROBOT
Aeris—Robots Laboratory with Dynamic Environment 171

1.2 Standard Robot Playground

In standard robot scenario there is defined static playground (track). Usually, this
track is for one robot goal, painted on the floor, created with tape, build from the
wooden board or Lego parts. Doing changes into track needs to manually reassemble
track parts.

1.3 Dynamic Robot Playground

In dynamic robot playground, the map is changeable at any moment. This fact brings
many new possibilities and applicability of such a system. This system can handle
all standard static robot scenario maps as line follower, sumo or mouse maze. It
is possible to extend these standard scenarios for example in a way such as a line
follower track could dynamically change while the robot is following, or maze could
have dynamic parts. It is easy to imagine many new parameters in standard static
robot scenarios.

1.4 Related Work

As idea of using robots for teaching is not new so let’s highlight some interesting
systems incorporating robots into education and research.
SyRoTek project that aims to create an e-learning platform for supporting teach-
ing mobile robotics and artificial intelligence [7]. In principle final version of Aeris
should be able to work also as remote e-learning platform thanks to its client-server
based programming model with high modularity.
Robot platform Colias. “Colias is a low-cost, open-platform, autonomous micro
robot which has been developed for swarm robotic applications. Colias employs a
circular platform with a diameter of 4 cm. It has a maximum speed of 35 cm/s that
gives the ability to be used in swarm scenarios very quickly in large arenas. Long-
range infrared modules with adjustable output power allow the robot to communicate
with its direct neighbors from a range of 0.5 cm to 3m′′ [8]. If Aeris system is a grey-
scale system Colias could be a possible choice of the robot to work with. Missing are
RGB sensors, WIFI module, and computation power, so instead of making changes
into Colias, own robot platform was developed.
The main advance in Aeris robot is more computing power with 72MHz ARM
CortexM4 with FPU on board and thanks to used API, robot programming stayed
user-friendly. This allows computing more complicated problems from artificial
intelligence. Such neural networks, reinforcement learning or genetics algorithms,
all of them require adequate memory and computing power to be processed in real
time.
172 M. Chovanec et al.

Cos𝜙. Cos𝜙 is artificial pheromone system that shows similarities with Aeris as it
uses also LCD screen and USB camera [9]. Cos𝜙 system is specialized on ant colony
optimization using Colias robot and is basically one possible grey-scale scenario of
Aeris system. Differences of Aeris system compare to Cos𝜙 are that Aeris is aiming
to be the universal system, with possible to create unlimited numbers of scenarios
and Aeris is recognizing RGB components in 10-bit resolution.

2 Aeris System

Aeris is aiming to be the universal tool for creating dynamic 2D environment simu-
lations able to interact with mechanical robots. Created environment itself does not
need mechanical robots as it is able to create virtual robots too because every part
of the system is the robot (as example wall is a non-movable robot with specific
diameters). Part of Aeris responsible for simulations and interaction with mechani-
cal robots is Supervising Control and Simulating center. Trough this part of system,
map and robots are able of interaction. System control center knows position of the
mechanical robot (thank’s to camera above playground or touchscreen) and is able
to communicate with the robot. Control center controls scenario which is displayed
on the map. The map is represented by horizontally placed LCD display. The robot
is reading a map and executing given tasks by given program.

2.1 Aeris Parts

Aeris, as is defined, needs playground/environment visualization device represented

by horizontally placed LCD display and control center represented by a computer
which runs program and control display. To connect Aeris with real mechanical
robots, control center needs to be equipped with camera or touchscreen. Real robots,
to work with system, needs to be able to read color information from display and
communicate with control center or communicate together in multiple robot tasks.
Basic functional diagram of the system is showed in Fig. 2.
Basic parameters of robot are RGB sensors to read LCD display and Wi-Fi device
to communicate with the system. For the system purpose, universal version of the
robot is aiming to handle as much as possible scenarios for the user, with space to
extend robot parameters for specialized tasks.
∙ Universal robot
– RGB sensors
– 3D gyroscope
– 3D accelerometer
– 3D compass
– Microcontroller
– Drives with controllers
Aeris—Robots Laboratory with Dynamic Environment 173

Fig. 2 Aeris simplified system functional diagram

– Wi-Fi communication module

– Slots for expansion boards
Visualization device is horizontally placed LCD display, protected with poly-
methylmethacrylate 2 mm thick transparent layer. As an extension, there is a goal
to add touchscreen layer to extend possibilities to track robots on display.
∙ Playground physically created with horizontally placed LCD display.
– LCD Display
– Protective layer
– Touch layer (optional)
Control center is a program with client-server based programming model with
high modularity running on a computer, controlling system tasks as LCD driving,
camera reading, Wi-Fi communication, environment simulation, and algorithms
processing.
∙ Supervising Control and Simulating Center tasks
– LCD display controlling (graphical input to LCD panel)
– LCD display External state reading (Camera reading the activity of physical
robots on the map)
– Mechanical robot communication (Wi-Fi module)
– Environment processing
Virtual robots processing
Real robots processing
Virtual environment processing
Real environment processing
Algorithms
174 M. Chovanec et al.

2.2 Platform for New Dynamic Playground—RGB Sensors

and LCD Display

The basic goal for the transition from a static playground to the dynamic playground
was to find the platform for dynamic playground which can be readable with sensors
in a similar way as standard static robot playground. For this purpose, LCD LED
display was used. In a horizontal position, it is “easy to get” equipment with sufficient
dynamic parameters dependent on display type (frame rate).
Next step was to find and test sensors, applicable on the robot, able to read infor-
mation from common LCD displays. RGB sensor APDS9950 was chosen and tested
as it is Digital Proximity, RGB, and Ambient Light Sensor. This device is capable of
up to 390 readings per second. It is characterized by easily applicable I2C communi-
cation interface. Practical tests showed that it is able to read surface of LCD display
from needed range, approximately 1.5 cm in our scenario.
At the beginning, obtained values showed noise in intensity of color components,
Fig. 3, which was partly degrading information from sensors. This did not prevent
the color identification as the ratio between color channels stayed in sufficient range
to distinguish basic colors.
It was found out that sensors were affected by noise in an intensity of color com-
ponents while displayed brightness value was under 100 %. With display brightness
level on 100 %, intensity noise disappeared (Fig. 4). It is supposed to be the result of
display backlight controlling.
Next thing to deal with was a frame rate of the display. To minimize this possible
effect, sensors were set to 250 readings per second, which was over 4 times higher
than 60 Hz frame rate of used display. In measuring, as is seen in Fig. 4, random
spikes in color components were present, which is supposed to be frame rate artifacts.
These were filtered out with a median filter.

∙ Parameters of Aeris dynamic playground

– LCD LED Display Dell E5515H [3]
Diagonal Viewing Size 54.6” (1386.84 mm)
Resolution 1920 × 1080 pixels
60 Hz Frame Rate
– RGB sensor APDS 9950
Up to 390 readings of color per second
I2C communication interface
Small Package Dimensions (3.94 × 2.36 × 1.35 mm)
Aeris—Robots Laboratory with Dynamic Environment 175

Normalized sensor response

1
0.9
0.8
0.7
0.6
0.5
0.4
0 500 1000 1500 2000

Time [ms]
Red component Green component Blue component

Fig. 3 Sensors response on white light with LCD LED brightness setting on 50 %, 250 Hz sensor
sampling
Normalized sensor response

1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0 500 1000 1500 2000
Time [ms]
Red component Green component Blue component

Fig. 4 Sensors response on white light with LCD LED brightness setting on 100 %, 250 Hz sensor
sampling

2.3 Aeris Supervising Control and Simulation Center

In proposed solution, it was necessary to split system parts into independent blocks
to maximize universality and variance of experiments.

∙ environment
∙ map
∙ server
∙ robots
∙ visualization

Main part of the system is running environment, which is handling incoming

agents (robots) requests Fig. 5. Environment is processing all interactions includ-
ing physics, communication, and rewards for agents. Environment is loading map,
which can be designed manually or randomly (to simulate real world). All these parts
are connected using common TCP server, which has open connections for robots and
176 M. Chovanec et al.

Fig. 5 Visualization of virtual robots in Aeris system in 3D

visualization. Each robot can be calculated on its own computer as this is necessary
for complex AI algorithms like huge neural networks real-time calculations. More
than one visualization computer can be also used, so the experiment can be visualized
on any machine connected to the server via Internet. It is planned to do visualization
on html5 currently, so no other application will be necessary.

2.4 Possibilities and Limitations

Aeris system is able to handle static and dynamic scenarios. The size of scenario is
limited by the used display. Dynamic scenarios are limited by the frame rate of used
display and reading rate of used sensors for display reading.
From its nature, the system is able to create 2-dimensional scenarios or
2-dimensional version of specific 3-dimensional scenarios. As example, in mouse
maze it is possible to represent near presence of wall by increasing intensity of color
near the wall, to use a camera on robot pointed on the surface or use system camera
and give the information to the robot that the wall is near.
Aeris Supervising Control and Simulation Centre is being constantly developed
to become map editor, map generator, center for handling various simulations of
virtual robots behavior or maps behavior center to create “living” maps.
Aeris—Robots Laboratory with Dynamic Environment 177

Fig. 6 Aeris robot version 1

as line follower

2.5 Actual State and Future Work

The actual state of Aeris system is far from being the final version.
The control program is able to create and simulate robots. The simple static map
editor was created. Static and dynamic map for line follower was created Figs. 6 and
7. The universal mechanical robot was created and tested. All development is under
Linux in C and C++.
Actual work is focusing on mouse maze scenario, visual communication LCD to
robot, camera localization of robot with help from students working on their thesis
but basically, all parts of the system are being constantly upgraded. In next step, there
is a goal to bring user-friendly version easily usable for teaching simple robotics on
schools.

∙ Actual state and work in progress

– Robot version 2—done Figs. 8 and 9
– Robot programs for specific scenarios—done/development
– Supervising Control and Simulation Centre—done/development

Fig. 7 Early version of

Aeris system
178 M. Chovanec et al.

Fig. 8 Aeris robot version 2, top view, 5 × 5 cm base dimensions without wheels

Fig. 9 Aeris robot version 2, bottom view, 5 × 5 cm base dimensions without wheels
Aeris—Robots Laboratory with Dynamic Environment 179

– Camera detection of robot position—development

– Optical monitoring of robot communication—development
– Optical robot-to-camera communication—development
– Expansion board to enhance robot functions—development
∙ Future work
– Charging stations for robots
– Touch screen addition to the system
– Version for teaching simple robotics on schools
– Create robotic competition for Aeris system

Recently, the second version of a robot with diameters of base 5 × 5 cm was

created Figs. 8 and 9. This robot replaced first unreliable but smaller version with
base 3 × 3 cm. Aeris robot was created with the aim on universality, where the goal
was that robot will be able to handle all tasks what the system could bring.
Project web page is being developed including information about the project [10].

2.6 Aeris System in Teaching Process

Aeris system is already incorporated into teaching process on Computer engineering

study program on Faculty of management science and informatics in subjects Project
1 and 2, and also, there are preparations for another subject. Also partial works on the
project are doing troughs bachelor’s and master degree thesis. Aeris is being used
on presentation activities for public with the aim to promote the study of science
and engineering. Co-author of this paper is master degree student. The process of
creating Aeris is itself incorporated into the teaching process.
The final version of the system could handle most of the present days task needed
for teaching robotics as simplest presentation scenarios, robotic disciplines scenarios
or support for the Ph.D. study of Artificial intelligence and computer engineering.

3 Conclusion

Aeris system is aiming to become powerful robotic laboratory equipment by its

unprecedented variability, ability to change on fly different scenario maps and robots
behavior programs. With Aeris, there is a possibility to bring till now simulation
only experiments closer to the real world, closer to audience. Evacuation problems,
multi-agent problems, robotic challenge scenarios with totally new possibilities to
create interactive maps by a new and fresh way, all together on the same platform.
Real robot interacting with simulated robots on “living map”. Implementing sim-
plest robot scenarios as line follower, trough robotic football to ant colony. Limits of
playing with robots are pushed forward with Aeris system.
180 M. Chovanec et al.

Acknowledgments This work was supported by Foundation of Kia Motors Slovakia in Foundation
Pontis in project Umelá inteligencia hravou formou.

References

1. 8-bit AVR Microcontroller with 1K Byte of In-System Programmable Flash (2002). http://
www.atmel.com/images/doc0838.pdf
2. Balogh, F.: Mobile robots contest ISTROBOT. Acta Mechanica Slovaca, 2-A/2006 (2006)
3. Avago Technologies: APDS-9950 Digital Proximity, RGB and Ambient Light Sensor (2015).
http://www.avagotech.com/docs/AV02-3959EN
4. First Lego League. http://www.firstlegoleague.org/
5. Miček, J., Karpiš, O., Kochláň, M.: Audio-communication subsystem module for Yrobot—a
modular educational robotic platform. In: EDERC 2014 [electronic source]: Proceedings of
the 6th European Embedded Design in Education and Research: 11–12 Sept., 2014, Milano-
Italy.-[S.l.]: IEEE, 2014.-978-1-4799-6842-8.-CD-ROM, p. 60–64 (2014)
6. Kochláň, M., Hodoň, M.: Open hardware modular educational robotic platform—Yrobot In:
RAAD 2014 [electronic source]: 23rd International Conference on Robotics in Alpe-Adria-
Danube Region: 3–5 Sept. 2014 Smolenice Castle, Slovakia: conf. proc. - Bratislava: Slovak
University of Technology, 978-80-227-4219-1 (2014)
7. Kulich, M., Chudoba, J., Kosnar, K., Krajnik, T., Faigl, J., Preucil, L.: SyRoTek–distance teach-
ing of mobile robotics. IEEE Trans. Educ. 56(1), 18–23 (2013)
8. Arvin, F., Murray, J., Zhang, C., Yue, S.: Colias: an autonomous micro robot for swarm robotic
applications. Int. J. Adv. Robot. Syst. 11(113), 1–10 (2014)
9. Arvin, F., Krajník, T., Turgut, A.E., Yue, S.: COS𝜙: Artificial pheromone system for robotic
swarms research. In: 2015 IEEE/RSJ International Conference on Intelligent Robots and Sys-
tems (IROS), pp. 407–412. Hamburg (2015)
10. http://aeris.fribot.sk/
UNC++Duino: A Kit for Learning
to Program Robots in Python
and C++ Starting from Blocks

Luciana Benotti, Marcos J. Gómez and Cecilia Martínez

Abstract We present UNC++Duino, an open source educative software for

learning to program a robotic kit in C++ and Python. Besides of these two indus-
try programming languages, UNC++Duino can be programmed using 2 high level
languages based on blocks are free of syntax errors. One of the block based lan-
guages included is completely iconic allowing for its use with preliterate children.
The hardware we use with UNC++Duino, the open RobotGroup robotic kit, can
be used to build different automated constructions based on an Arduino board, sen-
sors and actuators. UNC++Duino was developed within Argentinean K-12 schools
by the Universidad Nacional de Córdoba with the collaboration and support of the
Argentinean National Ministry of Science and the RISE program in Google for Edu-
cation. Its goal is to provide an engaging tool for learning to program in different
programming languages with increasing difficulty and control of the hardware.

1 Introduction

Why is it necessary to find effective and innovative ways of engaging more stu-
dents into Computer Science (CS)? Taking our country as an example, we know that
Argentinean universities graduate approximately 4000 CS students a year (compared
to 10,000 in Law and 15,000 in Economics) while the national industry needs to hire
twice that amount per year. The lack of human resources in CS has an economical
impact in Argentina. The Information and Communication Technology (ICT) indus-
try in Argentina, despite having grown intensely in the last ten years (four times in

L. Benotti (✉) ⋅ M.J. Gómez ⋅ C. Martínez

Universidad Nacional de Córdoba, Haya de la Torre y Medina Allende,
Córdoba, Argentina
e-mail: [email protected]; [email protected]
URL: http://masmas.unc.edu.ar
M.J. Gómez
e-mail: [email protected]
C. Martínez
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 181

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_17
182 L. Benotti et al.

the number of employees, and nine times in the amount of exports), is still struggling
to find qualified workers for its workforce.
Part of the reason why students do not choose to pursue a career in CS is that, nei-
ther programming, nor other CS techniques and concepts, are taught at school. Pre-
vious studies show that this lack of early CS education can influence career choices;
students may not be selecting CS simply because they do not know what CS is [4,
10]. Since it is not taught at school, misconceptions about what the discipline actu-
ally is are commonplace. Indeed, recent work found that although more than 90 %
of Argentinean high school students surveyed use computers, most of them believe
that programming means installing programs [2].
The typical K-12 student in Argentina never encounters CS topics during his/her
school years. Programming is never taught at school, not even as an optional course.
The ICT curriculum in Argentina focuses on user training classes rather than CS con-
tent. In these courses, computing entails little more than learning how to use a word
processor, a spreadsheet or how to write an online blog. Students often get bored
in their ICT classes and outperform their own teachers. This context is not unique
to Argentina; many developed countries share the same problem. For example see
the cases of the USA [20] and the UK [9]. There are some exceptions such as Israel,
where CS has been taught at (some) high schools for many years now [22], and other
countries are starting to follow. This is the case, for example, of New Zealand [1].
Children are using computers much earlier each decade. However, this intensive
use is not contributing to their knowledge of CS as a discipline nor to developing the
ability to understand a programming language. Children become software consumers
very early but they do not learn the basics of how this technology works.
Given the current situation, there is increasing consensus that teaching program-
ming in particular, and CS in general, in an engaging way in the school curriculum
is necessary. It is imperative to both include students in the technological world as
active and creative citizens, and to help them make educated choices about their
professional future.
Bergen [3] points out that programmable toys such as robots, give children the
possibility of creating, imagining, and exploring. Robot programming is at the same
time engaging, stimulating and rich of many important CS concepts where the digital
world connects to the real world.
With the goal of teaching programming and other CS important concepts in an
engaging and interdisciplinary way the main contributions of this paper are:
∙ Introducing a multilanguage robot programming platform that allows children to
change from one language to another making evident the unimportance of the
particular syntax, helping students discover new concepts and learning new tech-
niques when they are ready.
∙ Describing an open source educative robotic kit that can be used to build an
unlimited number of automated prototypes based on Arduino boards using just
a screwdriver—a 3-wheel robot, an elevator, a harvester, an automated house, etc.
∙ Proposing a set of original activities and suggestions for adapting the same activity
for preliterate, primary school and highschool students. The activities make use
UNC++Duino: A Kit for Learning to Program Robots in Python . . . 183

of the programming kit to teach not only programming but also propose how to
integrate content from other disciplines (e.g. astronomy and physics).

2 Previous Work

Starting very young, children can create, run and debug simple computer programs
using programming languages that are both challenging and attainable for their
age [5, 14]. Even children who are preliterate, have diverse background and different
developmental stages, can program tangible platforms using iconic commands [7,
11]. The effects and implications of learning CS at such an early age have been ana-
lyzed in previous work. According to Clements [5], young children who program
concrete objects have the opportunity to analyze a situation and reflect on the prop-
erties of the objects they have to manipulate.
While exploring how to teach CS to little children, researchers found that the dif-
ficulties in children programming laid in their immature motor skills and on syntax
problems [8, 16]. Thus, there has been an important development on specific pro-
gramming platforms to address the developmental traits of preschool children (such
as Toon Talk [17], Scratch Jr [15], Cherps [19], among others). The kit we present in
this paper is one of such programming platforms, particularly designed for program-
ming robots and other automated constructions. Differently from previous work, it
offers an integrated way to grow from simple block programming languages into full
fledged languages such as Python and C++.
Flannery and Berns [7] showed that as a result of robot programming in preschool,
children imagine, plan its action, and construct a robot. In their study, the authors
found that all 4–6 year old students could program short challenges and explore
robot’s capabilities.
Although most interventions to teach programming with robots achieve high stu-
dent engagement and task completion, most available robotic programming kits are
not accessible to schools for their high cost or their lack of flexibility. Some kits
are tailored to be used with one (sometimes proprietary) block-based, or otherwise
simplified, programming language. This rigidity hinders children’s and teacher’s cre-
ativity limiting the possibility of creating and solving new challenges appropriate for
different age groups.
The following studies compare different age groups performance on similar robot
programming activities. Magnenat and his colleagues [12] taught CS with robot pro-
gramming to different age groups of children using an event handler language to
program robot actions in response to different events. Comparing the groups per-
formance in the same task, they found that most children understood and solved
simple tasks such as moving a robot upon a touch of a button or identifying robot’s
instructions. However, older children performed better on complex programming
that required several conditions or events. Dagiene et al. [6] compared students from
Finland, Sweden and Lithuania from 7 to 12 years-old performing similar algorith-
mic thinking tasks exercises. Using multiple choice questions, they evaluated con-
184 L. Benotti et al.

cepts such as algorithm modularity, data structures, and control flow. They found no
strong difference across age groups, but rather among countries. The authors suggest
that educational context, academic quality and in particular, reading ability promoted
by each school system may be strongly related to learning programming. Thus, we
want to highlight the value of developing a completely open source robotic program-
ming kit1 such as UNC++Duino. Using it, teachers and students can choose not only
what kind of robotic construction they want to build—a 3-wheel robot, an elevator,
a harvester, etc—, but also the programming language they want to do it in—a high
level, iconic language or a low level language such as C++, or Python. Python is
one of the top programming languages used nowadays by universities in introduc-
tory programming classes. Having different programming languages accessible side
by side in an interactive development environment, help the students explore and
move into more complex languages when they are ready.
In this paper we present the robotic kit hardware and the multilanguage program-
ming software UNC++Duino. UNC++Duino was developed by the Universidad
Nacional de Córdoba in Argentina with the collaboration and support of the National
Ministry of Science and Technology and the RISE program in Google for Education.
The hardware, described in Sect. 3, can be programmed in parallel in different pro-
gramming languages organized in an increasing order of difficulty and control, as
described in Sect. 4. Section 5 propose a set of original activities and suggestions for
adapting the same activity for preliterate, primary school and highschool students.
The activities make use of the programming kit to teach not only programming but
also propose how to integrate content from other disciplines.

3 Hardware: The Robotic Kit

The hardware that we use was designed by the Argentinean company RobotGroup.
The RobotGroup kit, when used to build a 3-wheel robot as shown in Fig. 1b, is small
(13 × 13 × 12 cm) but it includes an interesting set of sensors and actuators. It has
2 engines and gearboxes of 200 rpm located in the two front wheels. It includes two
IR sensors located where the “eyes” of the robot are. These sensors can be used to
test for proximity. It also has two sensors on the bottom that measure the ground
reflectivity and its colour. As can be seen in Fig. 1a, the Arduino board contains
standard I/O connectors.
The kit also has a USB 2.0 port and 6 connectors for analogical 10-bit sensors. A
programmable array of user leds is also included, as well as leds indicating power,
and engine direction rotation. Extensions can be added such as the standard shields
Arduino-compatibles (WiFi, Ethernet, ZigBee, extra engines, etc.). Finally, the kit
also includes a microphone, a sound synthesizer and an IR sensor for remote control.

1
The robotic kit software as well as the hardware are open source and the sources are available at
http://masmas.unc.edu.ar and http://robotgroup.com.ar.
UNC++Duino: A Kit for Learning to Program Robots in Python . . . 185

Fig. 1 The arduino board used by the kit and a 3-wheel robot built with it

We use this hardware because of the following advantages. To begin with, it is

affordable and usable out of the box, allowing schools and parents to buy it with a
reasonable budget. At the time of writing this paper, it was 4 times cheaper than the
well known Lego® Mindstorms® (and even cheaper if less sensors and actuators of
those described here are included). Moreover, the hardware design is open and based
on Arduino boards supporting Grove System; which is widespread and continuously
evolving. The board used is illustrated in Fig. 1a. In addition, it includes several sen-
sors, actuators and ports in order to add extra ones. Finally, the kit is very flexible
allowing for the construction of an innumerable number of automated constructions.
For example, it can be used to build a 3-wheel robot (Fig. 1b), a mill or Ferris wheel
(Fig. 2a), a driverless cart (Fig. 2b), a toy combine harvester (Fig. 4a), a small crane
(Fig. 5b), etc.
The Ferris wheel (illustrated in Fig. 2a) can be programmed to turn on an off dif-
ferent leds and to play music when reacting to some programmed condition coming
from its sensors. Or it can be transformed, by just using a screwdriver and some
programming, in a driverless toy cart (illustrated in Fig. 2b) that can deliver toys to
different rooms of a house. In the next sections we give more examples of automated
constructions and how they can be programmed, but we believe that the possibilities
are endless.

4 Software: Programming with Blocks, Python and C++

The UNC++Duino programming environment includes different programming lan-

guages. Its two block languages are based on BlocklyDuino,2 a platform that builds
on Blockly. BlocklyDuino was developed by Google for programming Arduino
boards supporting Grove System3 with blocks. A fragment of code in the block lan-
guage created by us for preliterate children is shown in Fig. 3a, a fragment of the

2 https://github.com/BlocklyDuino/.

3
http://www.arduino.cc/.
186 L. Benotti et al.

Fig. 2 An interactive Ferris wheel and a driverless cart that deliver toys

Fig. 3 The same code written in UNC++Duino iconic language and Blockly

original BlocklyDuino code is shown in Fig. 3b. The code has a repeat while true
that includes an if-then-else instruction. This instruction means that if there is an
obstacle in front of the ultrasound sensor then the robot will move forward, other-
wise it will turn left.
Many educative platforms like Code.org (in their Hour of Code initiative [21]) or
MIT APP Inventor [18] use Blockly. Blockly is open-source under the Apache 2.0
License. UNC++Duino extends BlocklyDuino to adapt it to the robotic kit described
in Sect. 3 by creating new blocks for some sensors and actuators. For example, we
added a block to play different songs using the robot sound synthesizer.
More importantly, we also extended BlocklyDuino to add the iconic language
illustrated in Fig. 3a. UNC++Duino can also be programmed in other full program-
ming languages such as Python and C++, each language was selected because it
offers a different level of difficulty and expressiveness. The simplest one is the iconic
language, but it is also the one that offers less control to the student. In this language
the student cannot choose, for example, the speed of the robot, its speed is defined
UNC++Duino: A Kit for Learning to Program Robots in Python . . . 187

as a constant. The iconic language represents the move forward action by a straight
arrow that moves the robot forward a fixed length (set arbitrarily to 20 cm) in the
block code. If the student wants the robot to move exactly 1, meter she will have to
include 5 forward actions in her code (or a block that repeats 5 times). Likewise, the
turn arrow represents turning a predefined angle (set to 90 degrees). The language
also includes simplified blocks for conditionals as can be observed in Fig. 3a. This
design decision means that a different conditional block needs to be created for each
condition that wants to be tested. In the figure the condition is: is there something
blocking the view of the robot? This leads to a proliferation of blocks but we thought
it was a reasonable compromise for conditions that are frequent when teaching pre-
literate children.
UNC++Duino interface offers all programming languages in different tabs. The
automated translation into Python or C++ occurs when the student changes to the
Python or C++ tab and the interface detects that the block code has been modified.
If the Python code is modified then it is also automatically translated into C++. The
translation is not done backwards, it is not bidirectional. That is, if a student changes
directly the code in Python or C++, the block code is not updated and a warning
sign is showed in the interface. It is not possible to make the translation bidirectional
because some of the programming languages used are less expressive than others.
For example, in the iconic language it is not possible to express that the robot has
to advance exactly 10cms. We informed the students in advance that the translation
is not bidirectional, and they had no problem with it. We assume that children will
eventually explore text languages, seeking to have more control of the robot functions
and grow out of the iconic interface.
UNC++Duino is a web platform programed in JavaScript and runs in most web
browsers. The interface is minimalistic, it shows only one programming language at
a time, and the student can easily change from one language to another by tabs. It
includes a button that sends the code to the robot and another one to share the code
(e.g. with the teacher or a fellow student). The block languages include a menu of
the available blocks that the student can drag and drop building the code as a puzzle.
Since UNC++Duino is open source, it allows users to create a new block, and to
define its translation to Python and C++. We have defined already a set of blocks
that are useful for the specific hardware we have been working. But, the software was
designed to be easily extendable by adding new blocks as well as new programming
languages other than Python or C++.
We can create a new block by performing the following two steps. First, we define
the block; specifying its name, color, structure (that is, whether it requires some
nested code such as the loop or the conditional) and its parameters. The new block
definition can be directly written in JavaScript, referring by code to the jpg or png
image(s) that represent the icon. When we created the if-then-else block shown in
Fig. 3a we used two photos of the robot, one with and one without an obstacle. Instead
of programming the new block, it can be designed graphically using the Blockly tool
designed for this purpose.4

4 https://blockly-demo.appspot.com/static/demos/blockfactory/index.html.
188 L. Benotti et al.

Second, we define the translation of the new block to the specific program-
ming language we want. UNC++Duino already includes two files, the python.js and
arduino.js, that are responsible for merging the code generated by the translation of
a set of blocks into Python and C++. If we want to translate the new block (or an
existing block) in another language, we also have to develop the merging code file.
The following Python code is generated by python.js when automatically translating
the code in Fig. 3a.

def avoidObstacles(robot):
while True:
if robot.ping() <= 20:
robot.turnLeft(60,0.5)
else:
robot.forward(60,0.5)

The robot’s methods ping, turnLeft and forward chose the appropriate
sensors or actuators to use. The method ping uses the ultrasound sensor to check
that the road is clear for at least 20 cm; turnLeft rotates one wheel in order to
turn left at 60rpm during half a second, and forward rotates both engines at 60rpm
during half a second.
Once we have defined the block and generated the code, we have to the com-
pile the new code. The interface will then be updated to include the new block or
programming language.

5 Activities for Different Age-Groups in the Classroom

In this section we describe different activities developed in classroom with the

UNC++Duino platform. Our iconic programming language, allows people from a
wide range of ages to program: from preliterate children to students who are taking
their first steps in programming. This high level language let students focus on the
program structure rather than on the syntax and other technical details until they are
ready to do so. Indeed, although originally designed for little children and evaluated
with them [14], UNC++Duino proved to be engaging but also challenging for in-
service K-12 teachers when taking their first steps in programming [13]. One of our
goals of UNC++Duino, is that students can grow with the platform, so we describe
here a set of activities that we designed in order to illustrate how to take advantage
of the flexibility of the tool.
Consider again the challenge of programming a robot that could avoid objects. In
the previous section we showed how this challenge can be solved using the iconic lan-
guage and also the Blockly language (Fig. 3). We also showed how the same problem
can be solved using Python. In the Python method avoidObstacles the students
can directly modify the parameters to change, for instance, the speed of the robot,
or the distance from the obstacles that the robot tolerates. They can also make the
UNC++Duino: A Kit for Learning to Program Robots in Python . . . 189

Fig. 4 A combine harvester that collects grain and a planetary system

robot move backwards before turning by sending a negative parameter to the method
forward (something that will prove necessary if they define the ping distance too
low).
Once students have programmed the robot that avoids objects, we could add an
extra challenge: create a firefighter robot. Teachers and students should design a maze
with different objects that the robot should avoid. In a part of the maze, the teacher
turns on a candle. The robot has to find the candle and turn it off. The room has to
be dark, because the robot can detect where the candle is using photosensors.
Instead of programming a robot that avoids objects, students could program one
that collects them. With the robotic kit and a screwdriver students can construct a toy
“harvester” similar to that illustrated in Fig. 4a that can “harvest” a predefined field
inside the classroom collecting crumbled yellow paper balls representing the grain.
Fields can be delimited with color tape so that the robot bottom sensors are used to
stay inside. A competition between different students groups can be organized; the
student who collects more grain wins.
Teachers from other disciplines could integrate Robotics and CS into their cur-
riculum. A planetary project could be part of the activities. Using the educational kit,
as illustrated in Fig. 4b, we can show our students a solar system structure, and tell the
children that their challenge is to make it move like the sun and the earth move. We
can introduce programming concepts such as parameter and cycle. Younger children
would use iconic blocks, that would help them move the engines faster or slower, in
order to experiment with different speeds.
High school students could program the same experience, but they could build the
planetary system as well. In this way, they will interact directly with arduino board,
sensors and engines. The programming goal would be to decompose the problem,
and work with programming concepts such as conditionals, cycles, methods. They
could also program the constants and calculations necessary to mimic the movements
of the earth and the sun on a faster scale. Another challenge could be to turn on a led
when the country they live in is not facing the sun.
A good idea could be to create something related with students life. Elevators
make children feel curious about how they work. If we are in a mall or building,
190 L. Benotti et al.

Fig. 5 An elevator programmed with a queue and a crane. a An elevator. b A crane

children want to use the elevator. Here we define a challenge to develop an elevator
at school (Fig. 5).
For kindergarteners, we can talk about elevators and how, in general, they work.
With the educational kit, teachers could build the elevator, and little children could
program it with UNC++Duino iconic blocks. The elevator will be equipped with 3
buttons that students will activate when testing the elevator (ground floor, 1st floor
and 2nd floor).
High schools students, should construct the elevator using the educational kit.
Then, they will incorporate programming, using UNC++Duino language blocks to
begin with. Students could use events for programming the elevator. To resolve the
elevator problem, students will have to manage when different users of elevator touch
different floors buttons. How the elevator should behave to be fair to the people wait-
ing for it? More advanced students could move to Python and use the queue data
structure.

6 Discussion and Conclusions

UNC++Duino has been evaluated by teaching programming to kindergarten and

primary school children [14] as well as for in-service teacher training in a Profes-
sional Development program [13]. For precise evaluation data we refer the reader to
these articles. Below we summarize their findings.
These studies found that most evaluated subjects, starting from 6 years-old, could
learn fundamental programming concepts such as sequence, loops, parameters, con-
ditionals through UNC++Duino and were capable of applying these notions into
robot programming. However, older children (from 10 years-old) could combine
these concepts better to create a new program [14]. The study described in [13]
UNC++Duino: A Kit for Learning to Program Robots in Python . . . 191

found that, after a 50-hour teacher training program, teachers could use the program-
ming concepts but only those with previous background in CS were able to fully
explain them to their students. The authors in [14] report that students showed high
engagement and that UNC++Duino triggered exploration. Most students explored
and commented on the different tabs with other programming languages and were
able to notice the parallelism between the equivalent program structures. Indeed,
some students were able to modify parameters in the text languages that were not
modifiable in the block languages although the activity was not designed with this
goal in mind.
In this paper we presented UNC++Duino, an open source educational software
for learning to program a robotic kit in C++ and Python starting from drag-and-drop
programming languages. One of the block based languages included is completely
iconic allowing for its use with preliterate children as well as with beginners. Besides
simplifying the initial steps, the code resulting from translating block code into text
languages is designed to be highly modular, as illustrated in Sect. 4.
In general, children grow out of computer platforms designed for a particular
age group very quickly, this is good for children but not so for teachers who some-
times cannot keep up learning to use different complex (and many times proprietary)
interfaces. Our platform encourages students to grow with it and because of it. We
acknowledge that further research is necessary with multi-language platforms for
CS Education, but this, combined with robotics could be one direction to encourage
children (and teachers) growth in CS.

Acknowledgments This work was partially funded by the grants PICT-2014-1833, PICT-2012-
712, PDTS-CIN-CONICET-2015-172, and PID-2012-2013-R18.

References

1. Bell, T.: Establishing a nationwide CS curriculum in New Zealand high schools. Commun.
Assoc. Comput. Mach. 57(2), 28–30 (2014)
2. Benotti, L., Martínez, M.C., Schapachnik, F.: Engaging high school students using chatbots.
In: Proceedings of the 2014 Conference on Innovation & Technology in Computer Science
Education, ITiCSE’14, pp. 63–68. ACM, NY, USA (2014)
3. Bergen, D.: Technology in the classroom: learning in the robotic world: active or reactive?
Child. Educ. 77(4), 249–250 (2001)
4. Carter, L.: Why students with an apparent aptitude for computer science don’t choose to major
in computer science. SIGCSE Bull. 38(1), 27–31 (2006)
5. Clements, D.H., Sarama, J.: Teaching with computers in early childhood education: strategies
and professional development. J. Early Child. Teach. Educ. 23(3), 215–226 (2002)
6. Dagiene, V., Mannila, L., Poranen, T., Rolandsson, L., Söderhjelm, P.: Students’ performance
on programming-related tasks in an informatics contest in Finland, Sweden and Lithuania. In:
Proceedings of the 2014 Conference on Innovation; Technology in Computer Science Educa-
tion, ITiCSE’14, pp. 153–158. ACM, NY, USA (2014)
7. Flannery, L.P., Bers, M.U.: Let’s dance the “robot hokey-pokey!” Children’s programming
approaches and achievement throughout early cognitive development. J. Res. Technol. Educ.
46(1), 81–101 (2013)
192 L. Benotti et al.

8. Flannery, L.P., Silverman, B., Kazakoff, E.R., Bers, M.U., Bontá, P., Resnick, M.: Designing
ScratchJr: support for early childhood learning through computer programming. In: Proceed-
ings of the 12th International Conference on Interaction Design and Children, pp. 1–10. ACM
(2013)
9. Furber, S.: Shut down or restart? The way forward for computing in UK schools. Technical
report, The Royal Society, London (2012)
10. Grover, S., Pea, R., Cooper, S.: Remedying misperceptions of computer science among middle
school students. In: Proceedings of the 45th ACM Technical Symposium on Computer Science
Education, SIGCSE’14, pp. 343–348, ACM, New York, NY, USA (2014)
11. Kazakoff, E., Sullivan, A., Bers, M.: The effect of a classroom-based intensive robotics and
programming workshop on sequencing ability in early childhood. Early Child. Educ. J. 41(4),
245–255 (2013)
12. Magnenat, S., Shin, J., Riedo, F. Siegwart, R., Ben-Ari, M.: Teaching a core CS concept
through robotics. In: Proceedings of the 2014 Conference on Innovation and Technology in
Computer Science Education, ITiCSE’14, pp. 315–320. ACM, New York, NY, USA (2014)
13. Martinez, C., Gomez, M., Benotti, L.: Lessons learned on computer science teachers profes-
sional development. In: Proceedings of the 2016 ACM Conference on Innovation and Tech-
nology in Computer Science Education, ITiCSE’16. ACM, NY, USA. In press
14. Martinez, C., Gomez, M.J., Benotti, L.: A comparison of preschool and elementary school
children learning computer science concepts through a multilanguage robot programming plat-
form. In: Proceedings of the 2015 ACM Conference on Innovation and Technology in Com-
puter Science Education, ITiCSE’15, pp. 159–164. ACM, New York, NY, USA (2015)
15. Meerbaum-Salant, O., Armoni, M., Ben-Ari, M.M.: Learning computer science concepts
with Scratch. In: Proceedings of the Sixth International Workshop on Computing Education
Research, ICER’10, pp. 69–76. ACM, NY, USA (2010)
16. Morgado, L., Cruz, M., Kahn, K.: Preschool cookbook of computer programming topics. Aus-
tralas. J. Educ. Technol. 26(3), 309–326 (2010)
17. Morgado, L., Kahn, K.: Towards a specification of the ToonTalk language. J. Vis. Lang. Com-
put. 19(5), 574–597 (2008)
18. Perdikuri, K.: Students’ experiences from the use of MIT app inventor in classroom. In: Pro-
ceedings of the 18th Panhellenic Conference on Informatics, PCI’14, pp. 41:1–41:6. ACM,
New York, NY, USA (2014)
19. Portelance, D., Strawhacker, A., Bers, M.U.: Constructing the scratchjr programming language
in the early childhood classroom. In: International Journal of Technology and Design Educa-
tion, pp. 1–16 (2015)
20. Wilson, C.: Running the Empty: Failure to Teach K-12 Computer Science in the Digital Age.
Association for Computing Machinery (2010)
21. Wilson, C.: Hour of code: we can solve the diversity problem in computer science. ACM
Inroads 5(4), 22–22 (2014)
22. Zur Bargury, I.: A new curriculum for junior-high in computer science. In: Proceedings of the
17th ACM Annual Conference on Innovation and Technology in Computer Science Education,
pp. 204–208 (2012)
Usability Evaluation of a Raspberry-Pi
Telepresence Robot Controlled
by Android Smartphones

Krit Janard and Worawan Marurngsith

Abstract Telepresence robots in static pan-tilt form can be a viable and affordable
choice for tele-education. However, cost considerations may impose limitations on
usability and expected performance. The goal of this study was to explore the
usability of a low-cost, static pan-tilt telepresence robot operated using an Android
smartphone. Experiments were carried out with 26 participants from two age groups
(14 students M = 20 years, 12 staff M = 25 years) in a laboratory. Each participant
interacted with the robot to perform two tasks. The opinions of the participants pre-
and post-experiments, and the time they took to complete the tasks, were recorded.
The results show that the average latency (of 3.1 ± 0.8 s for one robot movement)
is quite acceptable. The students were faster than the staff when controlling the
robot remotely but slower when working at the robot site. Correlation analysis
shows that confidence in the robot and the likelihood of adoption is strongly related
to data privacy features. All the methods used to control the robot remotely show
positive interaction to each other. This implies that the majority of participants were
focussed on the control methods and data privacy provided in the robot platform,
and were willing to accept a small delay in robot movement.

Keywords Usability testing ⋅

Performance evaluation ⋅ Telepresence robot ⋅
Raspberry pi ⋅
Higher education

1 Introduction

Telepresence robots already play a large role in telemedicine, tele-education and

remote offices for enterprises, as they allow users to have ad hoc verbal communi-
cation in a remote, restricted, underserved or dangerous area. Telepresence robots
have been successfully used to extend accessibility to remote education around the

K. Janard ⋅ W. Marurngsith (✉)

Faculty of Science and Technology, Department of Computer Science, Thammasat
University, Bangkok 12120, Pathum Thani, Thailand
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 193

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_18
194 K. Janard and W. Marurngsith

world [1–6]. Some common uses are allowing students with special needs to attend
classes and to perform laboratory experiments remotely, allowing remote teachers to
give lectures to underserved or restricted areas, and conducting field trips (as
reviewed in [7, 8]). Despite their broad range of applications, commercial robots are
not affordable for many institutions. Thus, several affordable robots based on
Raspberry Pi (Raspi) computers, with essential autonomy over e.g., head movement
using a pan-tilt unit, have been proposed [9, 10]. Applications involving head
movements have shown positive impacts on user involvement when interacting with
telepresence robots [11]. Thus, a pan-tilt telepresence robot with the capability to
orient its screen to face the interlocutor can be a good candidate for educational use.
Although telepresence robots in static pan-tilt form can be an affordable choice
for remote learning, cost considerations may impose issues on usability. As
reviewed in [12], usability concerns not only the effort needed to use a system, but
also the extent to which the system may be used by specific users in a specific
situation to achieve particular goals. Usability covers the user’s experience before,
during and after the interaction with the system. The focus of a robot’s usability is
on the control method and the interaction between the human and the robot. Several
control methods have been developed, along with evaluations of their usability e.g.,
using wheelchair control [13], using tricycle-style control [3], control by tablets
[14], and control by smartphones [15, 16]. Among these investigations, the
usability of robots by school children has been addressed in [3]. To use a static
pan-tilt telepresence robot effectively in a higher educational context, a systematic
usability evaluation is needed.
The goal of the study described in this paper was to explore the usability of an
affordable, static pan-tilt telepresence robot (Fig. 1), called ACTR (the Android
Controlled Telepresence Robot) in Higher Education. The robot used a low-cost

Fig. 1 Prototype of the ACTR robot

Usability Evaluation of a Raspberry-Pi Telepresence Robot … 195

Raspi2B computer as its main controller. Thus, the cost to develop the robot
prototype was US$600. To explore the usability of the ACTR, usability testing and
assessment were performed in a laboratory. Experiments were carried out on
twenty-six volunteers from two age groups (14 students M = 20 years., 12 staff
M = 25 years). Each participant performed two tasks. In the first task, the partic-
ipants stayed at the remote site. He/she used an appropriately programmed smart-
phone to remotely control the robot using three control methods (i.e., using
navigation buttons, tilting the smartphone, or using an automatic face-following
mode). The aim of this task was to observe how the participants controlled the
ACTR robot to accomplish the task. In the second task, the participants stayed at
the robot site but had a discussion with an interlocutor via the robot. During this
task, the participants were asked to walk around so that the robot could turn its head
to follow the face of the participants. Empirical data, including the satisfaction
rating from the participants pre- and post-experiments and the time to complete
each experiment, were kept and analysed.
The results showed that a majority of participants agreed that the speed of the
low-cost robot and the ways of controlling it, were acceptable. The majority of the
participants also agreed that the robot could be integrated well into educational use.
Correlation analysis showed that users’ confidence in interfacing with the robot, and
their intention to adopt one, is strongly correlated with the data privacy feature.
Moreover, the robot’s design should focus on an efficient and effective control
method in order to ensure a satisfactory user experience.

2 Android-Controlled Telepresence Robot Using

Raspberry Pi

2.1 The ACTR Prototype

An Android-controlled telepresence robot based on a Raspi2B computer, called

ACTR was used in this study. The ACTR robot, the upgraded version of [10], is a
static pan-tilt robot which can rotate its display to follow the face of a speaker who
moves during a talk. This face-following capability imitates the natural syn-
chronous posture of a listener when he/she pays attention to the speaker. The
robot’s hardware consists of a monitor, two cameras (i.e., a webcam and a Raspi
camera module), two servo motors forming a pan-tilt unit (PTU), and an AC power
supply (see Fig. 1). The Raspi2B computer and an Arduino Nano controller were
used for real-time face detection and video conferencing simultaneously. The Local
Binary Pattern (LBP) algorithm from the Open Source Computer Vision (OpenCV)
library was used for face detection. The computation was accelerated by offloading
to the GPU of the Raspi using the multimedia abstraction layer (MMAL) appli-
cation program interface.
196 K. Janard and W. Marurngsith

2.2 ACTR Robot Usage Scenario

In Higher Education, the ACTR robot can be used for remote learning, remote
teaching and group discussion. The most common use is for remote learning in
which the robot can be used to allow a student to attend a class remotely via video
conferencing. In this case, a lecturer teaches the class at the robot site and interacts
with the robot. The student remotely connects to the robot and takes control of the
robot by using an application provided for any Android smartphone. The connec-
tion between the robot and the smartphone is done by using a virtual private
network (VPN) over the Internet to ensure data privacy. Once the connection has
been established, two modes of communication operate concurrently. First, manual
control of the robot PTU runs in background mode to control the head position.
Second, the webcam interface runs in foreground mode to display the images of the
student and the lecturer on the screen of the robot and the smartphone. The student
can set the robot to freeze at its current position or control the movement of the
robot by using three control methods. These are (1) using the navigation buttons,
(2) tilting the smartphone or (3) using the autonomous face-following mode.

2.3 User Interface Design

Since the ACTR robot and the smartphone cooperate over the Internet, the design of
the user interface (UI) also considered issues regarding the user experience of a
system with multiple devices. The key issues considered in the UI design are the
conceptual model, usability, consistency, continuity, latency and reliability, as
suggested in [17]. In terms of the conceptual model, the UI of the robot and the
smartphone have an icon to show the status of the remote device (Fig. 2A and B).

(A)

(B)

Fig. 2 User interface on the robot’s display (left) and on the smartphone (right)
Usability Evaluation of a Raspberry-Pi Telepresence Robot … 197

This is to allow users to clearly see the overall connection of the system concep-
tually, and to see what might have gone wrong if the connection is not successful.
For interoperability, all of the control functionality is implemented on the Android
device as it is a suitable device in the context of use. Consistency and continuity
require the use of consistent naming for the same system features, and the ability to
continue the task on other devices respectively. Since the ACTR framework has
only one control device, the UI design does not have to cover these issues.
Regarding latency and reliability, a transparent model was used. This means that
when a remote user tries to connect to the ACTR robot, the UI will show the real
situation while it is trying to connect. The connection light is red when the robot is
not connected and blinks while it is trying to establish a connection. The connection
light turns green, when the robot has sent back an acknowledgement message.

3 Design of Usability Evaluation Experiments

In order to improve user interaction, empirical evidence on how users use the
ACTR robot to execute tasks was gathered by performing usability evaluations. The
relevant ISO standard distinguishes five usability factors: efficiency, effectiveness,
satisfaction, absence of risk, and context coverage.1 Efficiency means users can
perform the key tasks within an acceptable time interval, while effectiveness also
specifies the level of satisfaction of the users to finish the tasks. To examine
usability of the ACTR robot in a Higher Education context, both usability testing
and assessment methodologies [12] have been used. The goal was to systematically
observe how well students and staff can use the robot for discussion in a lab.

3.1 Participants

Twenty-six volunteers participated in the experiments i.e., 14 students (7 females, 7

males M = 20 years.) and 16 staff (5 females, 11 males M = 25 years.). Their
subjective experience and characteristics are summarised in Table 1. The average
height of participants is 167.7 cm. (approximately 30 cm different between the
tallest and the shortest). Most of the participants made video calls three times a
month on average, with each call lasting for approximately twenty eight min. Very
few participants engaged in teleconferencing. Many participants used Android
phones, and thought that smartphones were the main device for making video calls.
However, most of them only used computers to do an online video conversation.

1
The original ISO 9241-11 standard defines efficiency, effectiveness and satisfaction. Later in
2011, the ISO 25010 added absence of risk, and context coverage to the ISO 924-11.
198 K. Janard and W. Marurngsith

Table 1 Measures of subjective experience and characteristics of participants

Factor Range x̄ SD
1 Height (cm) 154–183 167.7 7.9
2 Frequency of video call usage (times/month) 0–6 3 2.2
3 Average length of video call usage (minutes/call) 0–60 28.3 14.6
4 Frequency of teleconference usage (times/month) 0–6 0.79 1.43
5 Average length of teleconference usage (minutes/call) 0–60 20.77 22.82
6 Mobile phone’s OS (Android:iOS:Windows) 17:07:02
7 Awareness of devices used for video calls 17:23:11:1
(Computer:Smartphone:Tablet:Other)
8 Experienced of video calls on devices 12:2:0:1
(Computer:Smartphone:Tablet:Others)

3.2 Method

The local usability testing and assessment were done in a laboratory comprising two
rooms, one emulating the robot site and the other the remote site. For the usability
testing, all participants performed two sets of predefined tasks with the robot devel-
oper. In the first task, the participant stayed at the remote site while the robot developer
stayed at the robot site. The participant used a prepared smartphone to control the
robot remotely while pretending to discuss laboratory work with the robot developer.
The participants were asked to control the ACTR robot using three control methods
i.e., (1) using navigation buttons, (2) tilting the smartphone, or (3) using the automatic
face-following mode, using one method at a time. The aim of this task was to measure
the time and perception of the participants in turning the head of the ACTR robot to
face several marked points at the robot site. The participants were asked to complete
this task in three different ways i.e., (1) moving the robot’s head in the horizontal and
vertical (H and V) directions, (2) moving the robot’s head in a diagonal direction,
(3) allowing the robot’s head to follow the face of the interlocutor autonomously.
In the second task, the participants stayed at the robot site, pretending that the
participants had to give a talk to the remote user. In this task, the participants were
asked to walk around, while the listener at the remote site configured the robot to
perform autonomous face following.
For the usability assessment, the opinions of the participants pre- and
post-experiments, and the time to complete each scenario were recorded.

3.3 Measurement and Data Analysis

The amount of time the participants required to accomplish each task was recorded
in units of seconds. The participants’ behaviour was video recorded, with the
participants’ consent. Data from questionnaires were also collected pre- and
Usability Evaluation of a Raspberry-Pi Telepresence Robot … 199

post-experiments. The questionnaire had five parts. The first part had sixteen
questions on demography and prior experience. Later parts had satisfaction rating
(using the 1–5 Likert scale). The last section of the questionnaire had a box for
written comments.
All the empirical data gathered from the experiments were entered into an Excel
spreadsheet and analysed in five steps. First, the data obtained from the students and
staff were tested to see whether they showed different responses, using the
two-sample T-test. Second, the one factor ANOVA test was used to determine
whether there were any significant differences between the means of data when
steering the robot in different directions using one control method. Third, the
average time and satisfaction rate on the Likert scale were summarised by using an
arithmetic mean with 95 % confidence interval and presented in forest plots. Fourth,
the correlation of features related to the participants’ acceptability of the robot were
analysed by constructing a correlation matrix. Lastly, the change of satisfaction
rating scores before and after the experiments were analysed by calculating the
difference and using statistical summary.

4 Experimental Results and Discussion

4.1 Efficiency

The efficiency of the ACTR robot was measured by observing the average time to
complete tasks. All participants were able to finish all tasks successfully but the
results from the students and the staff were significantly different. As shown in
Fig. 3, the students used less time to finish the tasks when controlling the robot
remotely. The students were 20 % and 10 % faster than the staff using the button
and phone-tilting control methods respectively. Phone-tilting control allowed the
tasks to be finished faster than button control at 44 % and 58 % for the students and
staff respectively. However, as shown in Fig. 4, the students were slower and
showed more variation in the total time to finish the task when they were com-
municating with the robot site, approximately 10 % slower. The total time shown in
Figs. 3 and 4 is the time taken to move the robot twenty times. On average, the
latency was 3.1 ± 0.8 s. The results clearly show the greater efficiency of the
phone-tilting control method over button control.

Students using navigation button

Staff using navigation button
Students tilting the smartphone
Staff tilting the smartphone

Average time (in seconds)

Fig. 3 Average time to complete the tasks of participants using different control methods
200 K. Janard and W. Marurngsith

Students using face following mode

Staff using face following mode

Average time (in seconds)

Fig. 4 Average time to complete the tasks of participants using automatic face-following

The robot’s movement is accurate when using button control.

The robot’s movement is accurate when tilting the phone.
The robot’s movement is accurate when using face following.

Strongly Disagree 1 2 3 4 5 Strongly Agree

Fig. 5 Perception of accuracy when controlling the robot in horizontal and vertical movement

The robot’s movement is accurate when using buttons control.

The robot’s movement is accurate when tilting the phone.
The robot’s movement is accurate when using face following.

Strongly Disagree 1 2 3 4 5 Strongly Agree

Fig. 6 Perception of accuracy when controlling the robot in diagonal movement

4.2 Satisfaction

The opinions of students and staff showed no statistical difference when moving the
robot in different directions. However, the results obtained from moving horizon-
tally and vertically (Fig. 5) were different from moving diagonally (Fig. 6). Inter-
estingly, participants gave the button control the highest score for accurately
moving the robot in the H and V direction (avg. of 3.9). However, button control
was perceived as the least accurate method when moving the robot diagonally (avg.
of 2.7). Button control also received the lowest score for the robot’s responsiveness,
i.e., having an average rating of 2.9 (see Fig. 7). The smartphone-tilting control
method also received a similar rating pattern, yet having a narrower range of rating
scores (i.e., average rating of 3.6 and 2.8 between moving in H and V and diagonal

The robot’s movement is responsive when using button control.

The robot’s movement is responsive when tilting the phone.
The robot’s movement is responsively when doing face following.
Strongly Disagree 1 2 3 4 5 Strongly Agree

Fig. 7 All participants’ perception score on responsiveness of the robot using different modes
Usability Evaluation of a Raspberry-Pi Telepresence Robot … 201

Size of images displayed on the smartphone.

Lighting and color of images shown on smartphone.
Speed of image transmission from robot to smartphone.
Loudness of sound on the smartphone.
Control of sound echo sending from robot to smartphone.
Speed of sound transmission from robot to smartphone.
Synchronization between image and sound.

Very unacceptable 1 2 3 4 5 Very acceptable

Fig. 8 Acceptability rating of image and sound quality at the smartphone site

directions respectively). However, tilting the smartphone was rated as the most
responsive method (see Fig. 7, avg. of 3.5). The participants felt that accuracy
dropped when steering the robot to move diagonally (see Fig. 6 in comparison to
Fig. 5). When ranking the average participants’ rating of different control methods
from the highest to the lowest score, the ranking is phone tilting, face following and
buttons.

4.3 Acceptability

Acceptability was analysed using three criteria (a) the Likert-scale acceptable rating
of video conferencing at the remote site, (b) the correlation of features affecting the
participants’ acceptability rating and (c) the change of satisfaction rating after using
the robot. As shown in Fig. 8, the participants found that the image and sound
quality fell between neutral to slightly acceptable. Correlation analysis shows that
the data privacy feature is strongly correlated with confidence, and the likelihood of
adopting the robot. As shown in Table 2 in bold face, the look of the hardware
design was also very much related to the likelihood of adopting the robot. After
using the robot, most participants changed their opinions on five aspects (see

Table 2 Correlation matrix showing the related features in bold face

Confidence Data HW UI Adoption Should
privacy look look likelihood cont.
Confidence 1
Data privacy 0.706 1
HW look 0.094 0.081 1
UI look 0.192 0.218 0.287 1
Adoption 0.439 0.312 0.391 0.315 1
likelihood
Should cont. 0.024 0.157 0.140 0.163 0.721 1
202 K. Janard and W. Marurngsith

Table 3 Statistical summary showing the change of satisfaction rating after using the robot
Students (Age 18–22) Staff (Age 23–40)
Strongly Disagree 1 2 3 4 5 Strongly Agree (satisfaction rating using the 1–5 Likert scale)
Before After Changes Before After Changes
(1) Participant feels confident in using smartphone to control the robot
Range 2–5 2–5 % 57 % Range 2–5 3–4 % 92 %
Average 3.3 3.3 + 38 % Average 3.2 3.7 + 73 %
Std.dev. 0.83 0.91 – 63 % Std.dev. 0.83 0.49 – 27 %
(2) Using a telepresence robot is becoming common
Range 1–5 1–5 % 43 % Range 2–5 3–4 % 33 %
Average 3.2 3.6 + 83 % Average 3.4 3.7 + 75 %
Std.dev. 1.12 1.22 – 17 % Std.dev. 0.79 0.49 – 25 %
(3) Using the robot can increase the quality of interaction
Range 2–5 2–5 % 64 % Range 2–4 3–5 % 33 %
Average 3.3 3.9 + 78 % Average 3.6 4.3 + 100 %
Std.dev. 1.07 1.03 – 22 % Std.dev. 0.67 0.62 – 0%
(4) The price of the robot (US$600) is affordable for educational use
Range 1–4 2–5 % 29 % Range 2–4 3–5 % 67 %
Average 3 3.5 + 100 % Average 3.2 3.8 + 88 %
Std.dev. 0.96 0.85 – 0% Std.dev. 0.72 0.58 – 13 %
(5) The functionality and utilisation of the robot meets your expectation
Range 2–5 2–5 % 57 % Range 3–5 3–5 % 42 %
Average 3.8 4.1 + 63 % Average 4.1 4.2 + 60 %
Std.dev. 0.97 1 – 38 % Std.dev. 0.67 0.58 – 40 %

Table 3). The staff’s opinions were less varied than the students. After using the
robots, all participants showed more positive rating for the affordability of the robot
in educational use and agreed that using the robot can increase the quality of
interaction. A majority of the participants showed more positive agreement that
using telepresence robots is becoming common. After doing the experiments, some
of the students seemed to have less confidence to control the robot.

4.4 Discussion

The results confirmed that using a static pan-tilt telepresence robot in Higher
Education is efficient and effective. The participants were quite satisfied with having
three ways to control the robot. When informed of robot price, the participants felt
that the quality of the image and sound are quite acceptable. This shows that within
this price range, the participants did not expect much improvement in the video
conferencing quality. After using the robot, many participants felt less confidence to
control the robot. This suggested that the expectation of participants was different
Usability Evaluation of a Raspberry-Pi Telepresence Robot … 203

from what the system provided. Thus, the effectiveness and ease of use of the
control methods should be improved. As the experiments were done in a local
setting, the real communication delay has not yet been observed. Thus, more work
should be carried on to assess the impact of internet delays on the quality of the
communication offered via the robot.

5 Related Work

The PEBBLES mobile telepresence robot is one of the earliest developments of a

robot for students at elementary school and high school level used in Canada and
the USA [1]. The high school version of the robot has two monitors to display the
student’s face and his/her work separately. The unique feature of the robot is a wire
hand for waving and getting attention. The robot prototype has been tested with
users, however no usability study has been reported. A commercial robot, VGo, has
also been used to enhance the educational experience of a special needs student [2].
The case study showed that the student felt that the robot has given him back his
socialization and excitement about school. Moreover, the cost of using the robot (at
US$6,000) was only one tenth of the cost to provide private lessons for students
with special needs.
In Japan, a child-operated telepresence robot using tricycle style has been pro-
posed for early education [3, 6]. Usability testing on twenty children of 4–8 years
old was used to compare two controlling styles i.e., using a tricycle device and
using a video game controller. The results showed that the children performed tasks
faster when using the tricycle interface. This research emphasizes that the interface
used to control a remote robot must suit the physical and mental state of the
expected users.
Instead of a student-controlled telepresence robot, in Korea, the
teacher-controlled, mobile Tele-education robot has been proposed to offer an
English language course [4]. A 12-week pilot project was performed in 2011 using
29 English teacher robots controlled by native teachers in the Philippines, with each
teacher conducting a class of 8 students. The results show that majority of students
found the class useful, had a strong desire to re-attend the class and were more
interested in learning English.
An affordable telepresence robot for education at university level, with an arm
capable of performing remotely in a chemistry laboratory, has been proposed in [5].
The robot was built within the budget of $400 Canadian. The case study showed
that remote laboratory and telepresence robots can provide added flexibility to
online learning and can allow special needs students to experience learning activity
in real-time.
The usability of telepresence robots has been addressed in many publications.
Yan et al. studied the usability of a pan-tilt static robot with the capability to orient
its display in the direction of a user voice [18]. Experiments performed on 15
participants showed that the top rank usage scenario suggested by the participants is
204 K. Janard and W. Marurngsith

for a business meeting. Lewis et al. evaluated the usability of two commercial
telepresence robots, the VGo (US$6,000) and AVA500 (US$70,000) [19].

6 Conclusion and Future Work

Telepresence robots have been applied to education at every level ranging from a
well-equipped, full function commercial robot for business to an affordable design
for Higher Education. Telepresence robots in static pan-tilt form can be an
affordable choice to integrate with existing online activities for Higher Education.
Usability evaluation is a key step to gather empirical evidence about how users use
the robot. Therefore, within the cost limitation, the robot’s design can be suitable
and improve user interaction. This study aimed to examine the usability of a
low-cost, static pan-tilt telepresence robot controlled using Android smartphones.
Usability testing and assessment were carried out on 26 students and staff.
Empirical data, including, the time to complete tasks and subjective satisfaction
ratings were gathered and analysed. The results showed that the participants had
neutral to positive attitude to the performance, and to the ways to control the robot.
A majority of the participants agreed that the robot could integrated well into
educational use. The results of correlation analysis highlighted that the confidence
to interface with the robot, and the intention to adopt one, is strongly related with
data privacy feature. Thus, future work will focus on enhancing the data privacy
feature and improving the degrees of autonomy such as mobility, turning the
robot’s head by voice or by poking, calling for attention by raising a hand, or a light
signal.

Acknowledgments This project has been funded by the Faculty of Science and Technology,
Thammasat University. We thank the reviewers for their valuable comments. We thank contrib-
utors to the Raspberry Pi community website for discussions and lesson learned. We thank
Professor Roland Ibbett for improving the readability of this paper.

References

1. Yeung, J., Fels, D.I.: A remote telepresence system for high school classrooms. In: Canadian
Conference on Electrical and Computer Engineering (2005)
2. Bloss, R.: High school student goes to class robotically. Ind. Robot Int. J. 38, 465–468 (2011)
3. Tanaka, F., Takahashi, T., Morita, M.: Tricycle-style operation interface for children to control
a telepresence robot. Adv. Robot. 27, 1375–1384 (2013)
4. Yun, S.-S., Kim, M., Choi, M.-T.: Easy interface and control of tele-education robots. Int.
J. Soc. Robot. 5, 335–343 (2013)
5. Denojean-Mairet, M., Tan, Q., Pivot, F., Ally, M.: A ubiquitous computing platform—
affordable telepresence robot design and applications. In: 2014 IEEE 17th International
Conference on Computational Science and Engineering (CSE), pp. 793–798 (2014)
Usability Evaluation of a Raspberry-Pi Telepresence Robot … 205

6. Tanaka, F.: Robotics for Supporting Childhood Education. In: Cybernics: Fusion Of Human,
Machine And Information Systems, pp. 185–195. Springer, Tokyo (2014)
7. Tsui, K.M., Yanco, H.A.: Design Challenges and Guidelines for Social Interaction Using
Mobile Telepresence Robots. In: Reviews of Human Factors and Ergonomics, vol. 9,
pp. 227–301 (2013)
8. Kristoffersson, A., Coradeschi, S., Loutfi, A.: “A review of mobile robotic telepresence. Adv.
Hum.-Comput. Interact. 2013, 3 (2013)
9. Prabha, S.S., Antony, A.J.P., Meena, M.J., Pandian, S.R.: Smart cloud robot using raspberry
Pi. In: 2014 International Conference on Recent Trends in Information Technology, ICRTIT
2014 (2014)
10. Janard, K., Marurngsith, W.: Accelerating real-time face detection on a raspberry pi
telepresence robot. In: 2015 Fifth International Conference on INTECH (2015)
11. Bamoallem, B.S., Wodehouse, A.J., Mair, G.M., Vasantha, G.A.: The impact of head
movements on user involvement in mediated interaction. Comput. Hum. Behav. 55, 424–431
(2016)
12. Isabel, A., Queirós, M.A., Silva, A.G., Pacheco Rocha, N.: Usability Evaluation Methods: A
Systematic Review. In: Saeed, S., Bajwa, I.S., Mahmood, Z. (eds.) Human Factors in Software
Development and Design. IGI Global (2014)
13. Gironimo, G., Matrone, G., Tarallo, A., Trotta, M., Lanzotti, A.: A virtual reality approach for
usability assessment: case study on a wheelchair-mounted robot manipulator. Eng. Comput.
29, 359–373 (2012)
14. Clotet, E., Martínez, D., Moreno, J., Tresanchez, M., Palacín, J.: Development of a High
Mobility Assistant Personal Robot for Home Operation. In: Ambient Intelligence—Software
and Applications, vol. 376. Springer (2015)
15. Emami, M., Bahmani, M.R.: Design and implementation of a robot controlled by android. Int.
J. Appl. Eng. Res. 10
16. Tetzlaff, T., Zandian, R., Drüppel, L., Witkowski, U.: Smartphone controlled robot platform
for robot soccer and edutainment. Advances in Intelligent Systems and Computing, vol. 345,
pp. 505–518 (2015)
17. Rowland, C., Goodman, E., Charlier, M., Light, A., Lui, A.: Designing Connected Products:
UX for the Consumer Internet of Things. O’Reilly Media (2015)
18. Yan, R., Tee, K.P., Chua, Y., Huang, Z.: A user study for an attention-directed robot for
telepresence. In: Lecture Notes in Computer Science (including subseries Lecture Notes in
Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 7910 LNCS, pp. 110–117
(2013)
19. Lewis, T., Drury, J.L., Beltz, B.: Evaluating mobile remote presence (MRP) robots. In:
Proceedings of the International ACM SIGGROUP Conference on Supporting Group Work,
pp. 302–305 (2014)
On the Design and Implementation
of a Virtual Machine for Arduino

Gonzalo Zabala, Ricardo Moran, Matías Teragni

and Sebastián Blanco

Abstract Arduino has become one of the most popular platforms for building
electronic projects, especially among novices. In the last years countless tools,
environments, and programming languages have been developed to support
Arduino. One of these is Physical Etoys, a visual programming platform for robots
developed by the authors. Physical Etoys supports compiling programs into the
arduino. For this to work, a Smalltalk to C++ translator has been built. Although it
has been very useful, this translator has brought a new set of issues. In this paper we
will discuss some of these problems and how we decided to overcome them by
developing a simple virtual machine that will be used as the base for the new
Physical Etoys.

Keywords Arduino ⋅
Programming language ⋅ Virtual machine ⋅
Concurrency ⋅
Physical etoys

1 Introduction

Since the emergence of the Arduino board, the world has seen a significant increase
in the amount of people without technical training (artists, designers, hobbyists) that
have started to explore the world of microcontroller programming. The educational
field has not been exempt of this trend. Following the movement that promotes

G. Zabala (✉) ⋅ R. Moran ⋅ M. Teragni ⋅ S. Blanco

Universidad Abierta Interamericana, Buenos Aires, Argentina
e-mail: [email protected]
R. Moran
e-mail: [email protected]
M. Teragni
e-mail: [email protected]
S. Blanco
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 207

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_19
208 G. Zabala et al.

teaching programming and computer science in schools, robotics offers a highly

motivational medium to introduce concepts from several disciplines. Furthermore,
the amount of knowledge necessary to carry out small to medium size robotic
projects is increasingly smaller. The Arduino board, being open hardware and low
cost, invites students from all over the world to embark on the adventure of edu-
cational robotics.
The Arduino platform provides a simplified environment (based on the C++
programming language) in which most of the advanced microcontroller concepts
are hidden away from the user. However, this environment is still too complex for
some of the most inexperienced users, especially young children.
One of the main issues we found while teaching robotics to high school students
is the limited support for concurrency in the Arduino programming language.
Robotics competitions and exercises usually require the coordination of several
concurrent tasks to achieve a goal. For instance, a common competition is the
robot-sumo, in which two robots try to push each other out of a circle. This
competition involves performing two simultaneous tasks: finding the opponent, and
avoiding being pushed out of the circle. Other competitions involve more complex
tasks. Most of the students participating in this kind of activities struggle to specify
the concurrent behavior and as a result their robots perform poorly. The experience,
thus, becomes frustrating instead of fun and engaging.
For these reasons, and taking advantage of the fact that Arduino is open source,
there have been multiple attempts to provide a programming environment more
suitable for beginners. One of these attempts is Physical Etoys (PE), an extension of
Etoys [1] designed to provide children the tools to easily program different hard-
ware platforms, including Arduino. Etoys is a media-rich authoring environment
and visual programming system that allows children of all ages to create simula-
tions and small videogames using a tile-based scripting system. Up to the creation
of PE, Etoys only allowed to manipulate virtual objects, such as drawings on the
screen. By using PE, a child can interact as easily with the virtual world provided
by Etoys as with the real world through any of the supported robotic kits.
Etoys (and by extension, PE) provides a very simple concurrency model in
which all of the user scripts run at configurable intervals inside an infinite
loop. Although execution is entirely single-threaded, from the user perspective all
the scripts are running simultaneously. This simple model is usually good enough
for most students.
However, PE required the Arduino board to be constantly connected to the
computer, which is not allowed in some robotics competitions. In order to address
this issue PE can now also compile its scripts and upload them to the Arduino using
a special programming mode. This feature involves a great deal of complexity that
beginners are not usually able to handle. And since the PE firmware cannot coexist
with the compiled programs, once the scripts are uploaded to the Arduino all the
benefits provided by PE interactive environment are lost.
Moreover, this “compiled” mode has severe technical issues. In order for the
compilation to work, PE first translates the scripts (which are automatically gen-
erated Smalltalk code) to C++, then it uses avr-gcc to compile the C++ sources
On the Design and Implementation of a Virtual Machine for Arduino 209

into machine code, and finally it uploads the machine code to the Arduino board
using avrdude. This process is highly complex and slow. It forces the PE distri-
bution to include the AVR tools, increasing its size up to 5 times (from 47 to
231 Mb in its last version). Furthermore, the AVR tools are platform dependent,
which makes it difficult to provide a single cross-platform distribution for PE.
Even though the ability to choose the programming mode depending on the kind
of project being carried out has distinguished PE from other similar projects (such
as Scratch for Arduino [2] or Minibloq [3]) the above-mentioned problems require a
different approach.
The following requirements have to be met:
1. The script execution must be performed directly on the Arduino board without
the need for interaction with the computer.
2. If the arduino happens to be connected to the computer, all the interactivity
features provided by PE must be preserved.
3. The user should not be required to specify which programming mode to use
(either compiled or interactive).
With these objectives in mind, it was decided to implement a virtual machine
that could interpret the bytecode of a very simple programming language. This
virtual machine (which was called Uzi) would be uploaded to the Arduino board
once. The computer can then send individual instructions or entire programs for the
Arduino to run by communicating with the virtual machine through the USB port.
In this paper we will discuss the design and implementation of the Uzi virtual
machine, and we will compare it with other similar technologies in order to
highlight the benefits and limitations of this solution.

2 Related Work

The development of virtual machines and high-level languages for small micro-
controllers is not new. There have been a lot of attempts to provide a different
programming environment for Arduino. Most of them are based on pre-existent
general purpose programming languages such as Java, Scheme or Python.
HaikuVM is one of such attempts [4]. It is a Java VM based on leJOS [5] that
runs on Arduino. Its compiler analyzes the Java source code in order to generate a C
program that contains the user program (stored in Flash memory as a set of C
structs) and the virtual machine that will interpret it. The user must then use the
Arduino toolset to upload the program to the board. This implementation has
benefits, such as the low memory usage by storing the user program in the Flash
memory alongside the VM, but it needs the arduino tools to compile and upload the
programs. The fact that it outputs C code allows the compiler to easily introduce
special constructs that let the user inline C code, thus allowing him to choose the
level of abstraction required for the problem at hand. HaikuVM supports almost all
of Java semantics, including garbage collection, threads, and exceptions, but it lacks
210 G. Zabala et al.

support for reflection, object finalization, weak references, and type information for
arrays. In order to efficiently use the available memory in Arduino, the compiler
performs a static program analysis that allows it to discard unused classes and thus
generate more compact programs. Regarding performance, some benchmarks show
an execution speed of “about 55k java opcodes per second on 8 MHz AVR
∼ATmega8” [4].
Ocamm-pi is a variant of the Ocamm programming language [6] that supports
several platforms, including Arduino [7]. Ocamm is especially designed to write
concurrent programs, which are difficult to express using the Arduino language. It
requires a board with at least 32 KB of space for code and 2 KB of RAM, so the
smallest Arduino boards supported are the ones that use the ATmega328
chip. Similarly to HaikuVM, the occam-pi bytecodes are stored in flash memory
alongside the virtual machine. However, unlike HaikuVM, the bytecode can be
uploaded separately. Another similarity ocamm-pi has with HaikuVM is the static
program analysis that allows it to eliminate dead-code and generate compact pro-
grams. This process is not only performed on user-generated code but also on
ocamm-pi libraries. Ocamm-pi has a rich set of runtime libraries that provide
functions for interacting with Arduino features such as the serial port, PWM and
TWI. Most of these libraries are entirely implemented in occam-pi. This is possible
thanks to interrupts and memory being accessible from occam-pi code, allowing the
development of low-level libraries directly in occam-pi. However, handling inter-
rupts using occam-pi code has a performance cost that limits the amount of
information that can be processed. For example, handling serial communication in
occam-pi can only process characters at a baud rate of at most 300 bps. Regarding
performance, the execution of bytecodes has been reported be 100–1000 times
slower than the execution of native code.
Splish [8] is an interesting project because instead of providing only a virtual
machine it also provides a visual programming environment, much like PE. All the
instructions and programming constructs are represented as icons that can be
interconnected to define the program flow. The programs built using this visual
environment are then compiled into an object code for a stack virtual machine
designed specifically for this language. Uploading the compiled programs into the
Arduino board is done via USB. The Splish firmware includes a monitor program
that is in charge of the communication between the board and the computer; it
listens to the Serial port for commands to execute and periodically sends back status
information. This allows the computer to monitor the state of the Arduino pins and
the execution of the programs. It is designed with debugging facilities in mind, even
if that has a negative impact on the performance. If the Arduino board is connected
to the PC, it can run programs in “debug mode”, allowing step by step execution.
PyMite [9] (also known as python-on-a-chip) is a Python interpreter for 8-bit and
larger microcontrollers. It can execute a subset of Python bytecodes and it supports
almost all of Python’s most important data types (such as 32-bit signed integers,
Strings, Tuples, Lists, and Dictionaries) and some advanced features such as gen-
erators, classes, and decorators. It allows writing native code by marking a Python
function with a special keyword and writing the C code in the function’s
On the Design and Implementation of a Virtual Machine for Arduino 211

documentation string, thus making it easy to develop low level libraries. It supports
several platforms, but since it requires at least 64 KB of program memory and
4 KB of RAM, Arduino boards smaller than the MEGA are not supported.
The Scheme programming language has several implementations designed for
small microcontroller based embedded systems. Two of the most interesting ones
are Microscheme [10] and PICOBIT [11], which are very different in their
approach, even though they both are implementations of the same programming
language. Microscheme targets the 8-bit ATmega chips used by most Arduino
boards, while PICOBIT targets the Microchip PIC18 family of microcontrollers.
Microscheme differs from PICOBIT in that it uses direct compilation instead of a
virtual machine. Its compiler, written in C, generates AVR assembly code which is
in turn assembled and uploaded to the board using the avr-gcc/avrdude toolchain.
PICOBIT instead provides a Scheme virtual machine written in portable C that,
although being currently implemented for the PIC18 microcontrollers, could be
ported to any platform that has a C compiler. Another characteristic worth men-
tioning of the PICOBIT approach is that it does not only provide a custom
Scheme compiler and virtual machine but also a custom C compiler designed
specifically for developing virtual machines. This C compiler takes advantage of the
patterns commonly found in the implementation of virtual machines and it performs
a set of optimizations that result in a significant reduction of the generated code.
Both implementations support different subsets of Scheme.

3 Design Principles

The main goal of this project is to provide a tool that a visual programming
environment such as PE could use to compile and run its programs.
Given that PE has an educational purpose, Uzi was designed based on the
following principles:
• Simplicity: It should be easy to reason about the virtual machine and how it does
its job.
• Abstraction: It is the responsibility of the Uzi language to provide high-level
functions that hide away some of the details regarding both beginner and
advanced microcontroller concepts (such as timers, interruptions, concurrency,
pin modes, and such). These concepts can later be introduced at a pace com-
patible with the needs of the user.
• Monitoring: It should be possible to monitor the state of the board while it is
connected to the computer.
• Autonomy: The programs must be able to run without a computer connected to
the board.
• Debugging: Uzi must provide mechanisms for error handling and step by step
execution of the code. Without debugging tools, the process of fixing bugs can
be frustrating for an inexperienced user.
212 G. Zabala et al.

4 Implementation

In order to simplify the translation of PE scripts to Uzi programs the execution

model of the Uzi virtual machine was designed to be as similar as possible to the
Etoys model. Like Etoys projects, an Uzi program can include several scripts that
are executed concurrently. Each script runs forever in an implicit loop and its
execution rate can be configured independently. This model simplifies the needed
code to express concurrent tasks, as can be seen in the example section of the
present paper.
Some of the Uzi tools reside on the computer while others run directly on the
Arduino. The computer counts with all the necessary components to parse, compile,
and transmit the programs to the Arduino board through the serial port. All these
programs were developed using Squeak, an open source version of Smalltalk. The
Arduino, on the other hand, includes all the software required to execute Uzi
programs. These tools were written in C++ (Fig. 1).
The UziParser is responsible for parsing a string written in Uzi syntax and
generating a parse tree. It was implemented using PetitParser [12], a parsing
framework that allows you to define parsers using Smalltalk code. Although the Uzi
syntax is heavily inspired by Smalltalk, it should not be confused with Smalltalk
code. Uzi does not follow any of the Smalltalk semantics. It does not support
objects nor late binding. It is a domain specific language which main purpose is to
run PE scripts inside the Arduino board and it was designed to make the translation
process as simple as possible.
An example script written in the Uzi language can be seen below. This small
program blinks the LED on pin 13:
#blink13 ticking 1 / s [toggle: 13]
The UziCompiler is responsible for traversing the parse tree and generating a
compiled program containing the bytecodes that the Uzi virtual machine will
execute.
The UziEncoder is in charge of serializing the program to a custom binary
format designed for Uzi. This format is designed to be as compact as possible

Fig. 1 Uzi architecture

On the Design and Implementation of a Virtual Machine for Arduino 213

because it will not only be used to transmit the program to the arduino but also to
store it in the EEPROM, which has very limited space.
The UziSimulator is a Smalltalk implementation of the Uzi virtual machine. This
tool allows us to run on the computer the exact same process that the Arduino will
execute. This is currently useful to verify the implementation of new functionality
before making the change in the actual virtual machine that will be uploaded to the
board. In the future, the UziSimulator might also be used to add debugging features
such as step by step execution.
The UziProtocol is the last tool on the PC side. It is used by the other com-
ponents to communicate with the arduino. It can either send entire programs or
specific commands that the arduino will execute. It also listens for arduino state
updates. All the IO primitives implemented on the UziSimulator, for example, use
the UziProtocol to actually perform the operation on the board.
On the arduino side, Uzi is installed as a firmware that contains both the Uzi
virtual machine and also a small Monitor program that communicates with the
UziProtocol through the serial port. The Monitor acts as a bridge between the
virtual machine and the development tools. It listens on the serial port waiting for
commands to execute and periodically sends data back to the computer. The
commands that the Monitor understands include IO operations, executing a specific
program, and storing a program in the EEPROM memory. The data that the
Monitor sends includes the state of the pins and the state of the virtual machine
(global variables, instruction pointer, stack, and current script).
The VM class is responsible for executing Uzi programs. It requires, essentially,
two attributes: the instruction pointer (IP), an integer that refers to the next
instruction to be executed; and a pointer to the stack. In each tick, the VM iterates
over the entire list of scripts. For each script the VM knows the time it was last
executed and its ticking rate. If the time since it was executed exceeds its ticking
rate, the VM executes the script. Executing a script involves resetting the IP and
executing each of the script’s bytecodes one by one. The execution of a script must
leave the stack exactly as it was before its execution started. The bytecode exe-
cution is handled by a simple switch statement. Since, as mentioned before, the Uzi
compiler privileges small code size over execution speed, the Uzi instruction set
was designed to use as little space as possible. Each instruction occupies one byte,
where the most significant four bits are used to represent its operation code and the
least significant 4 bits are used to specify its operand. Since 4 bits can only address
a maximum of 16 values, a special instruction is used to extend a specific operation
by using the next byte as its operand. The value 0xFF is used to mark the end of a
script. Since only 4 bits are used to represent an operation code, the instruction set
only includes the most common operations, such as handling the stack, accessing
pins, calling primitives, and starting/stopping scripts. Other operations (such as
arithmetic or logical) are implemented as primitives.
The stack has a fixed size of 100 elements. In case of stack overflow, the VM
will stop execution immediately. The invalid state will be stored and transmitted by
the Monitor to the host PC (if connected).
214 G. Zabala et al.

5 Example

The following example, although admittedly simple, is useful to show the differ-
ences between code written in the simplified C environment provided by Arduino
(which we will call Arduino code), scripts built using the PE visual interface, and
scripts written in the Uzi programming language.
This program performs four independent tasks:
1. It blinks a LED once per second (BLINK13).
2. It blinks another LED twice per second (BLINK12).
3. It turns on a third LED when a button is pressed (BUTTON).
4. It controls the brightness of a fourth LED with a potentiometer (DIMMER).
These simple tasks are performed concurrently, which is something difficult to
express in the Arduino code. As it can be seen in the example below the Arduino
code mixes the statements that perform the tasks with the code required to schedule
them at the correct intervals. This makes the code’s intention less obvious and, thus,
harder to read and modify.
The Arduino conceptual model for the pins represents another issue. In order to
read or write, it differentiates analog and digital pins, forcing the user to use
different functions for different types of pins. The abstraction is not event correct:
the function that “writes” a PWM wave is called analogWrite() even though it does
not generate an analog wave and is not related to analog pins or the analogRead()
function in any way [13]. Additionally, each pin can either be in one of two modes,
which the user must explicitly specify: INPUT for reading, and OUTPUT for
writing. A simpler model could restrict the operations that can be performed on a
pin to simply “write” and “read”, handling each specific case without exposing the
details to the user. This model has its drawback but it would be simpler to
understand for a beginner than the Arduino functions. Moreover, while digitalWrite
() and digitalRead() functions work on the same range (either 0 or 1), analogWrite()
accepts a value from 0 to 255 and analogRead() returns a value from 0 to 1023. This
small difference forces the user to transform from one scale to the other when trying
to use the input from an analog pin to output a PWM signal, as can be seen in the
Arduino example code. Failing to do this can lead to incorrect behavior, which is
difficult to debug for an inexperienced user.
Additionally, since you can’t read the value of a pin configured as OUTPUT
(without accessing the registers directly, at least), in order to blink the LEDs, the
user is forced to store the state of the pins in a variable. This extra code adds
complexity to the solution.
Handling each LED blink rate also requires extra code. Using the delay()
function, which blocks the processor for a given amount of time, as it is a common
practice in Arduino examples, is not allowed here because it would disrupt the
execution of the other tasks (the Arduino board has only one microcontroller).
Instead, the user is forced to call the millis() function and check on every tick if it is
time to blink each led.
On the Design and Implementation of a Virtual Machine for Arduino 215

All these issues with the Arduino code greatly increases the complexity of an
otherwise simple project.

In PE this same example is very different due to the fact that PE is a completely
visual programming environment. First, the user needs to indicate which type of
device is connected to each pin by clicking and dragging on icons. Then the user
has to build each script by, again, clicking and dragging the different instructions.
Each script belongs to an object and it runs concurrently with all the others. The
concurrency is automatically handled by the PE scheduler, which simplifies
describing the execution of concurrent tasks (Fig. 2). Although it cannot be seen in
the figure, each script is configured to run at different rates: 1/s. for the first, 2/s for
the second, and 100/s for the third and fourth scripts. Such configuration is much
simpler to set up in PE than in the Arduino code. Each task is encapsulated into its
own script, which simplifies reading and understanding the code. The visual
interface presented by PE is easier for beginners to understand because it makes
syntax errors impossible and it exposes the user to an object oriented API in which
each graphical object represents a real object that the user can manipulate directly.
Although the user does not have to specify each pin mode, it does have to tell PE
which devices are connected to each pin, but doing it by clicking and dragging
216 G. Zabala et al.

Fig. 2 Graphical representation of the arduino inside PE and its corresponding scripts

devices into their corresponding pins feels much more natural and intuitive than
calling a function.
Finally, the Uzi program is the smallest of the three, with only four lines of code
describing four scripts. Each script can be configured with its own ticking rate and
the Uzi virtual machine will take care of executing it at the desired interval. If the
user does not specify a ticking rate (as it is the case with the “button” and “dimmer”
scripts, then the virtual machine will execute them on every tick). It is no longer
necessary to remember the state of each pin in order to blink the LEDs, because Uzi
handles it automatically when calling the “toggle:” primitive. There is no distinction
between analog and digital pins, the only operation available is “write:value:” and
“read:” (apart from others that can be built upon these two, such as “toggle:”) and
both accept values in the 0 to 1 range. This can be seen in lines 3 and 4, where the
scripts have essentially the same statements but with different parameters. And
finally, the user is not forced to specify each pin mode explicitly; Uzi configures the
pins automatically.
#blink13 ticking 1 / s [toggle: D13]
#blink12 ticking 2 / s [toggle: D12]
#button ticking [write: D11 value: (read: D9)]
#dimmer ticking [write: D10 value: (read: A1)]

6 Limitations

Some of the design decisions that were taken during the implementation of Uzi
resulted in limitations, performance being the most important. Using a virtual
machine makes it nearly impossible to obtain the same performance that can be
obtained using native code. Although no benchmarks have been run yet, we expect
the performance to be at least 100x slower. For most of the programs we expect the
users to write using PE this might not be a problem, but for others this might
On the Design and Implementation of a Virtual Machine for Arduino 217

impose a clear limitation. One of the solutions that is being considered is to

automatically generate, apart from the Uzi bytecodes, a C++ program that could be
uploaded to the board using the arduino software. This would allow the maximum
efficiency for the cases when it is needed while maintaining the benefits of the
virtual machine approach.
The small amount of memory available on Arduino boards poses a whole set of
limitations. Currently, the EEPROM is being used exclusively for program storage,
which means that the user cannot use it to store values from its programs. Ideally,
the Monitor, VM and user programs would all be stored in Flash memory (which is
much larger), but this is not implemented yet. Until then, no strings or arrays are
supported because they occupy too much space.
The current Uzi implementation does not allow dynamic memory allocation.
This design decision has several advantages, such as making the implementation of
a garbage collector unnecessary or allowing static analysis to determine how much
memory the program will need at compile time. However, it also restricts the type
of programs that can be written using Uzi.

7 Future Work

Uzi is still a work in progress. Although most of its design is finished, only a small
subset of all primitives is currently implemented, which allows to write only simple
programs like the one described above. Once the implementation becomes stable, it
will be integrated with PE so that visually scripted Etoys projects can be translated
to Uzi bytecodes.
The Uzi language also requires better tooling. Although debugging is one of the
project guiding principles, no debugger has been implemented yet. The develop-
ment of an integrated development environment is planned for the future.
Even though Uzi is currently designed with a special focus on PE, it is of interest
for the authors to evaluate its capabilities as an intermediate language in which
different programming models could be implemented.
Finally, since the Uzi virtual machine is small and simple, porting the Uzi virtual
machine to other educational robotics platforms (such as Lego Mindstorms Nxt or
even PIC microcontrollers) is also of interest.

8 Conclusion

The design and implementation of Uzi, a virtual machine for Arduino, was
described.
This virtual machine solves a specific problem encountered while using PE to
teach robotics to high school students.
218 G. Zabala et al.

The advantages of Uzi over the traditional Arduino tools were exemplified by
writing the same program in three different programming languages: the simplified
C provided by Arduino, PE, and Uzi.
Given the advantages of Uzi over the traditional Arduino tools, its use for
educational purposes is highly encouraged.

References

1. Freudenberg, B., Ohshima, Y., Wallace, S.: Etoys for one laptop per child. In: 7th International
Conference on Creating, Connecting and Collaborating through Computing—C5 2009, Kyoto,
pp. 57–64 (2009)
2. Citilab: Scratch for Arduino (2015). http://s4a.cat/
3. Rahul, R., Whitchurch, A., Rao, M.: An open source graphical robot programming
environment in introductory programming curriculum for undergraduates. In: 2014 IEEE
International Conference on MOOCs, Innovation and Technology in Education, IEEE MITE
2014, Patiala, pp. 96–100 (2014)
4. Bob Genom: HaikuVM: a small JAVA VM for microcontrollers (2014). http://haiku-vm.
sourceforge.net/
5. Rao, A.: The application of LeJOS, Lego Mindstorms robotics, in an LMS environment to
teach children Java programming and technology at an early age. In: 5th IEEE
Integrated STEM Education Conference, ISEC 2015, pp. 121–122(2015)
6. Elizabeth, M., Hull, C.: Occam-a programming language for multiprocessor systems. Comput.
Lang. 12(1), 27–37 (1987)
7. Jacobsen, C.L., Jadud, M.C., Kilic, O., Sampson, A.T.: Concurrent event-driven programming
in occam-π for the Arduino. Concurr. Syst. Eng. Ser. 68, 177–193 (2011)
8. Kato, Y.: Splish: a visual programming environment for arduino to accelerate physical
computing experiences. In: 8th International Conference on Creating, Connecting and
Collaborating through Computing, C5 2010, La Jolla, CA, pp. 3–10 (2010)
9. Python (2014). https://wiki.python.org/moin/PyMite
10. Suchocki, R., Kalvala, S.: Microscheme: functional programming for the Arduino. In:
Scheme and Functional Programming Workshop, Washington, D.C., pp. 21–29(2014)
11. St-Amour, V., Feeley, M.: PICOBIT: a compact scheme system for microcontrollers. In: 21st
International Symposium on Implementation and Application of Functional Languages, IFL
2009, South Orange, NJ, pp. 1–17 (2010)
12. Bergel, A., et al.: PetitParser: Building Modular Parsers. In: Deep into Pharo, pp. 375–410
(2013)
13. Arduino—analogWrite() (2015). https://www.arduino.cc/en/Reference/AnalogWrite
Model-Based Design of a Competition Car

Richard Balogh and Marek Lászlo

Abstract The paper shows how students used the modeling and simulation capabil-
ities of the Matlab/Simulink to improve the control design of their winning FEImine-
tors car for the worldwide known Freescale Cup competition. Creating and simulat-
ing the model gives us a better understanding of the processes and almost bug-less
transfer of the code to the embedded processor. Model was used also for the first
estimation of a device controller. We also summarize our experiences with the com-
petition organization.

Keywords Freescale cup ⋅ Model-based design ⋅ Automotive design ⋅ Student

competition

1 Introduction

The Freescale Cup [1] is a global engineering multilevel competition where student
teams build, program and race an intelligent autonomous model car around a track.
The fastest car to complete the track without going off the track wins the race. The
competition aims to deepen student’s knowledge about the embedded control sys-
tems design.
During the design phase students must tackle several Science, Technology, Engi-
neering, and Math (STEM) related issues such as embedded microcontroller pro-
gramming, closed loop control, modeling and implementation, as well as overall
vehicle dynamics (physics). Soft skills are also trained through team collaboration,
communication and project management [2].
Detailed car design documentation is an important part of the registration process,
but unfortunately, it is not published anywhere and remains unknown for other teams.
Newcomers have to search for some rarely published solutions or have to start from
the scratch each year. In previous years some papers were published focusing on

R. Balogh (✉) ⋅ M. Lászlo

Slovak University of Technology in Bratislava, Bratislava, Slovakia
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 219

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_20
220 R. Balogh and M. Lászlo

specific aspect of the car design, e.g. the interface board design [3], or the line detec-
tion algorithm [4, 5].
Instead of detailed algorithm description, in this paper we try to show how to use
a model based design instead of ad-hoc based and intuitive approaches for successful
implementation of the controller algorithm.
Brief introduction into the competition rules is sketched in the Sect. 2. In the
Sect. 3 we show the electronic differential design, creation of the whole car model
and the controller. In the last section we summarize our experiences and recommen-
dations.

2 The Freescale Cup Competition Rules

The Smart Car race was originally conceived in 2003 in collaboration with Hangyang
University in South Korea and global company Freescale Semiconductor to increase
student exposure to cutting edge industry tools [2]. For the first time, at the Hanyang
University, they hosted 80 teams of students. Later the competition significantly suc-
ceed in China, where the number of teams quickly reached few hundreds—in 2008,
China alone hosted over 1,800 teams from over 600 universities. Since that time the
Freescale Cup has spread from Asia to both Latin and North Americas and finally
also to the Europe, impacting more than 100 schools and 15,000 students a year [6].
The competition is multi-level, in Europe it consists of few regional qualification
tournaments, one EMEA regional championship and finally the world-wide champi-
onship. As an example, the total number of 75 students from 25 teams representing
their respective universities from 11 European countries raced their cars on the 2014
EMEA region Freescale Cup track at Fraunhofer IIS in Erlangen, Germany. The
180 sq/m racetrack consisted of speed bumps, intersections, hills and chicane curves
(Fig. 1).
The spirit of the game is that students demonstrate excellent hardware integration
and superior programming.
Competition is for teams of undergraduate students (2–4) from technical univer-
sities.
Vehicle must complete a full lap and pass the start/stop line to be recorded and
registered. The car must also stop autonomously within 3 m, otherwise it is penalized.
Team has 3 attempts to achieve a full lap. The time of the first successful lap is the
time recorded for the team. This is also the most significant change since the origins
of the competition, where full three attempts were allowed and the learning strategies
were supported. Unfortunately, the learning by trial is now not supported. If any part
of one or more wheels leave the race surface, the car is considered as failed and the
time is not recorded. Absolutely no modifications of the car are allowed after the
unknown track is revealed.
Competition car: Organizers of the competition provide the competition kit consist-
ing of the car chassis with motors and electronics (processor, interface board) with
Model-Based Design of a Competition Car 221

Fig. 1 Building the race track in Erlangen 2014

the battery. The foundation of the car is a model vehicle chassis, transmission, DC
motor and servo steering. For the construction, the original and unaltered car chassis
must be used. This is especially focused on mechanical parts (tires, dc motors and
their transmissions). Footprint of the frame and the distance between wheels may not
be altered. Other changes are allowed. The car should be obtained by the students
themselves, but due to the high costs, our approach is to support the team with the
car.
Even the car is offered with the controller based on the Freedom KL-25Z board
with the car-specific shield containing all the necessary interfaces (H-bridge, servo
control and camera interface), teams are allowed to design their own electronics,
assuming they follow some rules. Only single processor, a Freescale 32-bit MCU
must be used. No auxiliary processor or other programmable device is allowed. No
DC-DC boost circuits may be used to power drive or steering motors and the total
capacity of all capacitors should not exceed 2000 uF. There is an unlimited num-
ber of sensors, but Freescale products must be used whenever possible. The base
components, chassis, electronics, steering, and drive train are shown in Fig. 2.
Parameters of the Racing Track: The track layout is not known to the challengers until
race day. Each year changes are made to the tracks which contain several elements
of difficulty including hills, hairpin turns, S-curves, and high speed straight-aways.
Width of the racing track is 60 cm, its color is matte white, with a continuous black
line (2.5 cm wide) in the middle as the pilot line. Rules also specify the minimum
bending radius, intersection angle and slope of the track (see Fig. 3 for an example
of racing track).
222 R. Balogh and M. Lászlo

Fig. 2 The car components and modules

Fig. 3 FEI-minetors team (from left Norbert Gál, Marek Lászlo and Andrej Lenčucha) at the
Freescale Cup Competition with their car
Model-Based Design of a Competition Car 223

3 Modelling

Winners of the 2014 Freescale Cup EMEA competition, the FEI-minetors team from
the Slovak University of Technology in Bratislava were not a newcomers in this com-
petition. During the initial testing and programming of the car, we recognized the
need for better understanding of various parameters and car properties. For success-
ful design it was necessary to understand not only the basic physics beyond the car
and the properties of the proposed controller, but we also need to know how certain
parameter influences another and which of them are the most important factors.
As the students of the STU in Bratislava are using the Matlab/Simulink during
the standard academic courses, it was a natural choice to use it also for modelling
the car for the Freescale Cup. As an illustration, we will show two important models
created for the competition.

3.1 Electronic Differential

The car contains two DC motors in the Ackermann steering geometry chassis, so
it was necessary to implement so called electronic differential to safely drive all
the curves in high speeds. Its function is to modify the speed of inner and outer
wheels according the steering angle. For better understanding of its function and for
easier implementation of the function in the embedded microcontroller, model of the
steering geometry was created. It includes lot of parameters, starting from geometry
(dimensions of the wheels, radius, length of the axis), including motor properties
(speed, torque).
Resulted differential model was simulated in Matlab and results were displayed in
the 3D chart of the Speed/Torque and Steering Angle for both left and right wheels
(see Fig. 5). Excerpt of the code used for simulation is on Fig. 4. It was later imple-
mented in the C programming language also in the embedded microcontroller. Later
the model and its parameters were modified based on real tests and empiric knowl-
edge. As an example, the value of the speed exponentiality c was adjusted.

3.2 Controller Design

For modelling, simulations and testing much more complicated model was created
(see Fig. 7). In Simulink, it was quite easy to start with modelling subsystems (car
geometry, DC motor, controller, etc.) and then integrate them into the one complex
layered model. We started with the standard text-book model [7] of the DC motor
with some measured and some more empiric parameters (see Fig. 6). Model com-
bines standard electric and mechanic dynamic model into the single one. Even in
this early stage, saturation and non-linearity were included into the model.
224 R. Balogh and M. Lászlo

Fig. 4 Part of the code for calculation of the torques

Fig. 5 Ackerman drive model and 3D chart of the electronic differential. Two planes represents
the respective torques for the rear left and right wheels. See the Fig. 4 for the code
Model-Based Design of a Competition Car 225

Fig. 6 Model of one of the DC motors

Fig. 7 Model of the car for the speed controller design

Fig. 8 Study of the noise influence on the proper controller operation

Later we added also model of the car chassis including its weight, dimensions—
see Fig. 7. Model of the DC motor from the previous step is the essential part of it.
Most of the car parameters were obtained by the measurement (mechanical dimen-
sions) and from experiments (conversion factors). From the beginning, it shows that
considering model non-linearities is crucial for good correspondence of the model
with reality. In each step, the parameters were adjusted according the measurements
to obtain real results.
226 R. Balogh and M. Lászlo

Fig. 9 Main control algorithm

After the subsystems were tested and compared with the real system, everything
was combined into the one complex model (see Fig. 8). The main controller of the
system was tested and parameters were modified many times based on results of the
simulations. Numeric values were obtained with a multiple iterative processes from
the first estimation and later adjustments based on experiments and their fitting to
the model. One of the most important observations were the current spikes appear-
ing almost everywhere and the final design of the power stage electronics takes this
into the account. Later, based on the real measurements and observations, further
modifications of the model was included. Finally, the controller code was almost
without changes transferred to the microcontroller. In the future, we plan to use the
Matlab embedded coder [8, 9] for the Freescale Freedom Board [10] platform, which
Model-Based Design of a Competition Car 227

we are just testing. Advantages of such hardware-in-the-loop modelling and testing

were described many times, see e.g. [11].
Later we studied the influence of non-linearities in the system. Controllers behav-
iours without and with non-linear parts in the model were compared. For speed pro-
filing we needed to use PID controller which had bad quality without linearization.
We were measuring steady state speed in correspondence to selected input in steps.
Found non-linearity was than corrected.
The main loop runs with the sampling period of 20 ms (see Fig. 9). Control sig-
nal is based on the pre-processed camera output. Within the cascade of the two con-
trollers, upper level controls the turn-off and its control signal is based on the devi-
ation of the line from the center position. Bottom level controller is based on the
previous controller output error. Its PID algorithm adjusts the overall speed of the
vehicle. Zero controller error means that the line is perfectly centered and the car can
use its full speed. The largest the deviation of the black line from the central position
is, the slowest speed is used to drive the car, providing the enough time to turn-off
the vehicle and find the central position again. Output signals are then processed in
the module of the electronic differential where the actual speeds of both wheels are
calculated. Result is then feed to the appropriate DC motor in the form of PWM sig-
nal. The model was also used for testing the influence of the noise induced on various
levels (see Fig. 8) and its results were used when designing the shielded cables from
the camera sensor.

4 Conclusion

Organization: In China, the Ministry of Education has embraced the event making
it standardized curricula across several of the leading universities [2]. Also at some
other universities the competition car design was successfully included into their
curricula (see e.g. [12]) but this was not our case.
At the STU the students prepare for the competition in their free time only. We
attempted to involve the students within their compulsory team-projects, but those
were not so motivated and successful. Personal motivation of the students include the
deep understanding of the new technology, real problems and possibility to compare
with other teams.
Key to the success in our case was the ability and passion to spent many hours
with testing and redesigning the car again and again. The task seems to be an easy,
but for students, especially in their first years at the university this is not true. They
have to fight with a new software packages, their installation, mastering the develop-
ment cycle, and with all the hardware issues at the same time. Especially when the
team choose to design also their proprietary hardware, the time constraints were cru-
cial. Otherwise, probably the main challenge was to implement the image processing
routines into the embedded microcontroller.
228 R. Balogh and M. Lászlo

Experiences: Creating and simulating the model gives us (a) better understand-
ing of the processes and (b) almost bug-less transfer of the code to the embedded
processor and (c) first estimation of the controller parameters.
Model is not perfect, but even in this form it is better than nothing. It helped us
to inderstand the behaviour of the car, which physical quantities are important more
and which are less. This leads to understanding, what can be neglected and what is
important. It is a big step over the trial-and-error approach often applied by students.
Model adjusting shows us a lot of non-linearities of the real system, especially
limits are important. This is rarely seen in textbooks examples on controller design.
In this paper, we showed just selected parts of the car modelling, as described in
[13], which is available on request (in Slovak only). We intentionally didn’t mention
the modelling of the steering servo, and we also didn’t mention the problems with
the image sensor and with the line detection in the signal due to the lack of the space
available.
Racing is a challenge that virtually every human knows, has no language barriers,
and never fails to provide excitement, adrenaline and with it a platform to educate [2].

Acknowledgments Publication of this paper has been supported by the Slovak Research and
Development Agency, Grant No. APVV-0772-12.

References

1. The Freescale: The Freescale Cup 2014 EMEA Rules v 1.2 11 (2014). https://community.
freescale.com/docs/DOC-94949
2. McLellan, J., Mastronardi, A.: Engaging students: the growing smart car competition. In: 2009
Annual Conference and Exposition, Austin, Texas (2009). https://peer.asee.org/4983
3. Zhicong, S., Xuemei, L., Mei, C., Hongbin, Z.: The design of smart car based on Freescale
processor. In: 2010 International Conference on Computer and Communication Technologies
in Agriculture Engineering (CCTAE), Jun 12, vol. 2, pp. 508–510. IEEE (2010)
4. Wang, Z., Liu, Y.: Design of road tracing navigation control for smart car use CCD sensor. In:
2010 International Conference on E-Health Networking, Digital Ecosystems and Technologies
(EDT), April 17 vol. 1, pp. 345–348. IEEE (2010)
5. Xiuquan, W., Xiaoliu, S., Xiaoming, C., Ying, C.: Route identification and direction control
of smart car based on CMOS image sensor. In: ISECS International Colloquium on Comput-
ing, Communication, Control, and Management, 2008. CCCM’08, vol. 2, pp. 176–179. IEEE
(2008)
6. Mc Lellan, J.: History of the Freescale Cup (2016). https://community.freescale.com/docs/
DOC-1011
7. Krishnan, R.: Electric Motor Drives: Modeling, Analysis, and Control. Prentice Hall (2001)
8. de Sean, W.: MATLAB and Simulink for Embedded Systems and Robotics. The
MathWorks, Inc. (2014). https://www.buffalo.edu/content/www/ccr/support/software-
resources/compilers-programming-languages/matlab/_jcr_content/par/download_1/file.res/
MathWorks_MATLAB_and_Simulink_for_LEGO.pdf
9. Hahn, C.: Autonom fahren auf unbekannter Rennstrecke. MathWorks untersttzt den Freescale
Cup. Matlab Expo Deutschland, MathWorks (2015)
Model-Based Design of a Competition Car 229

10. Semiconductor, Freescale. FRDM-KL25Z: Freescale Freedom Development Platform for

Kinetis KL14. 15/24/25 MCUs. Documentation. http://www.nxp.com/products/software-
and-tools/hardware-development-tools/freedom-development-boards/freedom-development-
platform-for-kinetis-kl14-kl15-kl24-kl25-mcus:FRDM-KL25Z
11. Kelemenová, T., Kelemen, M., Miková, L., Maxim, V., Prada, E., Lipták, T., Menda, F.: Model
based design and HIL simulations. Am. J. Mech. Eng. 1(7), 276–281 (2013). http://pubs.
sciepub.com/ajme/1/7/25/
12. Zhao, Z.: The reform and practice of education mode in multi-disciplinary joint graduation
design. Procedia Eng. 15, 4168–4172 (2011). ISSN 1877-7058. http://dx.doi.org/10.1016/j.
proeng.2011.08.782 (2011)
13. Lászlo, M.: Navigation and robot control with speed optimalization. Master thesis. Slovak
University of Technology in Bratislava (2015). http://www.crzp.sk/crzpopacxe/openURL?
crzpID=42928&crzpSigla=stubratislava
 Part V
Measuring the Impact of Robotics
on Students’ Learning
Student-Robot Interactions in Museum
Workshops: Learning Activities
and Outcomes

Alex Polishuk and Igor Verner

Abstract Our study proposes an approach based on learning through interaction

with robots. We investigated how this approach can contribute to the development
of systems thinking skills in elementary school students. 346 students of 2nd to 4th
grades participated in the workshop “Animal-like robots” conducted at MadaTech.
The workshop activities focused on inquiry and design of robot behaviors. We
evaluated students’ performance of the workshop tasks and found indications of
significant progress in the systems thinking skills.

Keywords Science museum ⋅

Workshop ⋅ Learning through interaction ⋅
Robot behavior ⋅
Systems thinking

1 Introduction

Educational robotics facilitates experiences, in which learners acquire knowledge in

joyful ways. Playful robotics activities are especially meaningful for elementary
education to introduce complex concepts such as system, process, structure and
behavior to young students [1]. Elementary schools nowadays are being faced with
the challenge of teaching such concepts in the science-technology curriculum [2].
According to the curriculum, the students are expected to learn system concepts and
apply them to problems of the natural world, technology, and everyday life.
Studies show that the traditional approach focusing on learning through practice
in building robots is attractive not for all students, but for those who have technical
skills and motivation for hands-on activities [3]. In recent years, there has been a

A. Polishuk
Israel National Museum of Science, Technology and Space (MadaTech), Haifa, Israel
e-mail: [email protected]
I. Verner (✉)
Faculty of Education in Science and Technology, Technion – Israel Institute of Technology,
Haifa, Israel
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 233

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_21
234 A. Polishuk and I. Verner

growing tendency to develop and implement alternative “pathways into robotics”

that put emphasis on understanding, discussion and cooperation, combine science
and art. In this regard, robotics finds more and more use in science museums and
centers to bring the benefits of the technology enhanced environments into exhi-
bition and classroom [4, 5]. The literature discusses innovative projects, developed
through collaboration with academic researchers, where robots play diverse roles as
exhibits, entertaining guides and remote facilitators [4, 6–8].
Technion Center for Robotics and Digital Technology Education in collabora-
tion with MadaTech Gelfand Center for Model Building, Robotics & Communi-
cation conduct a long-term research aimed to investigate and develop new strategies
of learning through interaction with robots. The rationale behind the
interaction-based approach is that learning through interaction with a robot is in line
with learning from interactive exhibits, the central and inherent activity in science
museums. The study focuses on three directions: (1) Learning through inquiry and
creating behaviors of animal-like robots, (2) Public robot theatre performances [4],
(3) Science lesson mediated by a humanoid RoboThespian [6].
This paper presents a case study in the first direction that followed up the
workshop “Animal-like robots” delivered for elementary school students, the main
audience of the MadaTech museum. A theatre performance of animated robots was
given to the school groups visiting MadaTech [4], to inspire their participation in
the workshop. In the paper we discuss the learning activities, student-robot inter-
actions, and outcomes of the workshop.

2 The Workshop Outline

The workshop was given, depending on school’s choice, in two versions: the 9 h
basic version and the 18 h extended version. In our study 97 students participated in
the basic workshop and 249 in the extended one. Both workshop versions com-
prised the learning activities presented below:
• Introduction to robotics. The students were exposed to live demonstrations and
videos of different types of advanced robots and learned basic robotics concepts.
• Inquiry into robot behaviors. Each pair of the students got a pre-programmed
“RoboCroc” made of the LEGO WeDo kit (Fig. 1a). The students tested
responses of the robot to different “stimuli” (move around, voice, light, and
touch) and made conclusions about the robot behaviors.
• Creating behaviors of RoboCroc. The students designed and implemented new
reactive robot behaviors, using the graphic programming language. They
practically tested the robot behaviors and iteratively improved them (Fig. 2a).
• Exploring the RoboCroc pulley system. The students inquired different options
to connect driving and driven pulleys of the robot. They examined how the
speed and direction of rotation of the driven pulley depends on the connection.
Student-Robot Interactions in Museum Workshops: Learning … 235

Fig. 1 a The RoboCroc. (a) Motion sensor Motor USB Hub

b The RoboMonkey

(b)

Cam-follower mechanism

• Constructing a “RoboDog” and equipping it with a tilt sensor. The students in

pairs built the RoboDogs (presented in Fig. 3) and “trained” (programmed)
them to execute commands. The targeting was realized by means of positioning
the tilt sensor (Fig. 2b).
• Creating interactions between RoboCroc and RoboDog. Pairs performed the
assignment working in twos. The students developed scenarios of interactions
and implemented them through collaboration (Fig. 2c).
The extended 18 h version of the workshop included additional activities: the
students learned how the motor and the control system determine the robot
behavior. They also constructed, programmed and tested behaviors of one more
robot “Drumming RoboMonkey” (Fig. 1b). This robot had a sound sensor and a
cam-follower mechanism to drive RoboMonkey’s arms.

3 The Study

Our research aimed to evaluate learning outcomes of 2nd to 4th grade students
participated in the robotics workshop. Among more than 2000 participants of the
workshop, the research sample included 346 students (grades 2−4) from five dif-
ferent schools in Haifa. In the study the sample was divided into three research
groups. The first group was 97 students participated in the 9 h workshop. For this
group the consideration focused on the analysis of learning behaviors. The second
group of 163 students studied the 18 h workshop. With this group we evaluated and
improved the learning activities, as well as tested and validated the worksheets as
instructional and research tools. The third group consisted of 86 students who
236 A. Polishuk and I. Verner

Fig. 2 Testing behaviors: a. The RoboCroc; b. The RoboDog; c. Robot-robot interaction

studied the 18 h workshop, among them 46 s grade and 40 fourth grade students,
51 boys and 35 girls. The study of this group focused on evaluation of the
development of systems thinking skills, based on the use of the developed research
tools. Here we address the following research question: Are there indications of the
development of systems thinking skills in students participated in the robotics
workshop? If so, what is the students’ progress?
The study applies the model for evaluation of systems thinking skills develop-
ment in school students, proposed by Richmond [10].
Student-Robot Interactions in Museum Workshops: Learning … 237

Fig. 3 Worksheet “Design and creating behaviors of RoboDog”

238 A. Polishuk and I. Verner

3.1 Richmond’s Model of Systems Thinking

Educational literature emphasizes the need for importance of systems thinking and
calls for pedagogies that develop systems thinking skills starting from primary
school [9]. Richmond [10] proposed such pedagogies to be based on active learning
through interaction with and construction of engineering systems. He considered
systems thinking as a construct of seven constituent skills: structural thinking,
dynamic thinking, generic thinking, operational thinking, scientific thinking,
closed-loop thinking, and continuum thinking. Richmond proposed to evaluate the
development of systems thinking in students of grades 4−12 by three levels of
competence (basic, intermediate and advanced) and characterized these levels for
each of the seven skills in general terms.

3.2 Method

The evaluation study was designed to consist of three stages. At the first stage we
adapted the Richmond’s model for our case, using the inductive method and
qualitative analysis of learning behaviors [11]. The data collected by observations
and video recording were analyzed through the following steps: students’ expres-
sions related to systems thinking, grouping the expressions into patterns, and
characterizing each pattern by a feature related to one of the skills defined in the
Richmond’s model. Using these features, at the second stage we developed five
worksheets directed to facilitate systems learning and evaluate its outcomes. The
worksheets are discussed in Sect. 3.3. At the third stage we developed scoring
rubrics for evaluating the level of student competence, based on the worksheet
assignments performance.

3.3 Specific Features of Systems Thinking

Table 1 presents the systems thinking skills in the Richmond’s model (1st column),
their short interpretations (2nd column) and specific features found at the first stage
of the study (3rd column).

3.4 Worksheets as Instructional and Research Tools

Worksheets. The workshop activities were supported by a series of worksheets.

The worksheets were used by the students to record the data of experiments in the
tasks they performed. The worksheets also were research tools to evaluate the
Student-Robot Interactions in Museum Workshops: Learning … 239

Table 1 Specific features of systems thinking skills

Skills Interpretations Specific features
Structural Thinking in terms of units of Identifying components of the robot
thinking measure or dimensions structure and their causal
relationships. Robot construction
and planning
Dynamic The ability to see and deduce Testing robot behaviors and their
thinking behavior patterns description by flowcharts
Generic The ability to apprehend the Understanding the input-output
thinking similarities in the feedback-loop relations, their representation by the
relations that generate cycles if-then statement, and using the
statement for planning reactive
behaviors
Operational Thinking in terms of how things Understanding mechanisms behind
thinking really work robot operation
Scientific Rigorous hypothesis-testing with Understanding physical parameters
thinking modifying only one thing at a time of the robot, generating and testing
and holding all else constant hypotheses about its reactions on
stimuli
Closed-loop The ability to look to the circular Programming robot behaviors using
thinking cause-effect relations responsible for feedback control
generating the behavior patterns
exhibited by a system
Continuum The ability to see connections and Identifying, planning and modifying
thinking interdependencies continuous robot movements and
interactions between robots

students’ progress in systems thinking skills during the workshop. The time gap
between the lessons at which the students were assigned to complete the worksheets
was 3−4 weeks.
The worksheets were validated by two educational robotics researchers and two
science teachers. The list of the worksheets is presented below:
Worksheet 1: Inquiry and design of RoboCroc behaviors. In the 1st assignment
of the worksheet the students were asked to document the change of RoboCroc
behaviors in response to stimuli. The possible behaviors before enacting the stimuli
were given: breathing, attack, and relax (listed in the 2nd column of Table 2). The
assignment was to identify and fill in the 3rd column the behavior responses to the
stimuli. The students were asked also to identify and record in the table the changes
in the robot behavior after ending the stimuli. The 2nd assignment of the worksheet
was to propose a new behavior of the RoboCroc and describe it by an “if-then”
statement. In the 3rd assignment of the worksheet the students were given a picture
of the robot and were asked to list the components of the robot and show their
locations in the picture.
Worksheet 2: Design and creating behaviors of RoboDog. The students were
asked to document the RoboDog behaviors that they developed. Like in the first
240 A. Polishuk and I. Verner

Table 2 Stimulus-response scheme of RoboCroc behaviors

Stimuli State before enacting the stimulus State in response to the stimulus
Yes Breathing attack
Attack retreat-attack
Relax retreat-breathing
No Breathing breathing
Attack retreat-relax
Relax relax

worksheet, they described and recorded reactive behaviors of the robot, and iden-
tified the robot components (Fig. 3).
Worksheet 3: Design and creating behaviors of “Drumming RoboMonkey. The
assignments in this worksheet were similar to that in worksheet 2.
Worksheet 4: Exploration of mechanical transmissions. The students docu-
mented results of their inquiry on how the jaw’s rotation and body movement are
affected by changes of the pulleys’ connection.
Worksheet 5: Exploration of cam-follower mechanism. The students documented
the effect of changes of the cam-follower mechanism on the drum rhythm and arms
synchronization of the RoboMonkey.
The mapping of the seven systems thinking skills addressed in each assignment
of the worksheets is presented in Table 3. For instance, each of the first three
worksheets consisted of three assignments. The second of them addressed the
development of dynamic, generic and closed-loop thinking skills.
Scoring rubrics. The rubrics for evaluation of students’ systems thinking skills
were developed for all the worksheets. Every rubric relates to a certain category of
systems thinking and has criteria for evaluating the assignment performance at three
levels (basic, intermediate and advanced). For example, the rubrics related to
“Design and creating behaviors of a RoboDog” are presented in Table 4. For each
rubric the table presents an assignment name, a criterion, and its specifications for
three levels of performance.

Table 3 Mapping of the systems thinking skills

Systems thinking skills Worksheets 1−3 Worksheet 4 Worksheet 5
Ass. 1 Ass. 2 Ass. 3
Structural thinking ☑ ☑
Dynamic thinking ☑ ☑ ☑
Generic thinking ☑ ☑
Operational thinking ☑ ☑
Scientific thinking ☑ ☑ ☑
Closed-loop thinking ☑
Continuum thinking ☑
Student-Robot Interactions in Museum Workshops: Learning … 241

Table 4 Rubrics for evaluation of the worksheet assignments

Rubric Skills Criterion Task performance levels
Basic Intermediate Advanced
Design of Generic, Design plan of Incorrect design Incomplete Complete
robot’s Scientific, RoboDog’s of behaviors design of design of
behavior Continuum behaviors behaviors behaviors
Description Dynamic, Description of Use of Use of Use of
of robot’s Generic, RoboDog’s anthropomorphic “mixed technological
behavior Closed-loop reactive behaviors language language” language
by “if-then”
diagrams

4 Findings

In the study we analyzed worksheets performed by 89 students participated in the

workshop. In the 1st assignment of the 1st worksheet only 15 % of the students
identified the RoboCroc’s simple and complex behaviors, while 45 % were at the
basic level (Table 5). The analyses of the 1st assignment of the 2nd worksheet,
comparative to the similar assignment in the 1st one, indicated a significant increase
in the students’ performances of this task: 60 % of the students not only identified
the robot behaviors but also correctly design them and only 25 % remained at the
basic level, while in the 1st worksheet the correspondent percentage was 15 and
45 %.
Our findings from the analysis of “if-then” statements on the cause-effect rela-
tions between robot’s response and the stimulus compiled by the students in the 2nd
assignment were as follows (Table 6). In the RoboCroc worksheet most of the
students described these statements using the anthropomorphic language (in terms
of body parts), while only 7 % compiled at the advanced level using the techno-
logical language with more than half of the students (54 %) at the basic level. In the
RoboDog worksheet, already 29 % achieved the advanced level with only about
one third (35 %) remained at the basic level.
Table 7 presents and juxtaposes the evaluations of performance of the 3rd
assignment in the worksheets related to RoboCroc and RoboDog. As shown, for the
RoboCroc worksheet, the first one done by the students, only 15 % performed the
task of identification robot components at the advanced level, while 54 % were at

Table 5 Inquiry of robot behaviors

Performance levels Percentage of students
1st worksheet 2nd worksheet
Identifying simple behaviors 45 25
Incorrect identification of simple and complex behaviors 40 15
Complete identification of simple and complex behaviors 15 60
242 A. Polishuk and I. Verner

Table 6 Description of robot behavior

Performance levels Percentage of students
1st worksheet 2nd worksheet
Use of anthropomorphic language 54 35
Use of mixed language (anthropomorphic and 40 36
technological)
Technological language 6 29

Table 7 Identification of robot components

Performance levels Percentage of students
1st worksheet 2nd worksheet
Noting in terms of body parts 54 22
Incomplete specification of components 31 49
Complete specification of components 15 29

the basic level. In the RoboDog worksheet, offered at the later stage of the work-
shop, already 29 % of the students demonstrated the advanced level and only 22 %
remained at the basic level.
Figure 4 presents students’ visual expressions and notes made by the second
graders at the workshop and collected by the school teachers. The drawings show
students’ engagement and the notes indicate that they enjoyed the learning activities
(Fig. 4).

Fig. 4 Students’ expressions of the workshop

Student-Robot Interactions in Museum Workshops: Learning … 243

5 Conclusions

Our study indicates that the proposed approach, based on learning through inter-
actions with animal-like robots, can be effectively implemented in the museum
environment. The approach engaged children in the early years of elementary
school in joyful learning in which they gained understanding of the robot system
and the principles of its programming and operation. To answer the research
question, we applied the Richmond’s model to evaluate students’ progress in the
development of the systems thinking skills. We found that activities in inquiry and
design of robot behaviors can foster students’ systems thinking skills. The progress
was achieved in all the categories of systems thinking, defined by Richmond.
Children’s reflections on the workshop activities were very positive. They were
closely engaged in and excited by the interactions with robots and gladly collab-
orated in inquiry and creation of robot behaviors. Based on our positive experience
in the museum environment, we recommend to examine the proposed approach also
in formal education settings.

Acknowledgments The study was supported by the Technion Center for Robotics and Digital
Technology Education and by the MadaTech Gelfand Center for Model Building, Robotics and
Communication.

References

1. Hipkins, R., Bull, A., Joyce, C.: The interplay of context and concepts in primary school
children’s systems thinking. In: Hammann, M., Reiss, M., Boulter, C., Tunnicliffe, S.D. (eds.)
Biology in Context. Institute of Education, University of London, Learning and teaching for
the twenty-first century (2008)
2. National Standards: STEM Standards, (March 2015) http://www.clexchange.org/curriculum/
standards/stem.asp
3. Rusk, N., Resnick, M., Berg, R., Pezalla-Granlund, M.: New pathways into robotics:
Strategies for broadening participation. J. Sci. Educ. Technol. 17(1), 59–69 (2008)
4. Verner, I., Polishuk, A., Klein, Y., Cuperman, D., Mir, R., Wertheim, I.: Robotics education
through a learning excellence program in a science museum. Int. J. Eng. Educ. 28(3), 523–533
(2012)
5. Shibata, T., Wada, K., Tanie, K.: Subjective evaluation of seal robot at the national museum of
science and technology in Stockholm. In: Proceedings of the IEEE International Workshop on
Robot and Human Interactive Communication, Millbrae, California, pp. 397–402 (2003)
6. Hashimoto, T., Kobayashi, H., Polishuk, A., Verner, I.: Elementary science lesson delivered
by robot. HRI 133–134 (2013)
7. Horn, M.S., Solovey, E.T., Jacob, R.J.K.: Tangible programming for informal science
learning: Making TUIs work for Museums. Proceedings of 7th International Conference on
Interaction Design and Children IDC, pp. 194–201. ACM, NY (2008)
8. Nourbakhsh, I., Hamner, E., Ayoob, E., Porter, E., Dunlavey, B., Bernstein, D., Crowley, K.,
Lotter, M., Shelly, S., Hsiuand, T., Clancey, D.: The personal exploration rover: educational
assessment of a robotic exhibit for informal learning venues. Int. J. Eng. Educ. 22(4), 777–791
(2006)
244 A. Polishuk and I. Verner

9. Riess, W., Mischo, C.: Promoting systems thinking through biology lessons. Int. J. Sci. Educ.
32(6), 705–725 (2010)
10. Richmond, B.: Systems thinking: critical thinking skills for the 1990s and beyond. Syst. Dyn.
Rev. 9(2), 113–133 (1993)
11. Thomas, D.R.: A general inductive approach for analyzing qualitative evaluation data. Am.
J. Eval. 27, 237–246 (2006)
Robot Moves as Tangible Feedback
in a Mathematical Game at Primary
School

Sonia Mandin, Marina De Simone and Sophie Soury-Lavergne

Abstract We study how elementary school pupils give sense to the moves of a
mobile robot in a mathematical game. The game consists in choosing 3 numbers out
of 6, whose sum is a given target number. The robot moves on a game board have
been implemented to provide pupils with a tangible feedback about their answer.
We have studied strategies of pupils to solve the problem and their evolution. Our
methodology included interviews, aloud verbalization and video observations of 28
pupils in grade 1 and 2 while they are playing. The pursuit of a mastery goal
encourages a trial and error strategy for only some of the pupils. We conclude that
some aspects of the moves of the robot, like its position, are perceived as a form of
help and not as a threat, even if they are only partially understood.

Keywords Robot ⋅ Tangible feedback ⋅ Learning ⋅ Solving strategy ⋅

Mathematics

1 Introduction

The OCINAEE project1—Connected Objects and Digital Interfaces for Learning at

Primary School—aims to explore and design mathematics learning situations with a
system of connected objects. The system is a set of interacting devices either
concrete like cards or a game board, or digital, like a smartphone. Connection

1
Funded by the French Bank for Public Investments, it is a partnership between two companies,
digiSchool and Awabot, and two public institutions, Erasme and the French Institute of Education.

S. Mandin (✉) ⋅ M. De Simone ⋅ S. Soury-Lavergne

Eductice/S2HEP, IFE, ENS-Lyon, Lyon, France
e-mail: [email protected]
M. De Simone
e-mail: [email protected]
S. Soury-Lavergne
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 245

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_22
246 S. Mandin et al.

between the two classes of objects is implemented by a mobile robot that can read
some tangible objects such as playing cards or any specific printed material. In this
project, we develop several games, which address mathematics for pupils from
grade 1 to 6.
Our research deals with the interface of such a complex system of tangible and
digital objects including a robot: how are users taking action on the system? How
does the system provide feedback to the user? What should the designers choose?
How do users understand the feedback? We study specifically actions and feedback
of OCINAEE system in the tangible world, which are different from actions and
feedback on a computer screen. Tangible objects like a set of cards and a robot
moving on a board may be means of action and feedback.
Our theoretical framework crosses cognitive psychology and mathematical
education, while mainly referring to objects that belong to computer sciences. After
a presentation of this theoretical framework, which has framed the design of our
games and the experiment we present here, we will describe the OCINAEE system
for one of the games and the study of how the implemented feedback is understood
by 28 French pupils in grade 1 and grade 2.

2 Theoretical Framework

2.1 Handling and Tangible Devices

The physical handling of concepts promotes the children’s learning including the
transfer of learning [1]. Many research works in education, psychology or specific
to mathematics education have shown that mathematical concepts are bodily
embedded and concepts are developed on tangible and physical manipulations [2,
3]. Concreteness is also used as a way to produce tangible feedback in the theory of
didactical situations in mathematics [4].
Also, new interfaces have to be figured out as “cognitive tools for promoting
thinking and adaptive learning, rather than only emphasizing technology for
interpersonal communication” [5] (p. 25). But tangible and connected objects allow
to link physical and virtual worlds. Therefore, in a learning perspective, we want to
study their possible uses and the actions, manipulations and feedback they create.
Tangible interfaces can be described as objects handled by users and used to
control a computer [6]. Using such interfaces in a learning situation, like a peda-
gogical game supports learning in different ways. Involvement of the pupils in the
learning process is improved. Africano et al. [7] and later Kubicki et al. [8] have
shown that tangibility through interactive tables allows to increase the active
handling time of the concepts and a simultaneous activity and collaboration among
pupils. Nevertheless, involvement is necessary but not enough to provoke learning.
Tangible feedback provides information about mathematical solving strategy of the
pupils. According to Brousseau’s theory of didactical situations [4], learning
Robot Moves as Tangible Feedback in a Mathematical Game … 247

situations are modeled by a game in which the possibility of different choices of

actions for the player and appropriate material feedback are conditions for learning
to occur. Moreover, pupils have to know by themselves if they have succeeded or
not in solving the problem, independently from an external evaluation. Actions and
feedback of the tangible world offer additional system of representations for
mathematical concepts. According to Vergnaud [9] then Balacheff [10], mathe-
matical concepts and knowledge are defined by a set of elements among which a
semiotic system of linguistics and symbolic representations. Balacheff “emphasis
the importance of the manipulation of representation systems” [10] (p. 8). Learning
develops when using and translating from one system to another, including tangible
experience. Sylla et al. [11] use cards with some chosen representations of concepts
in their Tangible interface for storytelling. They are images to combine in order to
create stories. Tangible cards support pupils in developing meaningful stories by
easily changing their sequence.
Tangible objects and concrete phenomena may be used to drive the virtual
system. Feedback concerns the reaction of the system toward the user. It is usually
virtual, on the screen of a computer. Now with a mobile robot, the virtual system
may drive the behavior of a concrete object and therefore produce tangible feed-
back. In the OCINAEE system, we have implemented a tangible feedback, which
consists in the moves and the positions of a robot on a game board. It belongs to the
concrete world and may translate some mathematical properties into a physical
behavior. We want to study how learners interpret and give sense, a mathematical
sense if any, to the tangible feedback of the robot.

2.2 Feedback

As anticipated above, we are interested in analyzing the feedback of the system on

pupils. In general, we will refer to the term feedback as a return from action. We
will introduce studies concerning feedback in different research fields. In particular,
on the one hand, we consider feedback as “metacognitive” feedback and, on the
other hand, as “didactical” feedback more focused on the mathematical task.
In research about games in general, integrated feedback is studied by Salen and
Zimmerman [12]. For a meaningful game, users have to perceive feedback and
consequences of their choices and actions on the game. Moreover, to improve
players’ experiences, Sweetser and Wyeth [13] highlight the importance of
immediate feedback about the progression toward the goal of the game, for instance
by making the score always accessible.
From a cognitive point of view and in a learning context, feedback aims to allow
the learners to focus on the gap between their performance and the goals to reach
[14, 15] by providing information on the accuracy of the answer or clues to reach
the goals [16].
248 S. Mandin et al.

Noury et al. [17] link feedback and help, considering different types of feedback
as helps. They highlight the relations between self-fulfillment, metacognitive
judgment and nature of the helps used by learners. They distinguish between two
types of helps: the instrumental helps providing clues to learners and the executive
helps providing the answers. In their experiment, undergraduate students have used
both types of helps after an error feedback, especially when they were following a
mastery goal. However, learners use less instrumental help when they perceive it as
a threat to their need for autonomy. They also avoid the executive help when they
perceive it as a threat to their skills revealing their incompetence
(performance-avoidance goals). There is no observed effects of the students’ desire
to demonstrate their skills (performance-approach goal) on the use of the helps.
With OCINAEE, the moves of the robot can help pupils by showing the gap
between their answer and the expected one. Thus the robot moves could be qual-
ified as instrumental help. However, if this help is a necessary step to move on with
the game and if it is not available on demand as in the experiment of Noury et al.
[17] it can be perceived as a threat.
Rodet [18] distinguishes between cognitive feedback whose aim is the assess-
ment of a work (taking into account the final result or the learner’s approach),
metacognitive feedback on cognitive processes of learners (in order to encourage to a
personal reflection) and methodology feedback concerning appropriate strategies in
order to improve later uses. These types of feedback are related to the actions, the
cognitive processes and the knowledge of learners. Two types of feedback in Rodet’s
approach [18] are in line with a more didactical point of view. Didactical feedback is
a response of the system that has some meaning from the learners’ point of view in
relation to the knowledge at stake [4]. Mackrell, Maschietto and Soury-Lavergne
[19] distinguish between evaluation feedback in response to the achievement of the
task and its success or failure and strategy feedback in response to a strategy of
resolution in order to support evolutions of the strategy, therefore learning. More-
over, they highlight direct manipulation feedback as any immediate environment
answer to the action of users. The two former types of feedback are built starting
from the last one. From a didactical point of view, the design of feedback does not
rely on knowledge about the learner as an individual, but rather on the components
of the tasks and the knowledge involved in the solving strategies.

3 Description of the Game

We describe the implementation of the target number game with the OCINAEE
system of devices, first by presenting the tangibles and virtual objects, then the
scenario of the game, finally the design of feedback.
The target number is a problem solving situation which consists in choosing
three numbers out of six, whose sum is a given target number. The target number is
automatically chosen by the system, displayed on a smartphone. The six possible
numbers are printed on six playing cards. The player has to choose three cards
Robot Moves as Tangible Feedback in a Mathematical Game … 249

Fig. 1 The target number game devices and an example of a 6 card set (the purple one). To the
right, submission of a card to the robot

among the six and to submit them to the system. A game is played in 6 untimed
rounds, by a single player or a group.
The kit includes the following devices (see Fig. 1): a robot with a smartphone, a
game board with a picture on which the robot moves, and 39 playing cards.
Technically, two sensors placed under the robot allow to scan any code on the
game board and on the cards (the code is not visible by eye). At any time, the
system knows the robot location and can determine its trajectory and positions. The
robot eyes light up in different colors. The cards are grouped in 6 sets of 6 cards.
Each card presents a number and each set of cards is associated with a color. When
users select a card and want to submit it to the system, they scan it under the robot
(see Fig. 1, right). Three other cards are available to “validate” the scanned cards, to
“cancel” the scan of all the cards and to “listen” to the target number. The picture of
the game board represents a landscape with five characters aligned on the skyline.
They materialized different final positions of the robot.
At the beginning of the game, once pupils have placed the robot on the game
board, the system displays a target number on the smartphone with a colored
background. This indicates the color of the set of cards to use (see Fig. 2). Pupils
have to select and scan three cards whose sum has to be equal to the target number.
The robot eyes light up each time a card is scanned: in white for a correct scan and
in red in case of a card of the wrong color or twice the same card. There is no
immediate feedback if the number of cards is not three.

Fig. 2 Numbers on each of the six sets of cards

250 S. Mandin et al.

Then, pupils have to submit the “validate” card to get an evaluation from the
system. Hence, the system computes the sum of the numbers on the chosen cards.
According to this sum, the robot moves to one of the different characters on the
game board. The robot final position is: (i) on the marmot when the sum is lower
and far from the target number, (ii) on the sheep when it is lower but close to the
target number, (iii) on the shed when it is equal to the target number (success),
(iv) on the snowman when it is above and close to the target and (v) on the yeti
when the sum is above and far from the target. The characters are aligned to mediate
the number line.
Additional feedback is implemented. The smartphone displays a congratulation
message or messages telling the pupils that their sum is too small or too big or that
they have scanned a wrong number of cards (different from 3). In case of errors,
pupils have two other attempts to try again. The robot finally performs a “dance”
when pupils have succeeded in all of the six rounds of the game.

4 Research Questions

We are going to study tangible feedback through its effect on pupils’ solving
strategies and through the way pupils transform it into a possible help. The tangible
feedback in OCINAEE system is constituted by a robot, which moves and reaches a
final position on the game board.
Our question concerns the perception and exploitation of this tangible feedback
by pupils. It has been designed as an evaluation feedback produced when requested.
Does the position of the robot inform pupils about the validity of their answer and
does this feedback have consequences on pupils’ solving strategies? In other words,
we wonder if a tangible feedback may be an evaluation and/or a strategy feedback.
Moreover, pupils can consider feedback as possible helps. But according to the
kinds of goal they are following, mastery and/or self-fulfillment, they may consider
instrumental help as a threat to autonomy, therefore ignoring the feedback. Con-
sequently, in our observations, we are going to identify pupils’ goal and their
perception of threat through the feedback.

5 Methodology: Participants and Experimental Setting

To answer the research questions above, we have conducted an exploratory study

with some of the pupils of the 35 teachers involved in the OCINAEE project.
The OCINAEE project concerns 39 classes from grade 1 to 6 in 14 schools and 4
junior high schools. Ongoing studies concerning all the classes will also drive us in
the improvement of the games and the robot moves as feedback. As we are
interested in mathematical skills of the curriculum of the early classes, we limit our
study to pupils in grades 1 and 2. The 28 pupils of the experiment come from 5
Robot Moves as Tangible Feedback in a Mathematical Game … 251

different classes, 8 pupils belong to 2 grade 1 classes and 20 pupils belong to 3

grade 2 classes. The observations were held during three weeks around Christmas
holidays. To minimize the effect of isolating one pupil at a time, we have observed
pairs of pupils, which indeed is the usual configuration chosen by teachers when
they use the game in their class. In each class, pupils have been randomly selected
two by two and extracted from the classroom for about one hour corresponding to
the duration of the different steps of the experiment. These pupils have already
played with other versions of OCINAEE games, under the supervision of their
teachers. They have also played the target number game, but without any moves of
the robot. These teaching sequences could not be controlled. However it has no
consequences on our experiment, because the pupils use the dynamic version of the
game for the first time.
The experimental setting consists in four steps: think-aloud training, interview
before playing framed by a short questionnaire, play, interview after playing framed
by a long questionnaire. The questionnaires have a total of 19 questions (6
open-ended and 13 close-ended questions with requests for explanations some-
times). Some questions aim only to open dialogue or allow pupils to recall the play
(e.g., Have you played to target number before? Did you see the robot move?).
Other questions propose some choices to help the children to answer (e.g., When
you were playing, you would: (1) like to show that you are doing well? (2) avoid
showing you can make errors (3) improve yourself in calculations?)
Think aloud was a mean to perceive the pupils’ strategies during the play. It has
been used in many areas, like to identify processes involved in the understanding of
a text [20], the writing process [21] or mathematical problems solving [22]. It is not
an easy task for young pupils. To make them more comfortable, we trained them
with 2 exercises that we assumed to be familiar. The first is an individual exercise of
continuation of a necklace in accordance with a color code. The second is a game of
noughts and crosses. It is a strategic exercise, which requires to alternate the
speaking times. Throughout training, when pupils were not verbalizing aloud their
thoughts, the researcher was fostering it with questions (e.g., what are you doing?
Where do you put your cross? Why?). The training lasted approximately 15 min.
The first interview aims to catch how pupils understand the game board and the
dynamic version of the game when they discover it for the first time. In particular,
we have asked the pupils to comment on the usefulness of the characters, including
the shed, on the game board. Just after the first interview and before the play, a short
explanation of the breakpoints has been provided to the pupils by the researcher.
At the beginning of the game, pupils have selected the range of the target
numbers (number target up to 20, 40 or 100). However the difficulty of the game
doesn’t lie in the size of the target number but in some features of the solution, like
the possibility or not to find two cards whose sum is a multiple ten or the presence
of distractive cards. During the play, we have videotaped the pupils. In the analysis,
we focus on the handling of the playing cards and their thinking aloud. Despite the
think aloud training, we have anticipated two risks. A first risk is the simultaneous
recording of two pupils’ voices and therefore difficulty to understand their talk
afterwards. A second risk is that despite training, thinking aloud remains difficult
252 S. Mandin et al.

for young pupils. We therefore carried out a combined analysis focusing on both
their speech and their actions. All data have been manually analyzed according to
our theoretical framework.
The after game interview is more complete. It allows to compare the pupils’
declaration to their actual achievement during play. We analyze the self-fulfillment
of the pupils (mastery vs performance goals) and their representation and
exploitation of the robot moves on the game board. The after game interview was
administered to 25 of the 28 pupils observed (it was not possible to run the
interview with 3 pupils).

6 Results

The results focus on the interpretation of the game board by the pupils, their
involvement during the game and the evolution of their solving strategies.

6.1 Interpretation of the Game Board and the Moves

of the Robot

The answers of the pupils to the first interview show that the positions represented
by the different characters on the game board have a meaning within the task only
for grade 2 pupils.
In grade 1, 2 out of 8 pupils separate the game board in two opposite team sides
and identify the different spots as symbolizing a failure or a success for each team.
The other pupils propose purely fictional explanations. 8 out of 20 grade 2 pupils
interpret these characters as positions of the robot, meaning that the sum of the
numbers is too big or too small with respect to the target number. Among them, 4
pupils perceive a shorter and more rarely a longer distance to target number when
the robot stops on the Yeti, and 3 pupils correctly interpret the positions represented
by the characters. One grade 2 pupil reacts as grade 1 pupils above-mentioned, the
others do not give any explanation.
The most frequent explanation of the aligned positions of the characters is the
facilitation of the robot move (6 pupils out of 28). This justification highlights the
focus of the pupils on the robot. But 4 pupils in grade 2 mention that characters are
sorted in ascending order according to the sum of the chosen numbers. This can be
analyzed as an initial mathematical interpretation going toward the concept of
number line. The others were undecided or gave off topic answers.
After the play, pupils still have misconceptions of the game board, with a better
interpretation in grade 2. The absence of materialization of the starting area led to a
confusion with the first position of the robot represented by the marmot (6 pupils
out of 25). The shed is the best understood character (19 pupils out of 25) compared
to the other four spots (correctly interpreted by 13 to 16 pupils out of 25).
Robot Moves as Tangible Feedback in a Mathematical Game … 253

10 Grade 1 -
Performance-
Approach Goal
Grade 1 -
Mastery Goal

5 Grade 1 -
Performance-
Avoidance Goal
Grade 2 -
Performance-
Approach Goal
0 Grade 2 -
SomeƟmes rightly SomeƟmes wrongly Always rightly Always wrongly Mastery Goal

Fig. 3 Distribution of pupils according to their answer to the question of achievement (have you
found the answer at the first attempt always/sometimes or never?) in comparison with their
achievement observed in the game (rightly if the answer corresponds to the observation else
wrongly) and their goals of self-fulfillment

6.2 Involvement and Goals of Pupils During the Play

As shown in Fig. 3, interviews after playing reveal that 4 out of 6 grade 1 pupils try
to progress (approach performance goal) while 13 out of 19 grade 2 pupils try to
improve their skills in computing (mastery goal) and only 3 grade 2 pupils pursue a
performance-avoidance goal. Their main goal is avoiding errors. This may explain
why they wrongly answer the question about their success at first attempt (one of
them wrongly read the target number and found a correct sum according to the
targeted number). Indeed, none of the grade 1 pupils were pursuing
performance-avoidance goal and they all answered correctly the same question.

6.3 “Trial and Error” or “Compute and Check” Strategies

Pupils explaining what they do in order to play mainly refer to their mental pro-
cesses: count, search, think (4 grade 1 and 16 grade 2 pupils). Scanning the cards is
also frequently mentioned (2 grade 1 and 5 grade 2 pupils). They rarely mentioned
the testing of a combination of cards (1 pupil only at each grade). These declara-
tions are consistent with the observed behaviors during the play. Videos allowed us
to count the number of pupils showing a trial and error strategy. These pupils do not
anticipate the sum of their combination of cards but first submit the cards to the
robot to get its feedback. We call them “testers”. We oppose this strategy to a
compute and check strategy. In this case, pupils add the numbers on the cards
before scanning them, they are “checkers”. There are more “testers” than we pre-
supposed according to the pupils’ answers in the interview (7 pupils out of 25) but
they are much fewer than the “checkers” (16 pupils out of 25). Moreover, the “trial
and error’ strategy has been often triggered by the experimenter to help some pupils
facing difficulties and hesitations. The “testers” also seem to be mostly pupils
pursuing a mastery goal (5 “testers” out of 7) while the “checkers” are better
254 S. Mandin et al.

8
6
4
2
0
Yes No, Threat to her/his No, Threat to her/his No (without
skills need for autonomny comments)

Fig. 4 Distribution of pupils according to the perception of the moves of robot

distributed between the different goals of self-fulfillment (9 “checkers” pursue a

mastery goal out of 16).

6.4 Effects of the Robot Moves as Feedback on the Pupils’

Strategies

In case of successive trials, it is expected that the robot moves have consequences
on the evolution of strategies like the new selection of playing cards.
But, according to the interview after play, less than half of pupils speaks of the
robot moves (8 out of 19 pupils, see Fig. 4). Five of them mention the moves for
checking if their sum of numbers is too small or too large or right or wrong. The
other 3 pupils do not bring precision. Indeed, as we see below, pupils modify their
combinations of cards according to the overshoot of the target number rather than to
the distance between their sum and the target number. Among the 11 pupils who do
not claim to have used the robot, 7 pupils do not understand the meaning and
therefore the usefulness of the robot positions and moves. Only two pupils give a
reason for not using them while they correctly understand them. Both pupils claim
they want to succeed by themselves. Thus, the robot move is not generally per-
ceived as a threat neither on need for autonomy nor on skills.
The effect of robot feedback on strategies evolution is not direct. Among the 21
pupils who had another trial after errors (not due to a bug, a wrong number of
playing cards or scanning the same card twice), 10 pupils affirm they start again
from their previous combination by changing one or two cards sometimes or often
(3 grade 1 pupils and 7 grade 2). For them, the robot moves work as a strategy
feedback. The other pupils start again as with a new target number. However, in the
25 s or third attempts concerning 10 pairs of pupils, 18 combinations involve the
change of only one card. In 11 of these trials, the new card replaces the smallest
card of the previous combination (see Fig. 5). The number of cards that are
modified between two attempts seems very strongly linked to the sum, smaller or
bigger than the target number. If the sum exceeds the target number, pupils usually
change two playing cards (3 cases out of 7). With a sum smaller that the target
number, pupils change only one playing card (14 out of 18), usually the smallest
(10 out of 14).
Robot Moves as Tangible Feedback in a Mathematical Game … 255

10
Target Number less
8
6 than the erroneous
4 combinaƟon
2
0 Target Number
1 Play Card 1 Play Card 1 Play card 2 Play Cards No changes higher than the
(the lowest) (the highest) (in the erroneous
middle) combinaƟon

Fig. 5 Distribution of changes of cards according to the value of the previous combination

These observations show that these pupils seem guided by a strategy that leads
them to exceed the target number and then get closer by changing the smallest card
as if it would “go down” more gradually. The game as a whole allows to highlight
the strategies and foster their evolution. But, for the moment there is no evidence
that strategies evolve towards more efficient ones.

7 Conclusion

In this paper, we have described an experiment which aimed to highlight the

interpretation and exploitation of a robot moves on a game board as tangible
feedback by grade 1 and grade 2 pupils. The target number game is played with
cards that has to be scanned under a robot. According to the sum of the numbers
printed on the cards, the robot moves to a position, indicating the distance between
the sum and the target.
We have noted several differences between the pupils according to their grade.
First, the game board is designed so as to mediate the number line. However the
meaning of the picture and the aligned position of the characters are not understood
by all pupils, especially the pupils in grade 1. They do not make connections
between the game board and the mathematical concepts involved in the game.
Second, it concerns the goals of self-fulfillment. The grade 1 pupils want to
demonstrate their skills (performance-approach goal) while most of the grade 2
pupils want to improve them (mastery goal) and only some of them want to avoid
showing they may make errors (performance-avoidance goal). These different goals
of self-fulfillment appear to influence the pupil strategies in the way they use the
robot. Some of them test their combinations of cards without anticipation whereas
others scan their cards only once they are sure of their sum. Most of the “testers”
pursue a mastery goal as if they were aware that the robot may supports the
evolution of their strategies. Nevertheless, it remains interesting that the help
provided by the robot moves is not perceived as a threat. Therefore, we can assume
that most of the pupils have not ignored the tangible feedback of the system.
Finally, tangible feedback is an evaluation feedback for about three quarter of the
pupils. The others rely on the message on the smartphone instead of the robot
position to evaluate their answer. Also tangible feedback is a strategy feedback for
256 S. Mandin et al.

only about half of the pupils, the ones who modify their strategy according to the
robot moves. This is not surprising because many of them have not understood the
meaning of the robot positions except for the shed. Didactically, it is also interesting
to observe that pupils modify their invalid combination according to how it exceeds
the target number and not according to its distance to the target number. If the sum
exceeds the target number, only the lowest number is changed, otherwise two cards
are most often replaced. This strategy isn’t the most efficient one and asks for
further adaptations of the game and its feedback.
Actually, the evolution of pupils’ strategies could also be obtained by a per-
sonalization of the learning situation, based on a learner profile [23]. In our game,
for pupils using a strategy based on considering if the sum is just exceeding the
target number instead of considering the distance between the sum and the target
number, the system should provide new target numbers that make pupils aware of
the limitation of their previous strategy.
This study is interesting to frame research about how concrete objects and
phenomena produce tangible and immediate feedback and it has to be continued in
order to study long-term evolution of pupils’ strategies.

References

1. Martin, T., Schwartz, D.L.: Physically distributed learning: adapting and reinterpreting
physical environments in the development of fraction concepts. Cogn. Sci. 29(4), 587–625
(2005)
2. Lakoff, G., Nunez, R.: Where mathematics comes from: how the embodied mind brings
mathematics into being. Basic Books, New York (2000)
3. Edwards, L., Radford, L., Arzarello, F.: Gestures and multimodality in the construction of
mathematical meaning. Educ. Stud. Math. 70(2) (2009)
4. Brousseau, G.: Theory of didactical situations in mathematics. Springer, Netherlands (1997)
5. Oviatt, S.: Designing digital tools for thinking, adaptative learning and cognitive evolution.
CHI, Vancouver, Canada (2011)
6. Mellet-d’Huart, D., Michel, G.: Réalité virtuelle et apprentissage. In: Grandbastien, M., Labat,
J.-M. (eds.) Les environnements informatiques pour l’apprentissage humain. Traité IC2
Information Commande Communication. Hermes (2006)
7. Africano, D., Berg, S., Lindbergh, K., Lundholm, P., Nilbrink, F., Persson, A.: Designing
Tangible Interface for Children’s Collaboration. CHI, Vienna, Austria (2004)
8. Kubucki, S., Pasco, D., Arnaud, I.: Using a serious game with a tangible tabletop interface to
promote student engagement in a first grade classroom: a comparative evaluation study. Int.
J. Comput. Inf. Technol. 4(2), 381–389 (2015)
9. Vergnaud, G.: The theory of conceptual fields. Hum. Dev. 52, 83–94 (2009)
10. Balacheff, N.: cK¢, a model to reason on learners’ conceptions. In: Martinez M. V., Castro
Superfine A. (eds.) PME-NA Psychology of Mathematics Education, North America Chapter.
pp. 2–15. Chicago, IL, USA (2013)
11. Sylla, C., Branco, P., Coutinho, C., Coquet, E., Skaroupka, D.: TOK—a tangible interface for
story’s telling. CHI 2011, Vancouver, Canada (2011)
12. Salen, K., Zimmerman, E.: Rules of Play: Game Design Fundamentals. The MIT Press,
Cambridge, MA (2004)
Robot Moves as Tangible Feedback in a Mathematical Game … 257

13. Sweetser, P., Wyeth, P.: GameFlow: a model for evaluating player enjoyment in games.
Comput. Entertain. 3(3), 1–24 (2005)
14. Kluger, A.N., DeNisi, A.: The effects of feedback interventions on performance: a historical
review, a meta-analysis, and a preliminary feedback intervention theory. Psychol. Bull. 119(2),
254–284 (1996)
15. Hattie, J., Timperley, H.: The power of feedback. Rev. Educ. Res. 77(1), 80–112 (2007)
16. Shute, V.J.: Focus on formative feedback. Rev. Educ. Res. 78(1), 153–189 (2000)
17. Noury, F., Huet, N., Escribe, C., Sakdavong, J.-C., Catteau, O.: Buts d’accomplissement de soi
et jugement métacognitif des aides en EIAH. Environnement Informatique pour
l’Apprentissage Humain (EIAH 2007), pp. 293–298. INRP, France (2007)
18. Rodet, J.: La rétroaction, support d’apprentissage? Revue du Conseil Québécois de la
Formation à Distance. 4(2), 45–46 (2000)
19. Mackrell, K., Maschietto, M., Soury-Lavergne, S.: The interaction between task design and
technology design in creating tasks with Cabri Elem. In: Margolinas, C. (ed.) ICMI Study 22
Task Design in Mathematics Education, pp. 81–90. Oxford, UK (2013)
20. Magliano, J.P., Millis, K.K.: Assessing reading skill with a think-aloud procedure and latent
semantic analysis. Cogn. Inst. 21(3), 251–283 (2003)
21. Hayes, J.R., Flower, L.S.: Identifying the organization of writing processes. In: Gregg, L.W.,
Steinberg, E.R. (eds.) Cognitive Processes in Writing, pp. 3–30. Erlbaum, Hillsdale (1980)
22. Rosenzweig, C., Krawec, J., Montague, M.: Metacognitive strategy use of eight-grade students
with and without learning disabilities during mathematical problem solving: a think-aloud
analysis. J. Learn. Disabil. 44(6), 508–520 (2011)
23. Mandin, S., Guin, N., Lefevre, M.: Modèle de personnalisation de l’apprentissage pour un
EIAH fondé sur un référentiel de compétences. EIAH’15, Agadir, Maroc (2015)
Personalizing Educational Game Play
with a Robot Partner

Mirjam de Haas, Iris Smeekens, Eunice Njeri, Pim Haselager,

Jan Buitelaar, Tino Lourens, Wouter Staal, Jeffrey Glennon
and Emilia Barakova

Abstract Personalization of educational and behavioral training to the develop-

mental stage of the individual child is common practice in educational and thera-
peutic settings. Research on robot-based education training is only just starting to
adopt this approach. We present a pilot study on a behavioral intervention design in
which Pivotal Response Training (PRT) elements are embedded into a game played
by a robot and a child. Seven game levels are designed to cover different levels of
communication skills that are targeted by PRT. The levels do not differ with respect
to the logical steps in the game that the children should take, only with respect to
the social competence that the child has. The behaviors displayed at each stage were
observed and analyzed using qualitative and quantitative methods. Our results
indicate that the more socially challenging a game is, the happier children are and
the more children engage with playing the game, although the game challenge
remains the same.

M. de Haas (✉)
Tilburg Center for Cognition and Communication,
University of Tilburg, Tilburg, The Netherlands
e-mail: [email protected]
M. de Haas ⋅ E. Njeri ⋅ E. Barakova
Department of Industrial Design, Eindhoven University of Technology,
Eindhoven, The Netherlands
e-mail: [email protected]
E. Barakova
e-mail: [email protected]
M. de Haas ⋅ P. Haselager
Department of Artificial Intelligence, Radboud University Nijmegen,
Nijmegen, The Netherlands
e-mail: [email protected]
I. Smeekens ⋅ P. Haselager ⋅ J. Buitelaar ⋅ J. Glennon
Donders Institute for Brain, Cognition and Behaviour,
Radboud University Nijmegen, Nijmegen, The Netherlands
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 259

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_23
260 M. de Haas et al.

Keywords Social robotics ⋅

Initiation of interaction with a robot ⋅
Games with
⋅
robots Personalizing educational games ⋅
Social competences for children with
ASD

1 Introduction

Increasingly more attention has been given to the use of social robots in education
and behavioral training for typically developed children and for individuals with
Autism Spectrum Disorders (ASD) [1–4]. The use of robots for ASD education and
care is one of the most promising use of robots, yet a tremendous amount of
theoretical and practical work is required to make it come true.
Previous studies show that children with ASD normally respond well to robots,
sometimes even prefer them to humans in similar settings [2]. This has been
attributed to the fact that social robots are easier to interact with; due to their simpler
facial and gestural features that do not overwhelm the child. In addition, it was
shown that the effectiveness of question asking training was equally high when
performed with a robot or a human trainer [5].
The robots are programmed with social behaviors designed to elicit fundamental
social skills such as eye contact, turn taking, and imitation among others. Children
with ASD possess a number of social deficits including: difficulties in recognizing
nonverbal communication cues such as body language, lack of attention, inability to
switch from focusing on one thing to another, difficulty in communicating and
reciprocating in conversations and a lack of social initiations [6]. Recent research
reports enhanced levels of social behaviors such as attention and engagement, [1, 4,
5, 7] enhanced social initiations skills [1], imitation [2, 8], and improved turn taking
behaviors. Although these shortcomings are typical for individuals with ASD, the
typically developing children will also benefit from a personalized training corre-
sponding to their level of language comprehension.
In therapeutic and educational settings, game sessions with robots are often used
and have been proven to be an effective way of training social skills with children

J. Buitelaar
e-mail: [email protected]
J. Glennon
e-mail: [email protected]
I. Smeekens ⋅ J. Buitelaar ⋅ W. Staal
Karakter Child and Adolescent Psychiatry, University Centre Nijmegen,
Nijmegen, The Netherlands
e-mail: [email protected]
T. Lourens
TiViPE, Helmond, The Netherlands
e-mail: [email protected]
Personalizing Educational Game Play with a Robot Partner 261

[5, 9, 10]. In the existing educational and training practices there has been little
attention for the personalization of the training to the developmental stage of the
individual child. We have adopted the Pivotal Response Training (PRT) framework
which differentiates the level of training of the child to the child’s level of com-
munication. This framework targets the children’s pivotal areas of development
instead of targeting individual behaviors. This results into improvements in other
pivotal areas that are not directly targeted. To make the training scenarios more
individualized to the child’s development stage, 7 levels of game scenarios have
been created that target different communication skills, varying in difficulty
(competence levels). These scenarios help the children to increase their individual
level of communication.
In the current pilot study we investigate the effects of robots on children who are
typically developing, but do display differences in the level of proactive behaviors
and speech development. With our pilot study one of the games designed to be
tested with children with ASD, a card game, is analyzed. We aim to improve the
overall design of the personalized robot behavior within the games, and specifically
the flow of the interaction within each level of the game. The influence of the
therapist who is present during the training sessions is also evaluated.

2 Experimental Design

The main aim of the current study is to find an effective way of integrating PRT
principles within a game scenario, to facilitate a playful robot-child interaction at the
communication level of the child. Diehl et al. [3] argued that an experimental setup,
where the teacher or therapist is in the same room, should be given more attention.
Moreover, Mehrabian [11] suggested that a teacher or therapist that is present at the
robot session can help the child develop a more detailed conversation, or give the
child feedback. However, these papers do not discuss the possible decrease of
learning efficiency caused by the interference of the teacher or a therapist at the
moments when the child and the robot are successfully engaged in interaction.

2.1 Study Setup

The robotic platform used for this study is the NAO humanoid robot, with 25
degrees of freedom, 58 cm in height. NAO has simplistic facial features, with only a
mouth and eyes, and its face resembles to the age of a child’s face. Some of the
robot behaviors that were used in this game include text to speech functions, hand
gestures, and NAO LEDs. We used the same experimental setting as described in
[1]: the robot was placed in front of the child and the teacher or a therapist using a
laptop was seated next to the robot. TiViPE, which is a graphical programming
environment was used to program the robot and for interaction during the sessions
262 M. de Haas et al.

[12]. A special interface was used for real-time interaction between robot and child
and was connected to the previously programmed interaction scenario. The pre-
programmed scenario consisted of a dynamical system of behaviors for complex
interactions, emotions and behaviors. This interface for a real-time interaction
through speech and simple movements was created as a part of this program
especially for this experiment.
During the experiment, the therapist was in the same room to control the robot’s
responses to the verbal cues of the child. The robot was used during a card game
(Kwartet) to prompt the children in initiating actions (asking for a card), to react to
questions and to give rewarding expressions afterwards.

2.2 Competence Levels

The children were divided in different social competence levels, based on how
many times the children asked the robot questions, asked for help or made a
statement and reacted on the cues of the robot during the introduction phase. The
social competence level corresponded with a higher level of the game. At a higher
level of the game the robot made the social interaction harder for the children. The
robot would ask for more complex initiations (i.e. social instead of functional
initiations). For example in the lowest level children only needed to ask for the
cards, but in a higher level they also needed to protest when the robot intentionally
performed a wrong move or tried to cheat.
Scenarios were developed for the different levels, incorporating three versions of
a card game, which provided the child a choice for a game, and encouraged child
initiative. These scenarios consisted of the flow of interaction between the robot and
the child. Within the game scenarios, the therapist incorporated different levels of
help (prompts) to the child that are provided by the robot. The therapist had the
chance to type in extra robot speech utterances using a text-to-speech interface
connected to the programmed robot scenario and these speech utterances could also
be accompanied by corresponding gestures.

2.3 Participants

Eighteen children participated in this experiment (aged 5 to 8, mean 7, 15 boys and

3 girls). Table 1 displays the levels of the game and the number of the children per
level. None of the children fitted in the level 1, 6 or 7. Two children were excluded
from the analysis, one because of technical difficulties in the video material and the
other child because (s)he did not understand the game and therefore required a lot of
help from the therapist.
Personalizing Educational Game Play with a Robot Partner 263

Table 1 The number of children assigned to a game level and description of the social
competence levels
Level No of children Description of the social competency level
1 0 Speaking with 2/3 words
2 5 Asking for object/activity + multiple cues
3 8 Asking for object/activity + asking for help + protesting
4 3 Asking for help + protesting + www (what/when/where)
question asking
5 2 www-question asking + protesting
6 0 www-question asking + asking through
7 0 Asking through + making statement + asking begin question

2.4 Observational Protocol

We designed a protocol for the observers in line with the objectives of this study.
The protocol was divided into four sections. In the first section, the observer was
required to note where the child looked (gaze direction) the observer could choose
between the following options: (1) the child was looking at the robot, (2) looking at
the experimenter or (3) shared their gaze towards a card with the robot. With these
options, we aimed to examine the relationship between the gaze behavior and the
level of pleasure or engagement during the game session.
In the second section, we asked the observer to rate the level of arousal of the
child during the game session, based on three levels from neutral (0) aroused (+) to
very aroused (++). The observer was required to look out for cues like clapping,
winning gestures, raising hands or screaming or jumping among others.
In the third section the observer was asked to rate the valence observed during
the interaction. Valence can either be negative or positive. Some of the cues that
indicate negative valence include that the child closes eyes, looks down, yawns,
expresses negative verbal responses, and is impatient. Positive valence cues are:
smiling, positive verbal responses, winking teasingly at the robot etc.
In the last section the observer was required to rate the perceived emotions
during the game, consisting of happy, angry, bored, confused, afraid, surprised or
another emotion, inspired by the categorical scheme of Eckman [13]. To gather
additional qualitative information and to verify the results we also asked for
evaluation of the arousal level and the emotional valence, although we are aware of
the consequences of mixing dimensional and categorical schemes for measuring
emotions.
The observers were students at the university campus, who were taking a class
on social robotics and as an initial exercise they were asked to rate videos. They
were provided with the guidelines for observation. The observers viewed the videos
on a computer and record their observations on the provided paper forms. The
observers were allowed to stop the video to record their observations or to view it
again if necessary. A total of 16 observers coded the videos and every child was
264 M. de Haas et al.

observed by two observers each. The reliability of the observers can be found in the
results section. Furthermore, we used the mean of all the scores to process the data.

2.5 Experimental Procedure

Every child played a card game with the robot. Introducing the robot to the children
prior to the study was added as a first step to the experiment as robots are very
unusual and exciting as playing buddies, but it also provided information to
determine the social competence level of the children. The whole procedure took
approximately 40 min for every child of which 20 min was the duration of the
actual card game. A digital camera was used to record all sessions. This camera was
placed in a way that the face and the action of the child could be observed at all
times. During the whole session the robot was present and was leading the playful
interactions.
The same therapist was present at each session to control and assist the robot, but
also to offer help when children asked for help. After the child finished his/her turn,
the therapist could choose the next behavior for the robot on a laptop. If the action of
the child was unexpected, the therapist could prompt the child to interact by typing
an additional robot utterance at the specially created interface. This interface was
connected to the overall scenario but also permits interactions and modifications of
robot behaviors to be made at every moment. The notebook on which this interface
was displayed was on the same table as the card game and the robot.
All sessions started with an introduction in which the children had to build a
tower with blocks. The blocks were provided by the robot. The performance of the
children in the introduction scenario resulted into one of the 7 levels of the game
scenarios. After the introduction the robot would explain the rules of the card game
to the child and ask the child for help placing his cards in the robot’s card holder.
The children were given the same card holder as the robot to put their cards in, so
they would not be distracted with holding the cards. The purpose of the game was
to get four cards of the same category and therefore the robot and a child would take
turns asking for a card. When the child asked the robot for a card, the robot would
point at that card so the child could grab it, which can be considered as a natural
reward. In the other case the robot would ask the child for a card and would also
ask whether the child wanted to place the card in an empty place in his card holder.

2.6 Data Collection

All sessions were videotaped. From each video, several segments of 20 s interac-
tions were extracted by the experimenter. These segments represent the interaction
moments when a verbal utterance was produces either by the robot or by the child
during the actual card game. There were two types of verbal interactions: the robot
Personalizing Educational Game Play with a Robot Partner 265

asked the child for a card and the child asked the robot for a card. These questions
were accompanied with nonverbal behaviors that the robot used during the sessions.
These behaviors can be described as cheering, pointing at card and turning its head
to add more expressiveness to its speaking. This resulted in around 12 videos per
child (depending on the course of the card game, some children had more inter-
action moments with the robot). The videos were watched in a randomized order by
the observers. The coding of the videos was done in random order since we wanted
the observers to remain uninfluenced by the context.

3 Results

The results section features first the qualitative analysis of the robot-child interac-
tion. Second, the data analysis involved determining the inter-observer agreement
(IOA). Further, we did a visual inspection of the data and calculated the mean of the
results of the two observers for each competence level. Pearson correlation was
used for determining whether the child was more confused when (s)he looked
towards the therapist instead of the robot [14]. For the analysis, the mean ratings of
the expressed emotions at the different levels of the children development were
compared. A Shapiro-Wilk test was performed to test for normality of the data [15].
The data was not normally distributed, therefore a Kruskal-Wallis test was per-
formed to check whether there was a difference over the levels [16]. For the levels
for which a significantly difference was found a Bonferroni corrected one-tailed
Mann-Whitney test was performed (significance border p <= 0.0125) [17].

3.1 Qualitative Analysis of the Robot Child Interaction

The behaviors that the robot used during the sessions can be described as speaking,
cheering, pointing at card and turning his head to add more expressiveness to its
speaking. Of the 16 observers 10 gave extra annotations about the robot’s behavior.
Most observers noted that more variation of the interaction would improve the
design. They repeatedly mentioned that the voice of the robot should match the
robot’s behavior, for example the voice did not match with the robot cheering. In
addition, the robot had no variation/intonation in its voice, because of the imper-
fections in the computer generated voice. In one of the levels the robot would make
a joke to provoke the children into protesting. The text-to speech engine is not able
to enact laughter well instead it pronounces “hà-hà-hà-hà”. These weaknesses were
causing confusion by the children.
During the session the robot would ask for a card. If the children gave the right
card the robot would confirm that fact, he would ask the child to place the card in an
empty place in the card holder and finally, the robot would cheer. Another
repeatedly noted observation was that the children already placed the card in the
266 M. de Haas et al.

Table 2 The reliability of Condition PABAK Condition PABAK

recording calculated with the
prevalence adjusted and Gaze to human 0.62 Angry 0.97
bias-adjusted kappa. All Gaze to robot 0.03 Bored 0.56
scores higher than 0.4 are Joint attention 0.28 Afraid 0.81
included in the analysis Valence 0.42 Confused 0.51
Arousal 0.45 Surprised 0.80
Happy 0.40 Other emotion 0.36

card holder, and the robot would still say that the children had to place the card in
the card holder. The cheer of the robot was at that moment redundant as the children
waited for the robot’s turn.

3.2 Reliability of Recording

The inter-observer agreement (IOA) was determined with the prevalence adjusted
and bias-adjusted kappa (PABAK, see [18]). Table 2 shows the PABAK for every
category (gaze, emotion dimensions and cognitive states). Agreement was defined
when both observers identified the same behavior as present or rated the child equal
on the scale. The agreement between observers was only moderate (Mean PABAK
was 0.52). The observers showed a higher agreement on the categorical emotions
than on the dimensional ratings (i.e. there was no agreement in the level of arousal,
or valence). For the analysis we only considered the categories with a score higher
than 0.4 (moderate agreement).
Furthermore, the mean of all scores was used to process the data. We did this
under the impression that when the two observers disagreed on the score for a
behavior the mean provides the shared view of the observers. For example when the
child was slightly positive, some observers rated this child neutral and others
positive. In that case the child had a score of 0 and 1, the mean score of these two
observers (0.5) will be accurate enough to describe that the child was slightly
positive. We still checked with a visual inspection of the data whether the observers
did not disagree completely, since rating in both directions can result in neutral. In
only 1 % of all the cases the observers disagreed completely.

3.3 Levels of Social Competence

It was expected that the dimensional measurements of emotion would be similar for
the children in all different levels since the levels were designed for the social
competence level of the children, while the presence of the different emotions can
differ. Figure 1 shows in percentages the times observers rated the children in a
certain emotion plotted versus the estimated level of social competence (which
Personalizing Educational Game Play with a Robot Partner 267

Fig. 1 The percentage of the 80

happy, bored, confused and 70
60

Percentage
surprised emotion per child Level 2
50
developmental level. The Level 3
40
observers were allowed to rate 30 Level 4
more than one emotion as the 20 Level 5
observed emotion in the child 10
behavior 0

d
y

d
pp

se

ise
re
Bo

fu
Ha

pr
n

r
Co

Su
Fig. 2 The average of the 10
perceived arousal in
8
children’s behavior at each
competence level
Arousal 6

0
1 2 3 4 5 6 7
Social competence level

corresponds to the level of the game). Figure 2 shows the average of the perceived
arousal in children’s behavior at each developmental level, which corresponds to
the level of the game. The absolute arousal is highest in game level five, and lowest
in game level four.

3.4 Gaze Direction

The observers were asked to record whether the child looked at the robot, the
experimenter, or looked at the same card as the robot. It was expected that children,
who were rated to look more at the experimenter, were also rated as more confused
or more surprised and children who were rated to look less at the experimenter were

Table 3 The correlation and corresponding p-values between the observed gaze towards
experimenter and observed confusion, surprised and happy. Confusion has a correlation of 0.358,
which is not high, but still significant
R p
Gaze * confusion 0.358* 0
Gaze * surprised 0.037 0.635
Gaze * happy −0.1 0.199
268 M. de Haas et al.

more often rated as happy in this game situation. As Table 3 shows, the confusion
of the children correlated with the moments when children also looked at the
experimenter.

4 Conclusion

In this paper we present the results of a pilot study that seeks to establish how
children react to a robot in different scenarios. Children with a higher level of social
competence were assigned to an interaction scenario with the robot in which the
social behaviors that accompanied the game were more complex. The children were
rated as most happy and with a high arousal in level three and five. In level two and
four the children were observed as the most bored. Mann and Whitney [17] stated
that the emotion boredom can be classified with a low arousal and valence and the
emotion happy can be classified with a high arousal and valence, which is con-
sistent with our results and thus verifies the reliability of the observations.
The results indicate that the more complex the child robot interaction on a level
of social competence is, the more engaged the children are. A possible alternative
explanation of this result is, that the levels two and four were not challenging
enough (not well enough designed) for the children or that they were misclassified.
However, if the levels were not challenging enough for the children it does not
explain why the children in game level three are observed differently. Game level
three expects more social interactions of the children than game level two, however
less than in game level four.
In level four and five the robot asked sometimes on purpose for the wrong card
in order to initiate protesting behavior. This resulted in the child looking at the
therapist with confusion. When the robot answered the child’s confusion with an
explanation that he asked for the wrong card, the child had more attention to the
robot afterwards. When the therapist answered that the robot asked the wrong card
and the robot also said this after the therapist, the child kept asking questions to the
therapist.
As Robins and Dautenhahn [19] suggested, the teacher or therapist in the room
should be an active part of the interaction between child and robot. In this study the
therapist was used as an assistant of the robot and this presence was needed for
responding to each unexpected initiation of the child. When the robot could not
grab his cards, the robot asked the therapist for help. In this pilot, a triadic inter-
action between teacher, robot and child is used. However, in a previous study of
Barakova et al. [1] a triadic interaction was used between the robot and two chil-
dren. The authors found that the children appreciated the personal attention of the
robot. Therefore, in the current experimental setting the therapist is an assistant of
the robot instead of part of the main interaction.
To further improve the design of the experiment the therapist should interfere
less with the child. It is possible to pre-program a few explanation behaviors for the
robot beforehand to execute during the game, since text-to speech explanation is
Personalizing Educational Game Play with a Robot Partner 269

slow and influences the interaction. More engagement between the robot and a child
results in improvement of the communication skills of the child [7]. A faster
response of the robot can result in higher engagement of the child with the robot
and therefore in more effective improvement of his/her communication skills.
Moreover, the robot used a computer-generated laughter, in a next experiment this
can be changed into a recorded (more natural) laugher.
Extensive research has been done towards research with children and robots in a
Wizard of Oz setting (see [15]) with the researcher being invisible for the child.
However, less research has been done with the experimenter in the same room,
which in some educational settings is preferable, for instance in therapy with
children with ASD. In this case the therapist stayed in the same room to observe
how the child was interacting with the robot and this will have an influence on the
robot-child interaction, mostly favoring the human-human interaction. The current
study is a preparation for such an experiment and the social competence levels are
designed for the children with ASD. However, the description of the levels of social
competence are general enough can be useful for enhancing the engagement of a
general education, for instance of language and social skills training.

References

1. Barakova, E.I., Bajracharya, P., Willemsen, M., Lourens, T., Huskens, B.: Long-term LEGO
therapy with humanoid robot for children with ASD. Expert Syst. 32(6), 698–709 (2015)
2. Kose-Bagci, H., Ferrari, E., Dautenhahn, K., Syrdal, D.S., Nehaniv, C.L.: Effects of
embodiment and gestures on social interaction in drumming games with a humanoid robot.
Adv. Robot. 23(14), 1951–1996 (2009)
3. Diehl, J.J., Crowell, C.R., Villano, M., Wier, K., Tang, K., Riek, L.D.: Clinical applications of
robots in autism spectrum disorder diagnosis and treatment. In: Comprehensive Guide to
Autism, pp. 411–422 (2014)
4. Gruarin, A., Westenberg, M.A., Barakova, E.I.: StepByStep: Design of an Interactive Pictorial
Activity Game for Teaching Generalization Skills to Children with Autism. In: Entertainment
Computing–ICEC 2013, pp. 87–92 (2013)
5. Huskens, B., Verschuur, R., Gillesen, J., Didden, R., Barakova, E.I.: Promoting
question-asking in school-aged children with autism spectrum disorders: Effectiveness of a
robot intervention compared to a human-trainer intervention. J. Dev. Neurorehabil. 16(5),
345–356 (2013)
6. Shapiro, S.S., Wilk, M.B.: An analysis of variance test for normality (complete samples).
Biometrika 52(3–4), 591–611 (1965)
7. Feil-Seifer, D., Mataric, M.: Robot-assisted therapy for children with autism spectrum
disorders. In: Proceedings of the 7th International Conference on Interaction Design and
Children, pp. 49–52 (2008)
8. Barakova, E.I., Gorbunov, R., Rauterberg, M.: Automatic interpretation of affective facial
expressions in the context of interpersonal interaction. IEEE Trans. Hum.-Mach. Syst. 45(4),
409–418 (2015)
9. Fujimoto, I., Matsumoto, T., Ravindra, P., Silva, S.D., Kobayashi, M., Higashi, M.:
Mimicking and evaluating human motion to improve the imitation skill of children with
autism through a robot. Int. J. Soc. Robot. 3, 349–357 (2011)
270 M. de Haas et al.

10. Vanderborght, B., Simut, R., Saldien, J., Pop, C., Rusu, A.S., Pintea, S., David, D.O.: Using
the social robot probo as a social story telling agent for children with ASD. Interact. Stud.
13(3), 348–372 (2012)
11. Mehrabian, A.: Basic Dimensions for a General Psychological Theory Implications for
Personality, Social, Environmental, and Developmental Studies (1980)
12. Lourens, T., Barakova, E.I.: User-Friendly Robot Environment for Creation of Social
Scenarios. In: Foundations on Natural and Artificial Computation, pp. 212–221 (2011)
13. Ekman, P.: An argument for basic emotions. Cogn. Emot. 6(3–4), 169–200 (1992)
14. Cohen, J.: Statistical Power Analysis for the Behavioral Sciences. Lawrence Erlbaum
Associates, Inc. (1977)
15. Riek, L.D.: Wizard of oz studies in HRI: a systematic review and new reporting guidelines.
J. Hum. Robot Interact. 1(1), 119–136 (2012)
16. Kruskal, W.H., Wallis, W.A.: Use of ranks in one-criterion variance analysis. J. Am. Stat.
Assoc. 47(260), 583–621 (1952)
17. Mann, H.B., Whitney, D.R.: On a test of whether one of two random variables is stochastically
larger than the other. Ann. Math. Stat. 50–60 (1947)
18. Byrt, T., Bishop, J., Carlin, J.B.: Bias, prevalence and kappa. J. Clin. Epidemiol. 46(5),
423–429 (1993)
19. Robins, B., Dautenhahn, K.: The role of the experimenter in HRI research-a case study
evaluation of children with autism interacting with a robotic toy. In: The 15th IEEE
International Symposium on Robot and Human Interactive Communication, ROMAN 2006,
pp. 646–651 (2006)
Robot as Tutee

Lena Pareto

Abstract This paper explores the possible advantages of substituting teachable

agents in a learning environment, with a humanoid robot as the non-human tutee.
Teachable agents are used as an extension to educational games in order to leverage
engagement, reflection and learning. The learning environment is engaging and
shown to be effective for learning and promote self-efficacy in experimental studies
in authentic classroom settings. Features beneficial for learning which are further
enhanced by a robot compared to an agent are identified. These include embodi-
ment of the robot; a social, empathic behaviour, better conversational abilities
which together provide a better role model of an ideal learner for the student to
identify with.

Keywords Robotics ⋅ Tutoring ⋅ Teachable agent ⋅ Robot tutee ⋅

Role-model learner

1 Introduction

Robots are starting to enter the classrooms, not as substitutes for human actors but
as teaching tools or teaching assistants [1, 2]. According to [2], robotics is mostly
used in computer science education, in domain-specific subjects such as geometry
or physics where movement and spatial cognition are involved, in foreign language
learning or as assistive tools for cognitive or social support, for example as sto-
rytelling assistant in pre-school [3]. Most applications are devoted to the engi-
neering field not to the wider scope as a general educational tool [1]. However, in
these classes students also learn general 21st century skills such as communication,
collaboration, creative thinking, and problem-solving skills [4, 5].
Robots can play various roles in education; from design material in engineering
to virtual companions in other learning situations. Our concept belongs to a learning

L. Pareto (✉)
University West, Media and Design, Trollhättan, Sweden
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 271

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_24
272 L. Pareto

situation where the human acts as a teacher/tutor to the virtual companion but where
the purpose of the virtual companion is to act as a productive counterpart for the
human learning [6]. There is a distinction between virtual companions that can be
taught and those that can learn: to be teachable it is enough to appear to learn from
the tutor, they do not need to ‘learn’ [7]. To simulate a tutee there are two objec-
tives: to model a behavior that appears to progress due to teaching activities and that
is understandable to the human tutor in order to monitor the teaching activity. In
contrast to [8] where a robot learns social behavior from a child by machine
learning, our robot tutee merely appears to learn and the focus is on features that
support the human learner in becoming more competent.
The aim of this paper is to propose the Student Tutor And Robot Tutee (START)
learning concept, based on a virtual tutee enhanced mathematical game, which is
shown effective for learning conceptual understanding and reasoning in authentic
classroom situations [9, 10]. There, teachable agents are used as an extension to
educational games in order to leverage engagement, reflection and learning. Here,
we explore robots as alternative to agents, and address the following question: Can
a robot tutee enhance learning experiences and increase learning effects compared
to a virtual tutee such as a teachable agent?
The paper describes features of a teachable agent game that are beneficial for
learning, motivated by learning theories and experimental classroom studies. An
analysis based on learning theories identifies features that can be further enhanced
by migrating the tutee to a humanoid robot compared to the virtual teachable agent.
The fields of robotics and artificial intelligence in education are thus combined [11].

2 Questions and Learning

Questioning is fundamental to learning and education, in particular for promoting

deep-level reasoning of complex material. Questioning is an ordered event of three
stages: an incentive of action; articulation; and a response action [12]. An ideal
learner is an active, self-motivated, creative, inquisitive person who asks and
searches for answers to deep questions [13]. Deep questions (such as why, why not,
how, what-if, what-if-not) target difficult scientific material, whereas shallow
questions (who, what, when, where) more often target simple facts.
There is substantial evidence that human tutoring is effective [14]. Tutees learn
from peer tutoring; tutors gain in understanding as well [15]. Typically, the tutor is
untrained and the knowledge gap between tutor and tutee can be minimal [15]. The
tutor benefits from preparing material and reviewing their own understanding, and
trying to respond to tutee questions or confusion if explanations are inadequate,
incomplete or contradictory. Such tutoring effects make grounds for the
learning-by-teaching paradigm and thus teachable agents [16] and virtual tutees in
general.
Robot as Tutee 273

3 The Learning Environment

We are currently using teachable agents as an extension to educational games in

order to leverage engagement, reflection and learning. Teachable agents are
computer-simulated peers that “learn” based on input from the student. Artificial
intelligence techniques guide the agent’s behaviour based on what it was taught and
students can revise their agents’ knowledge (and their own) based on such
behaviour.
In most teachable agent systems, the pedagogical model is that of direct
instruction. Students teach their agents by representing knowledge in a
pre-determined format such as a concept map. This presumes that the
student-as-teacher is able to articulate her knowledge in the required format. Such
an assumption may not be appropriate when targeting younger students and
domains such as causal reasoning and mathematical understanding, and we have
taken a different approach: students teach their agents by demonstrating their game
playing know-how and the agent acts “the ideal learner” and poses numerous of
deep, explanatory questions prompting the student to reflect on and explain her
playing behaviour. The purpose of the teachable agent dialogue is to challenge
student tutors to reflect on their actions; to support transfer of tacit game playing
skills to verbal language; and to provide a role-model of the ideal learner to imitate.
The teachable agent is designed as an inquisitive, active learner that takes the
initiative and challenges the student to explain and justify her actions and thinking.
Successful learners use self-explanation as a learning strategy [15]. Most students
do not spontaneously self-explain but they do so when prompted and can learn to
do it effectively. The level of difficulty in the agent’s questions follows the game
performance level of the human tutor to challenge the student without being
de-motivating.
The tutee learns from observing the tutors’ playing-behaviors and from the
tutor’s responses to questions. Good as well as less good game playing behavior is
considered. Similarly, having robots learn from positive as well as negative
examples in [8] created better behavior. The tutee has a representation of the
domain knowledge, which is hidden until the tutor demonstrates the corresponding
knowledge, and that way the tutee appears to learn. The tutee’s question asking
behavior follows a progression tree; if the tutor does not manage to answer ques-
tions at one level, it will not proceed with more advanced questions. This way, the
agent will not become more competent than the tutor.
A teachable robot, claimed to be the only one, is a care-receiving robot that
promotes language learning for young children [17]. It is taught by direct
instruction by moving the robot, telling the robot what to do, or by showing an
action. Our tutee is taught indirectly by observing the tutor act and asking questions
based on the tutor’s actions, under the pretext that the tutee wants to understand.
That way it plays the role of a curious, inquisitive learner, eager to understand. It is
an active role where the tutee and the human tutor interact in a dialogue around the
learning topic with the purpose to stimulate the student tutor to reflect on and
274 L. Pareto

explicate his/her actions. This process resembles scientific inquiry, since the tutee
asks deep why-questions related to the learning material and the reasoning becomes
explicit and visible. Hence, the learning process takes place in a social constructive
environment [18].
The learning environment has shown to yield significant learning gain for
playing students compared to controls, it engages children in advanced mathe-
matical thinking in early education, and young primary students can act as suc-
cessful tutors [10]. According to [1], few studies support their claim of effectiveness
with quantitative evidence. This idea combines the motivational power of games
with the reflective power of a virtual tutee asking thought-provoking, deep ques-
tions on the learning material during game play.

4 Features of Student as Tutor

The learning is based on a joint, engaging activity, where the tutor and tutee have a
task to perform together. It is the interaction between the tutor and the tutee that
further leverage the learning already designed in the activity. It must be a mean-
ingful activity with an explicit learning goal, and games are good candidates.
Assigning the role as tutor to students have the following features related to
learning.
The students are assigned the role as tutor, but since the virtual tutee asks
insightful questions and prompts plausible explanations they might not accept the
fiction. However, 92 % of the students claimed that they taught the agent, and not
vice versa [10]. Such fiction adoption is more likely to occur if the tutee behaves in a
way natural to the tutor [19]. By natural we mean that (1) the questions should be of
the form that the students could have asked themselves, (2) the timing of the question
should be reasonable, and 3) the tutee should not become too clever too soon.
To teach the agent was ranked as the most engaging activity, compared to
watching the agent play or play the game without the agent [10]. Still, the agent was
not visually present besides as a small face image on the screen, so the idea of the
teachable agent was enough to stimulate motivation. Collaborating with a huma-
noid robot compared to a simple image and the idea of an agent ought to enhance
motivation a great deal.
The learning environment improved students’ self-efficacy beliefs [9] compared
to students not playing the game. An explanation can be that the tutee was ignorant
from start and tutors who feel more capable exert more effort toward tutoring [15].
Also, novice peer tutors can feel anxiety about tutoring a human peer, but when the
tutee is a computer agent such responsibility is removed. The tutee introduces the
so-called protégé effect, i.e., an ego-protective buffer, since it is the tutee’s
knowledge that is in focus instead of the student’s [20].
Features beneficial for student learning include students acting in an expert role,
a role seldom used in education, but is a method lauded among learning theorists.
Taking on a role or identity is one of the most effective ways of learning to think in
Robot as Tutee 275

new ways and learn new subject matter [18]. The teaching activity is known to be
beneficial, and having to respond to the tutee’s reflective questions ought to
enhance learning, since question-driven explanatory reasoning appears to be the
primary factor that explains why one-to-one tutoring is one of the most effective
methods of learning a body of knowledge or a skill [16].
Finally, the tutor needs to evaluate and judge the tutee’s behaviour and per-
formance in order to teach, as well as negotiating and reasoning with the tutee who
ensures that the conversation remains around domain-relevant topics.

5 Features of Robot as Tutee

The virtual companion is assigned the role as tutee, which become a genuine
situation compared to when teachers asks questions. Teachers’ questioning is not
genuine [16]: they are not interested in the answer to learn, rather to judge the
student’s knowledge. The robot will be programmed to act as a learner, ignorant at
start and behave as if it learns by observing the tutor’s action and responses to
questions. A low competency pedagogical agent is more motivational than a
high-competency agent [21].
The embodiment and social behaviour of a robot makes the collaboration and the
dialogue more believable compared to the teachable agent. There is evidence from
neuroscience that the more human-like technology appears; the easier it is to accept
it having intelligent features [11], and human-like robots are most believable after
humans. The tutor-tutee dialogue is highly situational and interactional: the tutee
robot reacts in direct response to student tutor’s actions. More human-like actions
from a social robot ought to enhance motivation.
Features beneficial for student learning where a virtual companion actually beats
a human peer, concern directing attention to pre-defined learning issues and staying
on the topic. Moreover, such behaviour is accepted since virtual companions need
not be social. The virtual tutee acts according to its knowledge, which is a reflection
on the tutors observed and explicit knowledge. Hence it’s behaviour is related to the
interacting partner, and could be personalized according the student’s learning style
or preferences [3], or to learners’ special needs such as children with autism [22].
Also, this makes the tutee ask questions within the tutor’s zone of proximal
development [18], which cannot be assured with human collaboration.
The student interacts with the tutee and constructs knowledge through dialogue,
an approach argued for in [6]. The dialogue is essential, and can be controlled since
the robot is pre-programmed for the intended type of dialogue and topic. Hence, our
approach is similar to [6], where students can negotiate their ideas with a humanoid
robot and learn by means of socio-cognitive conflicts. Their study indicates that the
robot-child dialogue was more effective than the human-child counter part. Their
results are promising despite the small sample size and a novelty-effect of robots.
276 L. Pareto

The tutee’s thought-provoking questions encourage the tutor to self-explain, i.e.,

when a learner asks herself deep, explanatory questions and searches for answers.
A robot ought to act an ideal learner more convincingly than an abstract teachable
agent, and thus provide a better role model for the tutor to identify with.

6 Conclusion and Future Work

The START concept where a virtual companion is migrated from a teachable agent
to a robot tutee, is argued to further enhance the learning situation due to (1) the
embodiment of the robot; (2) a social, empathic behaviour (eye gaze, facial
expressions, gestures) possible to implement in the robot, (3) better conversational
abilities which all together provide a better role model of an ideal learner for the
student to identify with.
Future work includes setting up a Wizard-of-Oz experiment with the same
dialogue protocol as in the teachable agent-based learning environment, with a
social, humanoid robot.

References

1. Benitti, F.B.V.: Exploring the educational potential of robotics in schools: a systematic review.
Comput. Educ. 58(3), 978–988 (2012)
2. Mubin, O., Stevens, C.J., Shahid, S., Al Mahmud, A., Dong, JJ.: A review of the applicability
of robots in education. J. Tech. Educ. Learn. 1, 209-0015 (2013)
3. Fridin, M.: Storytelling by a kindergarten social assistive robot: a tool for constructive learning
in preschool education. Comput. Educ. 70, 53–64 (2014)
4. Eguchi, A.: Educational robotics for promoting 21st century skills. J. Autom. Mobile Robot.
Intell. Syst. 8(1), 5–11 (2014)
5. Alimisis, D.: Robotics in education & education in robotics: Shifting focus from technology to
pedagogy. In Proceedings of the 3rd International Conference on Robotics in Education,
pp. 7–14 (2012)
6. Mazzoni, E., Benvenuti, M.: A robot-partner for preschool children learning english using
socio-cognitive conflict. Educ. Tech. Soc. 18(4), 474–485 (2015)
7. Brophy, S., Biswas, G., Katzlberger, T., Bransford, J., Schwartz, D.: Teachable agents:
combining insights from learning theory and computer science. In: Lajoie, S.P., Vivet, M.
(eds.) Artificial Intelligence in Education, pp. 21–28. IOS Press, Amsterdam (1999)
8. Cakmak, M., DePalma, N., Arriaga, R.I., Thomaz, A.L.: Exploiting social partners in robot
learning. Auton. Robot. 29(3–4), 309–329 (2010)
9. Pareto, L., Arvemo, T., Dahl, Y., Haake, M., Gulz, A.: A Teachable-agent arithmetic game’s
effects on mathematics understanding, attitude and self-efficacy. In: Biswas, G., Bull, S., Kay,
J., Mitrovic, A. (eds.) Proceedings of the International Conference on Artificial Intelligence in
Education, pp. 247–255. Springer, Heidelberg, Germany (2011)
10. Pareto, L.: A teachable agent game engaging primary school children to learn arithmetic
concepts and reasoning. Int. J. Artif. Intell. Educ. 24(3), 251–283
11. Timms, M.J.: Letting artificial intelligence in education out of the box: educational cobots and
smart classrooms. Int. J. Artif. Intell. Educ. 1–12 (2016)
Robot as Tutee 277

12. Van der Meij, H.: Student questioning: a componential analysis. Learn. Individ. Differ. 6,
137–161 (1994)
13. Graesser, A.C., Person, N.K.: Question asking during tutoring. Am. Educ. Res. J. 31, 104–137
(1994)
14. Craig, S.D., Sullins, J., Witherspoon, A., Gholson, B.: Deep-level reasoning questions effect:
the role of dialog and deep-level reasoning questions during vicarious learning. Cogn. Instr. 24
(4), 563–589 (2006)
15. Roscoe, R.D., Chi, M.T.: Understanding tutor learning: knowledge-building and
knowledge-telling in peer tutors’ explanations and questions. Rev. Educ. Res. 77(4),
534–574 (2007)
16. Biswas, G., Katzlberger, T., Brandford, J., Schwartz D.L.: TAG-V.: extending intelligent
learning environments with teachable agents to enhance learning. In: J.D. Moore, Redfield, C.
L., Johnson, W.L. (eds.) Artificial Intelligence in Education, pp. 389–397 (2001)
17. Tanaka, F., Matsuzoe, S.: Children teach a care-receiving robot to promote their learning: field
experiments in a classroom for vocabulary learning. J. Human Robot Interact. 1(1) (2012)
18. Vygotsky, L.: Mind in Society: The Development of Higher Psychological Processes. Harvard
University Press, Cambridge, MA (1978)
19. Chan, T.-W., Chou, C.-Y.: Exploring the design of computer supports for reciprocal tutoring.
Int. J. Artif. Intell. Educ. 8, 1–29 (1997)
20. Chase, C., Chin, D.B., Oppezzo, M., Schwartz, D.L.: Teachable agents and the protégé effect:
increasing the effort towards learning. J. Sci. Educ. Technol. 18(4), 334–352 (2009)
21. Kim, Y., Baylor, A.L.: Pedagogical agents as learning companions: the role of agent
competency and type of interaction. Educ. Technol. Res. Dev. 54(3), 223–243 (2006)
22. Weiss, P.L., Cobb, S.V.G., Zancanaro, M.: Challenges in developing new technologies for
special needs education: a force-field analysis. In: 10th International Conference on Disability,
Virtual Reality and Associated Technologies, Sweden (2014)
Concept Inventories for Quality Assurance
of Study Programs in Robotics

Reinhard Gerndt and Jens Lüssem

Abstract Robotics is gaining an importance among the subjects to be studied in

scientific and engineering disciplines. However, being a quite new discipline with a
high degree of inter- and trans-disciplinary aspects, the teaching community cannot
rely on experience, gained by a long history of experiments. With this paper we
propose means to help assess and improve study programs in robotics. The
approach is based on the idea of concept inventories (CIs) and related tests to derive
objective information for a comparison of student groups with each other and over
time, e.g. to assess learning gain for specific measures. The approach helps to
further establish quality assurance in the design of robotics study programs and
equips teachers with measures for a formative assessment of their work.

Keywords Robotics education ⋅

Concept inventory ⋅
Teaching assessment ⋅
Learning assessment ⋅
Quality assurance ⋅
Accreditation

1 Introduction

Quality assurance as a dedicated task has been introduced to many universities with
the Bologna process, targeted to harmonizing the European Higher Education Area.
There are a number of organisations working on the definition of standards and
guidelines. As a representative for others, [1] requests “Institutions should have a
policy and associated procedures for the assurance of the quality and standards of
their programmes …”. However, the requirements typically are not broken down to
the operational level. Moreover, often, the quality assurance procedure is based on

R. Gerndt
Ostfalia University, Wolfenbuettel, Germany
e-mail: [email protected]
J. Lüssem (✉)
University of Applied Sciences Kiel, Kiel, Germany
e-mail: [email protected]

© Springer International Publishing Switzerland 2017 279

M. Merdan et al. (eds.), Robotics in Education, Advances in Intelligent
Systems and Computing 457, DOI 10.1007/978-3-319-42975-5_25
280 R. Gerndt and J. Lüssem

review by independent external agencies, either at the level of individual study

programs or at system level.
This introduces a subjective element to the process, which may result in a
unification of programs and hamper development of individual fields of expertise
and excellence. Whilst definition of standards is a commendable objective of these
activities, they often lack sufficient objective components, e.g. a formative
assessment of students and knowledge gain.
Currently, quality assurance often is biased by personal experience of the,
internal or external, auditors or members of advisory boards or by a given industrial
or research community involved. However, this only insufficiently allows for a
more agile reaction to new requirements by industry, service providers and
academia.
To overcome these limitations, new means of quality assurance for study pro-
grams should therefore include both, organisational processes and formative
methods.
The remaining part of this paper is organized as follows: After this introduction
we will shortly revisit the organisational measures for quality assurance in higher
education. Then we will add some notes on concept inventories as means of quality
assurance. Section 4 will specifically address the robotics concept inventory and
present experimental data on its application. Eventually, we sketch how organisa-
tional means and concept inventories can be combined for a long-term management
of study programs and present our conclusions.

2 Quality Assurance of Study Programs

Since the start of the Bologna process, higher education institutions have estab-
lished internal quality management systems for their study programs.
Quality assurance agencies had taken the role of external auditors. Initially,
quality assurance agencies audited study programs. Thus, the—internal and
external—evaluation of study programmes was in the centre of interest (see Fig. 1).
Since ten to fifteen years, quality assurance agencies have moved towards an
institutional audit approach to quality. Thus, higher education institutes have started
to build organisation-wide quality assurance systems.
For at least a decade, we see a growing quality assurance community within
higher education institutions. Networks are growing across Europe.
Today, a number of higher education institutions with mature quality manage-
ment systems have already moved away from programme accreditations to a so
called system accreditation. An example for a quality assurance system can be taken
from Fig. 2.
The evaluation of courses and the curricula seem to be still in the centre of
interest (see Fig. 2), but higher education institutions focused more and more on
their internal processes—in preparation for external audits.
Concept Inventories for Quality Assurance of Study Programs … 281

Fig. 1 Regular internal evaluation (from [2])

Fig. 2 Model for a quality assurance system (adapted from [3])

Thus, higher education institutions shifted away little by little from the evalu-
ation of the content of their study programs.
The reasons for this shift are manifold:
• Quality assurance agencies require sound quality management systems—the
content of a study program or a course is only a part of these management
systems;
• Higher education institutions focus on processes to meet these requirements;
• Higher education institutions use evaluation criteria that can be used on an
organisational level—this tends to exclude evaluation criteria related to the
content of a study program or a course
282 R. Gerndt and J. Lüssem

• Study program managers have not the right means to evaluate courses or
curricula.
We think that concept inventories can be very helpful to refocus institutions to
the content dimension.

3 Concept Inventories as a Means of Quality Assurance

Concept inventories have been developed in order to determine whether a student

has a working knowledge of a specific set of concepts. Concept inventories are
designed and evaluated to ensure test reliability and validity.
Therefore, in our opinion, concept inventories can be a great help for study
program managers to evaluate courses, the linkage of courses or even whole study
programs.
Figure 3 shows a crucial part of a (typical) conceptual scheme for curriculum
development and evaluation: Expected learning outcomes may differ considerably
from the real learning outcomes. Concept inventories may be a reliable instrument
to determine the real learning outcomes.

Fig. 3 Conceptual scheme

for curriculum development
and evaluation (adapted from
[4])
Concept Inventories for Quality Assurance of Study Programs … 283

4 Overview Over the Robotics Concept Inventory

Concept inventories intend to list the relevant concepts that are required to master a
specific scientific field. With every concept inventory there is a respective test that
allows assessing the understanding of students of the relevant concepts indepen-
dently of the actual knowledge. Students typically undergo the test twice, first as a
‘pre-test’ at the beginning of the course and second as a ‘post-test’ at the end.
Since tests do not change over some time or only develop slowly, results can be
used to assess the overall input level of students within their peer group. Applying
the test a second time, after attending a specific course, allows the assessment of the
concept learning gain.
A series of similar test methods has been developed, i.e. [5]. Because of the
concept learning gain we think that concept inventories can be integrated a bit
easier in our quality assurance system.
Figure 4 visualizes the gains for a number of teaching approaches in “Signals
and Systems” courses [6].
A number of CIs have been developed for different fields of studies (e.g. [7, 4])
one of the first CIs with a direct relation to our robotic CI is the force concept
inventory ([8]).
Our specific robotics CI has been presented in [9]. The relevant concept classes
—derived from textbooks, curricula, and course syllabi (e.g. [8, 10–13])—are listed
here:
• Math/Numerical Methods
• Mechanics
• Control Theory
• Stability
• Kinematics
• Dynamics

Fig. 4 Averaged gains of

different course types (from
[6])
284 R. Gerndt and J. Lüssem

• Sensing
• Perception
• Planning
• Navigation
• Decision-making
• Uncertainty
The partial robotics CI presented in [9] has been verified with pre- and post-test
at 50 % of the course time in a robotics course in the year 2015 and with a pre-test
only in 2016. Both courses took place at the same university, with master students
with comparable background. In 2015, almost 30 students participated in the
pre-test only, of whom 8 students also took part in the post-test, whilst up to the
time of writing this paper in 2016, 14 participated in the pre-test only. The numbers
are small and thus conclusions need to be drawn with care.

4.1 Evaluating the Learning Gain

The formula to evaluate the learning gain as visualized in (Fig. 4) is calculated by

the following formula:

post − pre
gain =
100 − pre

Results showed a value of pre = 40 % for the percentage of positive answers in

the pre-test and post = 53 % for the percentage of the positive answers in the
(early) post-test. Applied to the formula, the gain can be computed to:

53 − 40
gain = = 0.22
100 − 40

Comparing this learning gain to the results for the Signals and System CI
(Fig. 4), following information can be derived:
(1) A pre-test performance of roughly 40 % would make the student group
comparable to undergraduate students. However, the Robotics CI still lacks
sufficient verification and calibration, such that a classification with respect to
the entry level of students may be too early. However, there is some credibility
to the result, since many students came from the software domain, possibly
making them comparable to undergraduate Robotics students.
(2) A post-test performance of 53 % results in a gain of roughly 0.21, which
groups the course with low-gain courses. However, since the ‘post’-test was
taken at about half time of the course (after having finished the theoretical
part), further improvement was to be expected. This would have moved the
course at least to the border of low and medium gain courses.
Concept Inventories for Quality Assurance of Study Programs … 285

Fig. 5 Answering behaviour:

transformation question

Fig. 6 Answering behavior:

rigid body mechanisms

Significantly more interesting from a lecturer’s point of view was the individual
answering behaviour. This can be illustrated with two examples.
(1) The changes between individual answers for a specific question related with
transformations are shown in Fig. 5. The left column indicates the answers
given at the pre-test, whilst the right column indicates the answers given at the
post-test. In this case, answer b was the correct one. The arrows indicate how
answering behaviour of individual students changed and the respective
numbers. This example clearly shows convergence towards the correct
answer, which can be considered an outcome of the course.
(2) The changes between individual answers for a specific question related with
rigid body mechanics are shown in Fig. 6. In this case, answer “a” was the
correct one. This example shows how many student, after answering the
question correctly in the pre-test, got detracted and picked a wrong answer in
the post-test. This kind of ‘unlearning’ may be a necessary step for students to
overcome incorrect concepts, which may be the case here, with the test taken
at half time, or it may indicate an unfavourable teaching approach.

4.2 Comparing Student Groups

A single CI (pre-) test may also allow comparing student groups with respect to
their state when entering the course. The information may be helpful to focus course
work on relevant aspects. However, comparing results from different students with
comparable developments also allows calibration of the CI tests.
286 R. Gerndt and J. Lüssem

Table 1 Percentage of correct answers

# Question Correct answers 2015 (%) Correct answers 2016 (%)
1 Transformation 34 36
2 Time shift 52 71
3 Accelerationa 52 71
4 Small-signal/linearization 55 50
5 Mass-damper 21 7
6 Segway 86 79
7 Maze 28 7
8 M-Bot 52 50
a
Text of question was changed to avoid potential ambiguity

First we compared the percentage of students of the group who selected the
correct answer in the pre-test of the 2015 and 2016 student groups. The results are
shown in Table 1.
The comparison of results between 2015 and 2016 student answers shows quite
the same results for questions number 1, 4, 6 and 8. In summary, at the pre-test in
2015 correct answers were given with an average of roughly 47 % and a standard
deviation of 20 %. In 2016 the average for correct answers reached the same value,
however the standard deviation of 28 % was larger than in the year before.
Next we compared results with respect to the majority of the answers that were
assumed to be correct (Table 2).
This table shows quite the same results with respect to the remaining questions 2,
3, 5 and 7. In all cases a majority consistently considered a right, respectively
wrong answer to be correct one. However, qualitatively the latter ones show a
significant deviation in the range of 14 to 21 points on the percentage scale (which
relates to 2–3 students in the 2016 test). Noteworthy to mention is that question 2
(time shift), which shows a significant deviation of results, is taken from the mature
and well calibrated Signal and Systems concept inventory [6].

Table 2 Majority of answers

# Question Majority of answers 2015 Majority of answers 2016
(%) (%)
1 Transformation 34 (correct) 36 (correct)
2 Time shift 52 (correct) 71 (correct)
3 Accelerationa 52 (correct) 71 (correct)
4 Small-signal/linearization 55 (correct) 50 (correct)
5 Mass-damper 45 (wrong) 71 (wrong)
6 Segway 86 (correct) 79 (correct)
7 Maze 55 (wrong) 79 (wrong)
8 M-Bot 52 (correct) 50 (correct)
a
Text of question was changed to avoid potential ambiguity
Concept Inventories for Quality Assurance of Study Programs … 287

With respect to the number of correct answers of individual students in the

pre-test, the following numbers have been recorded: In 2015 the average number of
correct answers was 3.25 out of eight with a standard deviation of 1.3. Average
number for 2016 has been 3.7 with a standard deviation of 1.8. This indicates a
slightly higher level of expertise in the 2016 student group, however, it also indi-
cates a wider diversity compared with the 2015 student group. Results can be
compared to an average number of correct answers of the respective post-test of
4.25 with a standard deviation of 1.5. These numbers indicate an overall gain of
knowledge in the 2015 group through the lecture. However, numbers also indicate
that the spread between students increased. This is reflected by the stronger devi-
ation of correct answers for individual questions in the 2015 pre-test.
Aside from aggregated information from all questions, the individual answers
provide further information. We will provide some details on question 1 of the
Robotics CI (topic: transformation), as shown in Fig. 7.
With 34 % and 36 % choosing the correct answer in 2015 resp. 2016, 76 % and
74 % considered wrong answers to be correct (Table 3). Furthermore, the answers
initially are quite equally distributed among answers “a” to “c”, which shows no
pre-occupation about this concept. Consequently, learning a suitable concept can
possibly be achieved more easily.

Fig. 7 Question 1 of robotics

CI Question 1
Given coordinate systems C0 and C1 as in the
following image:

For transformation of the coordinates of a point P

from coordinate systems C1 to C0 we can find a
system of linear equations. Identify the suitable
equations to calculate x0 and y0 values of point P
with its coordinates given in x1 and y1 values.

a) x0 = x1 + 2, y0 = y1 + 3,
b) x0 = - y1 + 2, y0 = x1 + 3,
c) x0 = - y1 - 2, y0 = x1 - 3,
d) x0 = 2y1 -3, y0 = -3x1 + 2.
288 R. Gerndt and J. Lüssem

Table 3 Distribution of answers for question 1 (Transformation)

Answers (%) a b c d none
2015 31 34 21 3 10
2016 21 36 29 14 0
(numbers may not sum up to 100 due to rounding errors)

5 Management of a Study Program Using the Robotics

Concept Inventory

So far, we have used the robotics concept inventory mainly

• to adjust our teaching methodology in our robotics courses;
• to change the structure of our robotics courses;
• to rearrange and adjust the content of our robotics courses;
• to redefine the “interfaces” between two courses.
But we think that it is even possible to work with a range of concept inventories
in order to help evaluating a whole study program.

6 Conclusions and Outlook

In this paper we presented how definition of conception learning outcomes and

formal assessment can be combined with organizational procedures to assure
quality in higher education in the field of robotics. We provided positive indicators
on the viability of the concept inventory approach. Moreover, we provided infor-
mation on how the approach can be applied at different levels of quality assurance,
from assessing and managing the learning outcome of individual students, over
student groups to entire higher-education organizations.

References

1. European Association for Quality Assurance in Higher Education: Standards and Guidelines
for Quality Assurance in the European Higher Education Area, Helsinki (2009). http://www.
enqa.eu/pubs.lasso
2. Crosier, D., Purser, L., Smidt, H.: Trends V—Universities shaping the European higher
education area. European University Association, Brussels (2007)
3. Vroeijenstijn, T.: A journey to uplift quality assurance in the ASEAN universities. Report of
the AUNP (2006)
4. Ogunfunmi, T., Herman, G.L., Rahman, M.: On the use of concept inventories for circuit and
systems courses. IEEE Circuit Syst. Mag. Third Quarter (2014)
Concept Inventories for Quality Assurance of Study Programs … 289

5. Ahlgren, D. Verner, I.: 2006–2015: Robotics Olympiads: a new means to integrate theory and
practice in robotics. In: Annual Conference, American Association for Engineering Education,
Chicago (2006)
6. Wage, K.E., Buck, J.R., Wright, C.H.G., Welch, T.B.: The signal and systems concept
inventory. IEEE Trans. Educ. 48(3), 448–461 (2005)
7. Lindell, R.S., Peak, E., Foster, T.M.: Are they all created equal? A comparison of different
concept inventory development methodologies. PERC Proc. 883, 14–17 (2006)
8. Hestenes, D., Wells, M., Swackhamer, G.: Force Concept Inventory. The physics teacher
(1992)
9. Gerndt, R., Lüssem, J.: Towards a robotics concept inventory. In: 6th International Conference
on Robotics in Education. Yverdon-les-Bains, Switzerland (2015)
10. Featherstone, R.: Rigid Body Dynamics Algorithms. Springer (2008)
11. Kelly, A.: Mobile Robotics—Mathematics, Models and Methods. Cambridge University Press
(2013)
12. Thrun, S., Burgard, W., Fox, D.: Probabilistic Robotics. The MIT Press, 2005
13. http://ocw.mit.edu/courses/mechanical-engineering/2-12-introduction-to-robotics-fall-2005/
syllabus Accessed 12 Mar 2016
14. Totté, N., Huyghe, S. Verhagen, A.: Building the curriculum in higher education—a
conceptual framework. In: Enhancement and Innovation in Higher Education, Glasgow,
United Kingdom (2013)