Lecture 1: Introduction to Robotics
MAE 345/549
Anirudha Majumdar
Princeton
Sept. 6, 2022
Agenda for today
• Course logistics
• Motivation and introduction to course content
• Course plans, syllabus
2
A little bit about me
• Faculty in MAE and Associated Faculty in COS
• Visiting Researcher at Google Arti cial Intelligence Lab @ Princeton
• I spend ~1 day a week at the Google of ce
• My rst name is dif cult to pronounce!
• I usually go by “Ani”
3
fi
fi
fi
fi
A little bit about me
• Research interests: Robotics
• Controlling agile robots (e.g., UAVs, legged robots) with safety guarantees
irom-lab.princeton.edu
4
Course logistics: AIs
Sasha Eric An-Ya Allen Nate
Bodrova Lepowsky Olson Ren Simon
5
Course logistics: support
Jon Prevost (MAE)
6
Course prerequisites
• Multivariable calculus (e.g., MAT 201 or 203)
• Linear algebra (e.g., MAT 202 or 204)
• Basic probability (e.g., ORF 309)
• Basic differential equations (e.g., MAE 305)
• Some programming experience (e.g., COS 126)
• We will be using Python
• Sasha will do an intro to Python
(next week; not mandatory)
7
Course logistics: grading
• Problem Sets: 45% (assigned: Wed, due: Wed midnight)
• Mixture of theory, coding, and hardware implementation
• Midterm Exam (Take-home): 25%
• Early-November (see syllabus for tentative dates)
• Final Project: 30%
8
Course logistics: OHs
• Of ce hours: doodle poll
9
fi
Course logistics: policies
• Collaboration is permitted (encouraged) on problem sets. But, you must
turn in your own write-up, code, etc. (except for hardware portions).
• Write-ups should be clear/legible and explain all steps.
• Late policy: 10% deduction for every day of lateness (won’t accept if more
than week late).
• No collaboration on midterm exam (open notes, no internet).
• Regular attendance is strongly encouraged!
10
Course logistics: references
• No required textbooks
• References:
• Nourbaksh and Siegwart, “Introduction to Autonomous Mobile
Robots”
https://pdfs.semanticscholar.org/4a37/fe05d825ae2554be2c0c90a19a39fe51c26b.pdf
• Lavalle, “Planning Algorithms”
http://planning.cs.uiuc.edu/
• Thrun, Burgard, and Fox, “Probabilistic Robotics”
https://docs.ufpr.br/~danielsantos/ProbabilisticRobotics.pdf
11
Other logistics
• Gradescope (through Canvas)
• Ed Discussion (through Canvas)
12
MAE 345 vs. MAE 549
• MAE 549 is the “grad track”
• There are 12/12 students enrolled
• MAE 549 students will have some extra problems on
assignments (these will be clearly marked)
13
MAE 345/549: Introduction to Robotics
What is this course about?
14
What is a robot?
An embodied agent that can be programmed to perform physical tasks
Side note: asking this question is a good way to start a ght at a robotics conference
15
fi
What is a robot?
• Other reasonable de nitions:
• A machine — especially one programmable by a computer — capable
of carrying out a complex series of actions automatically [Wikipedia/
dictionary de nition]
• A reprogrammable, multifunctional manipulator designed to move
material, parts, tools or specialized devices through variable
programmed motions for the performance of a variety of tasks
[Robotics Industries Association]
• An autonomous machine capable of sensing its environment, carrying
out computations to make decisions, and performing actions in the
real world [https://robots.ieee.org/learn/]
16
fi
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
17
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
18
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
19
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
20
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
21
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• Is this a robot?
22
fi
What is a robot?
• Ultimately, all proposed de nitions have some issues
• See these articles for a discussion of issues with de nitions:
https://www.wired.com/story/what-is-a-robot/
https://robots.ieee.org/learn/
23
fi
fi
Robotics vs. Arti cial Intelligence
• This is something most
roboticists agree on
• A robot needs to be embodied
• Arti cial Intelligence (AI) need
not be embodied
24
fi
fi
Degrees of roboticity?
An embodied agent that can be programmed to perform physical tasks
• Lack of universally acceptable de nition hints at
some deep philosophical questions
• Could also be an indicator of the youth of the eld
• Probably need to measure degree of “roboticity”
• In terms of degree of embodiment, autonomy,
complexity, programmability, …
• But we don’t have formal de nitions for these
concepts
25
fi
fi
fi
Anatomy of a robotic system
Actuators
Sensors
Computation (Inertial Measurement Unit,
Optical Flow camera)
“Sense-Think-Act” 26
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
27
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
28
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
29
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920 Descartes: Animals as automata
• Unimate (George Devol, Joe
Engelberger): ~1950s
30
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
31
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
32
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe Rossum’s Universal Robots
This is where the word “robot"
Engelberger): ~1950s
comes from
In Czech: “work”, “labor"
33
A brief (pre-)history of robotics
• Greek mythology (Talos): ~1000BC
• Early automata: ~300BC - 100AD
• Leonardo da Vinci: ~1500s
• Descartes: ~1600s
• More complex automata: ~1700s
• Charles Babbage: ~1800s
• Karel Capek: 1920
• Unimate (George Devol, Joe
Engelberger): ~1950s
34
A brief history of robotics
• ~1960s - today: robotics has grown into its own academic discipline
• Huge amount of progress on theory, algorithms, and hardware
But still a LONG
way to go!
35
Why is robotics hard?
https://www.youtube.com/watch?v=g0TaYhjpOfo 36
Why is robotics hard?
https://www.youtube.com/watch?v=eTLMGjp-AIQ 37
Why is robotics hard?
https://www.youtube.com/watch?v=Ujk_5KD5G9o 38
Why is robotics hard?
39
Why is robotics hard?
• Uncertainty will be a recurring theme in the course
• Most challenges in robotics can be interpreted as the robot (or robot
designer) facing uncertainty about something
• Uncertainty in the dynamics of the robot
• Uncertainty in the dynamics of the world
• Uncertainty in the geometry of the world
• Uncertainty in sensor measurements
• Uncertainty in the user’s intent
• …
40
Why should you care about robotics?
• Important topic with massive interest from industry and government
41
Why should you care about robotics?
• Important topic with massive interest from industry and government
42
Why should you care about robotics?
• Important topic with massive interest from industry and government
43
Why should you care about robotics?
• Important economic and social consequences
44
Why should you care about robotics?
• Lots of really fascinating technical challenges (many of
which we’ll discuss)
• Beautiful connections with many elds
• AI, machine learning, control theory, computer vision,
optimization, information theory, theoretical computer
science, applied math, …
• Provides a lens on the really BIG questions
• What is intelligence? What makes us human? What is
consciousness? What is free will?
• It’s really cool!!
45
fi
How should we organize robotics?
• One option: by applications
• e.g., aerial robotics, medical robotics, humanoid robotics,
underwater robotics, etc.
• Another option: by concepts/techniques
• Allows us to understand “core” ideas
• Allows us to appreciate that many application domains
share very similar challenges
46
Course plan: theory/algorithms
• Dynamics & Feedback control (~Lectures 2-5)
• Equations of motion, feedback controllers, linear quadratic regulator
• Motion planning (~Lectures 6-9)
• Discrete planning, planning in continuous spaces, trajectory optimization
• State estimation, localization, and mapping (~Lectures 10-15)
• Bayes ltering, Kalman ltering, particle ltering
• Computer vision and Learning (~Lectures 16-22)
• Camera models, optical ow, neural networks
• Broader topics in robotics (~Lectures 23-24)
• Robotics and the law, ethics, policy, etc.
47
fi
fi
fl
fi
Course plan: hardware implementation
• Hardware implementation:
• Best way to appreciate challenges
• Nothing like seeing things work (or not) in practice!
• Will use Crazy ie 2.1 quadrotor to introduce concepts
• Open source platform, small, light-weight, ideal for testing
48
fl
Course plan: hardware implementation
• Goal: make the drone navigate autonomously
49
Course plan: theory/algorithms
• Dynamics & Feedback control: make drone hover
• Equations of motion, feedback controllers, linear quadratic regulator
• Motion planning: nd paths that will reach desired goal
• Discrete planning, planning in continuous spaces, trajectory optimization
• State estimation, localization, and mapping (won’t implement on hardware)
• Bayes ltering, Kalman ltering, particle ltering
• Computer vision and Learning: optical ow and obstacle detection
• Camera models, optical ow, neural networks
• Broader topics in robotics
• Robotics and the law, ethics, policy, etc.
50
fi
fi
fi
fl
fi
fl
Course plan: hardware implementation
• Final project: put pieces together to navigate autonomously
51
52
Course plan: hardware implementation
• Teams for hardware implementation portions of assignments
• Please form teams of 4 by next Wed [Sept. 14th] and email
course staff (will hand out one drone per team in class on 9/15)
• Can use Ed Discussion to facilitate team formation
• Might be a good idea to have students from multiple majors in a
team
53
Course plan: hardware implementation
• Logistics for hardware implementation
• We have set up netted space in G105 and ACEE 012
• General lab-safety training (if you haven’t completed this)
• https://ehs.princeton.edu/training
• Online and in-person components (through EHS)
• Needed to access lab
54
Course plan: comments
• Robotics borrows a lot of ideas and techniques from many
different elds: control theory, computer vision, machine
learning, optimization, …
• Tried to focus course on topics that are “quintessentially
robotics”, but can’t avoid other topics altogether (e.g.,
control, computer vision)
• May feel like “survey” at some points, but the goal is to
introduce you to major concepts in robotics
• Different parts may feel more or less challenging
depending on your major
55
fi
Course plan: comments
• Many topics we won’t cover or do justice to:
• Inverse kinematics, grasping and manipulation, hardware
design, algorithms for walking/running, …
• Robotics and X
• X = ethics
• X = the law
• X = the economy
• etc.
56
Course plan: comments
• Things will not be perfect
• Need help from you!
• Feedback, suggestions, patience :)
57
Course plan: theory/algorithms
• Dynamics & Feedback control (~Lectures 2-5)
• Equations of motion, feedback controllers, linear quadratic regulator
• Motion planning (~Lectures 6-9)
• Discrete planning, planning in continuous spaces, trajectory optimization
• State estimation, localization, and mapping (~Lectures 10-15)
• Bayes ltering, Kalman ltering, particle ltering
• Computer vision and Learning (~Lectures 16-22)
• Camera models, optical ow, neural networks
• Broader topics in robotics (~Lectures 23-24)
• Robotics and the law, ethics, policy, etc.
58
fi
fi
fl
fi
