Development of a Supervised Robotic Prototype for

Citrus Harvesting
Nasasira Jossy
Kampala International University
Umezuruike Chinechere
Bowen University
Afam Uzorka

Kampala International University

Yuma Erick
Kampala International University
Balyejusa Gusite
Kampala International University

Research Article

Keywords: Robot, harvesting, algorithms, DOF, prototype

Posted Date: May 13th, 2024

DOI: https://doi.org/10.21203/rs.3.rs-4277555/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License

Additional Declarations: No competing interests reported.

 Development of a Supervised Robotic Prototype for Citrus
Harvesting
Nasasira Jossy1, Umezuruike Chinechere2, Afam Uzorka1, Yuma Erick1, Balyejusa Gusite1

1
Kampala International University; City: Kampala, Zip Code: 256, Country: Uganda
2
Bowen University; City: Iwo, Zip Code: 234, Country: Nigeria
Corresponding Author: Afam Uzorka
Email: [email protected]

Abstract
Citrus harvesting has been done manually for the past Sixty-Two (62) years in Eastern Uganda,
where commercial citrus cultivation started in the 1960s. Farmers use employees to help in
harvesting so that fruits can be picked on schedule. At this point, the hired harvesters gather in
gardens and divide up the harvesting tasks among themselves. Some climb citrus trees to reach
the ripe citrus fruits on the branches, while others gather fruits on the ground, pack them into
sacks, and then transport the sacks of fruits to the required stores. Thereafter, a prototype was
developed using a design science approach, and a field-oriented algorithm (FOC) was used to set
power and torque during the programming of the actuators for speed regulations. After that, the
validation process of the robot was determined by using a stopwatch such that to calculate the
time it takes for the developed system to carry out the tasks successfully while in the field and
the results showed that it picked up six citruses (6) fruits in five seconds (5 sec) from the ground
per two (2) revolutions of the picking mechanism. Compared to hand harvesting, the prototype's
storage capacity findings showed that it could only hold forty (40) citrus fruits, which were filled
in 33.4 seconds, maximizing harvesting efficiency. And this was achieved by getting 6 citrus
fruits collected per two revolutions divided by the 40 which is the maximum storage capacity
that was determined after the robot loading for 6 rounds as shown below in the full load capacity
= (40/6 = 6.67kgs), then during the study, the time it takes to load to full capacity = 6.67 * 5 =
33.4 seconds. Therefore, this research is highly recommended to citrus farmers in Soroti such
that low labor, high quality, and less time are achieved by farmers during harvesting.

Keywords: Robot, harvesting, algorithms, DOF, prototype

1 Introduction
According to (A. Gasparetto, 2019), robots spanned from 1950s and they were basically
programmed machines without the abilities of controls during operation since they had no
communication between them and the controllers. With respect to the hardware, the first
generation of these robots had low-technology equipment, even servo controllers were not
present at that time. These robots were capable of performing important quite trivial such as
package loading or simple material handling operations. Even though some developments in
automation had been initiated like the "program-Mable" which was a device invented by Poland
and Roseland in 1938 and also a tele-operated operated "manipulator" which was also invented
in 1949 by Goertz (Das et al., 2023). However, the point of robotics was in 1954 when George
Devol designed a "programmable article transfer" which was the base for robotics development
and this was the first true industrial robot in history.
According to (Jingjuan Zhao, 2020) , Prof Noashi kondo of Kyoto University in 1984, developed
the robot which was used specifically in tomato fields, with this advent in robotics boosted
agriculture globally. Though most agriculture sectors where still using manual methods even
researchers were working hard to invent more robotics solution to problems faced by agriculture
sector.
Even though robotics development in agriculture started some good years back, but still there are
many gaps to be filled by different solutions and this is because of many reasons like being
expensive to some farmers, and also the methods of agriculture being practiced (greenhouse)
which has hindered the success of agribots in agriculture. Therefore, with this technology, it
paved away for the development of a citrus harvesting robot for citrus famers in Uganda mainly
in Soroti district.

1.1 Background
Citrus was first produced commercially in Uganda in the 1960s (Kongai et al., 2018). Following several
attempts, commercial citrus farming output started to pick up in Kiige, Ongino, Odina, Soroti, and
Lambori in Eastern Uganda with the introduction of various improved varieties planted under government
schemes. Through Operation Wealth Creation and a few non-governmental organizations (NGOs), as well
as the National Agricultural Advisory Services (NAADS), which was founded in 2001, commercial
production gained momentum and received strategic development and promotional support. Farmers
received training in the application of grafting and budding technologies thanks to this assistance. The
Gross Domestic Product of Uganda increased as a result of the application and adoption of these
technologies, which raised output.
Together with rising yields and GDP, there are observable difficulties. Globally, there is a need for
more food production because the global population is predicted to hit 9.6 billion people by 2050
(sheikh, 2022). The high chances of jobs in other sectors have reduced human labor availability
in the agricultural sector hence boosting the usage of Agribots, which has tried to replace the
human labor required for agricultural activities such that society can meet 2050 agricultural
needs. For the past six decades, robots have proven a fundamental role in improving efficiency
and costs as production is concerned in agriculture.

2 Related studies
According to (Jia et al., 2020), studies on citrus harvesting robots started in the United States of
America, where Mehta a robotic harvesting prototype was developed. This robot had a seven-
DOF, closed-loop control for timely feedback, and it consisted of cameras that were used at the
end effect or center. This robot was tested in greenhouses and its success rate was identified to be
95% and was appropriate for large and small fruits, but the researcher came to realize that this
robot had poor performance for small fruits. This led to the development of another robot that
was proposed by Hannan, this robot was capable of recognizing citrus fruits under canopy using
machine vision, and this was able to reorganize small and medium fruits. However, these
researchers said that there is still room to achieve orchard operations and even
commercialization.

Whereas (Liu et al., 2019), developed an autonomous visual system of citrus picking robots
using Convolutional Neural Networks, with this technology the system was able to recognize the
position of the fruits and also obstacles such as branches, fruits, and leaves of citrus on a citrus
tree. The system was also equipped with binocular cameras to estimate fruit depth in a 3-D
environment and also Region-based Convolutional Neural Network model (R-CNN) was used
for detection and the results indicated that the system was accurate at 85.12% in case of fruits
being covered by leaves and the system achieved 95.35% recognition of fruits in a natural open
environment. But the system did not have a mechanism for storing harvested citrus and a
transportation mechanism to the desired stores.

And confirming to (Didamony & Shal, 2020),a cabbage harvester robotic prototype was welded
and assembled at a private workshop at El Ibrahima, Al Sharkia Governorate, and Egypt. After
its assembling, it was connected to a tractor where disc cutters were observed to be cutting
cabbage heads from the main stems approximately 60-90 heads of cabbages in a minute, and
conveyer belts from the harvester to storage or pelleted box were observed transporting
harvested heads of cabbages to the pallet storage section. The operation of this prototype was
assessed based on two shapes of disc cutters (serrated and smooth disc), speed of the disc, and
angles of the disc cutter, where these cutters are in a position to cut the stem of the cabbage and
this was achieved at 88% at 1.5km/h speed of the disc. With this prototype, it was identified that
the robotic prototype did not cause major damages because 4% of damages were recorded and
for the missed cabbage heads, a scale balance model was put into use such that the speed of the
motor shaft and disc is measured and also the diameter of cabbage stems which gave an accuracy
of 0.01mm and this prototype proved to be time-saving, cost-effective, and not forgetting that it
was able to reduce labor force needed during cabbage harvesting because it was meant for
harvesting operations in Egypt. Although the robotic performance showed good results, the
robotic system was mounted on a tractor for it to operate, this system called for air pollution
since there was need to use fuel as source of power for the tractor to move the mounted robot.

According to (Alatyrev et al., 2020), the Chuvash state agricultural Academy in Russia designed
a multivariate cabbage harvester system that helped farmers to harvest cabbages in the field. The
robot had a cutting device that was in a position to cut the cabbage plants aligned longitudinal
and transverse as it is moving, after cutting the cabbage head, they could be directed to the cloth
of a conveyer belt at a joint with a climbing conveyor where a roller was used to load them into
the containers accompanying the robotic machine after separate cabbage leaves from cabbage
heads, and the leaves, unripe heads had to be off loaded at the end of the conveyer to the ground.
This study was achieved by the researcher who simulated the cabbage machine for harvesting by
stacking cabbage heads into storage containers on a low-frame semi-trailer enabled them to
quantify the relationship between the performance of the harvesting machine, structural, and
technical parameters of the variant studied and the study showed that this machine was excellent
because 92.6% of cabbage heads were properly harvested with two operators of the machine in
the field.

In consonance with (Fue et al., 2020), they developed a robotic cotton detection algorithm that
was using a stereo camera that was in a position to accurately discover and track cotton yields or
bolls using deep learning technology. The stereo camera was deployed to evaluate the bolls spot
and unearthed the mean error standard deviation which increased the speed of the robot and also
installed camera focus increased to 0.64 km/h. with this, the robot performed well with 9 mm
Root Mean Square Error (RMSE) and an average R2 value was 99% when stationary, but when
the robot started moving at approximately 0.64 km/h, there was a drop of R2 to 95%, and Root
Mean Square Error increased to 34 mm. This research was able to demonstrate the exact
detection and estimation of cotton yields in a garden condition in real-time based on an
embedded system, though those crops that were under shade were not detected because the
system lacked effective manipulators and sensing that help in detecting cotton bolls under shade.

In accordance, (Yusoff et al., 2020), developed an Integrated Loose Fruit Collector machine in
Malaysia which was operated by a single man whereby he or she drives the robotic machine
along the harvesting path looking for loose fruits and when the fruits are identified the machine
is stopped from moving and the driver goes down holds the suction nozzle which he points it to
the loose fruits of palm oil on the ground, and by doing this the loose fruits of palm oil were
sucked by vacuum and the collected loose fruits will be stored in a tipping container which
consists of a cleaner, and when the machine is fully loaded to 3 tons the operator drives it to the
bin location, and the process is repeated simultaneously until all loose fruits are harvested. With
this machine, productivity increased to 50% and provided clean loose fruits compared to the
manual loose fruits in palm oil gardens. Even though the system was developed, the researchers
recommended the integration of the Internet of Things (IoT) such that the Integrated Loose Fruit
Collector can be remotely monitored.

(Wan Ishak et al., 2010), developed an eggplant harvester for gantry robotic system that was
specifically for harvesting eggplants this robot was controlled by a programmable Logic System
which was the hub for switches and DC motors. Since these were connected to it, was easily
programmed using a Visual Basic program that helped to come up with its interface. After these
researchers employed a videogrammetry method, such that the distance between eggplants and
stems is determined such that the end effector is able to harvest the right eggplant. With this
method, the videogrammetry achieved a successful harvesting rate of 67.2% because the robotic
arm was able to lift up to the eggplant branches such that all eggplants to be harvested are easily
detected and then the peduncle of the eggplant is cut by the end effector, after the operator
checking on the interface, whether the eggplant fruit is ready to be harvest. With this method, it
was identified to be time-consuming though at a low rate because the operator was the one to
initiate the end effector to take action.

(Peng et al., 2021), came up with a strawberry harvest-aiding system with crop-transport co-
robots such that manual harvesting of fresh market products like strawberries and grapes are
mechanized which was the main challenge in the fruits industry. This robot was designed in
dynamic predictive scheduling that was mathematically modeled so that it is able to handle
uncertain requests. In addition, simulation was deployed to help in algorithm testing after the
overall system was deployed into commercial strawberry harvesting. where manual harvesting of
fresh market crops was identified to be time-consuming because harvester could walk while
carrying loaded trays to collecting stores at the end of the fields which was hectic for farmers,
but when evaluations were done co-robots improved the harvesting speed by 10% when
compared to manual harvesting bring the ratio to 1:3. In the interaction between pickers and the
system, the researchers suggested improvements such that the robot can be able to stop at the
exact point where the pickers are to avoid time wasting on walking to the robot because the robot
was programmed to stop 5 meters away from pickers.

(Jawad et al., 2020), developed a four-wheeled robotic prototype that was driven by four high
torque DC motors with a speed of 200-RPM, and it had a storage capacity of 30kgs.The robot
was equipped with a rechargeable battery, which could be recharged by a solar in case of a low
battery because its power rate was 115.25 w. This prototype was not only used to harvest
tomatoes but also had the capability to irrigate plants and maintain the ecological fields. For
harvesting and irrigating mechanisms, a sensor-based watering system to detect the humidity and
temperature such that the need for water by plants is confirmed and computer vision was
deployed such that the fruit picking and irrigation are achieved at the end of the day. The
harvesting mechanism of this robot by the use of a Convolution Neural Network (CNN) was
used to train their robot such that it is able to locate fruits was placed in front of the circuitry box,
and a robotic hand was also attached to the robot which was able to extend 30 inches such that
the gripper is able to grab a fruit and cut the peduncle of fruit with the help of the attached cutter.
This robotic was successful after testing its performance measurements were taken on pepper
fruits and it was able to detect the ready fruits to be harvested, but its storage unit was found not
heat resistant because of the materials used to develop the storage unit because out of 30kgs
stored 1kg of the fruits seemed to have been affected by heat produced by the sunrise while
operating.

Basing on (Mohamed, 2017), came up with a system for mango harvesting in Egypt, it was
powered by a rechargeable battery of 12v, and 3A was used to recharge the battery whenever
power could go low, this system also had a chassis that was made of aluminum material with a
thickness of 3mm, with a thread holder attached together with a telescopic picking which
consisted a switch of on / off and two bolts connected at the end of cutting discs with an outer
diameter of 100mm and inner diameter was 20mm. This system was equipped with a bucket that
had an elliptical shape at the entrance such that mangoes can easily be collected in the bucket
while the system is moving. Experiments indicated that the machine operator had to hold the
harvesting stick from many angles with both hands in order to reach the targeted fruit to be
harvested and this was achieved by pushing the button on the harvesting stick such that the disc
cutter can be able to cut the peduncle of a mango which seemed to be time-consuming during
harvesting, though the results showed that manual picking a hook type of a stick indicated that
6.85% fruits were injured while an innovative picker injury rate was at 2.43%.

Whereas (Arad et al., 2020), developed a robotic system for harvesting sweet pepper fruits. This
robot was composed of a standard 6 degree of freedom and it had a customized end-effector
design, which was mounted on the cart that could drive on a pipe rail and on a concrete floor by
the Arduino-base Programmable Logic Controller. This cart had the capability of elevating the
end effector to the harvesting level such that fruit is easily removed from a plant and held picked
fruit during transportation to the storage container with the help of deep learning technology and
a mounted camera, which helped the machine to detect the fruit. 24V rechargeable batteries
powered the completely robotic system. The average performance was 29% for the unmodified
crops, and for the modified crops, where farmers removed leaves that are covering fruits to be
harvested for easy identification, the performance hit 61% that proved a success for this
autonomous robotic harvester. Although this was achieved, the researchers said that reducing the
size of catching fingers at the end effector increased reachability and reduce the collisions with
crop leaves; they also suggested that the speed of the manipulator should be increased during
fruit transportation to the storage bin for a better output of the system.

Whereas in India (Jhala et al., 2018), they developed a battery-operated mango harvesting
machine to replace manual mango harvesting. This machine consisted of a 12V DC battery, a
high-speed DC motor, a camera of 2mp, cutting blades, a switch, and a display screen of 7
inches, which helped the operator to easily identify the fruit to harvest. With the help of the
mounted camera, the operator could just check on the display screen so that to check whether the
focused mango fruit by the camera is ready to be harvested and if the fruit is ready the operator
had to push the button to turn on the end effect or which consists of circular fitted blades
connected to the speed motor which cut the peduncle of fruit, then after the fruit is conveyed to
storage bucket without any damage. Results showed that in 10 hectares, the machine was able to
harvest 350 mango fruits per hour compared to manual harvesting where after harvesting 5 to 7
mango fruits in a bucket a person has to empty the bucket since the arms cannot withstand the
weight of the harvested fruits in a bucket, and on emptying it leads to time wastage. However,
we noticed that the success of this machine, the researcher left the untold solution of how the
battery being used by the machine can be recharged while the machine is operating.
(Yeshmukhametov et al., 2019), developed a continuum robot arm and gripper for harvesting
tomatoes which they named TakoBot for working in confined and extremely constrained spaces.
The robot had grippers designed in a semi-spherical shape for grasping spherical objects like
tomatoes which require gentle harvesting methods. The grippers had cutting blade edges which
made it easy for the gripper to separate the tomato fruit from a stem without damage. And the
researchers used machine learning to train neural networks such that the robot is able to
differentiate between non-ripe and ripe tomatoes and this was achieved by developing filters
such as a shape filter, and a color filter to discriminate tomatoes from other fruits that may also
be similar to the tomato. The robot was designed in a way that it can harvest tomatoes from
different angles and heights because the arm of this robot had a flexible backbone structure
which enabled it to rotate the end effect or at different angles. Results indicated that the proposed
manipulator achieved a manipulation of 200gms of objects, however, the speed was slow
compared to other robotic prototypes, but the noticed beneficial point of the robot was
reachability and tomato selectivity. The researcher suggested the use of machine learning to
control manipulators such as the harvesting time could be improved since the robot took more
time in tomato grasping.

2.1 Development of harvesting robotic prototype

Based on the spherical structure of citrus fruit forms, (Xu Xiao, 2022) designed three-finger end
effectors so that the grippers could readily carry out their grip-related activities. The purpose of,
this end effector was to harvest citrus fruits from citrus tree branches. This system's picking
accuracy at 250 rpm was 95.23%. But it was discovered that several of the gathered citrus fruits
had skin abrasions due to the effectiveness of their end effector design. This led to the design of
a robotic prototype controlled robotic system for citrus picking in the Soroti district which is
equipped with ahorizontal roller mechanism that picks citrus fruits from ground without
damaging them.

(Jiacheng Rong, 2022), developed an autonomous watermelon harvesting robot for use in
greenhouses. The robot's components included an industrial computer for overall system control,
a mobile robotic platform with six degrees of freedom (DOF) for motion planning and pose
control, and RGB-D Camera technology for end effector detection and placement to facilitate
harvesting watermelon fruits. The developed robot operated with an overall success rate of
93.3% and a positioning inaccuracy of 8.7% in the unconstrained circumstances of watermelon
fruits in a greenhouse. Despite the robot's excellent performance, it was found that it could only
move from one location to another. Although the robot had an excellent performance rate, it was
found that its moving mechanism was tied to the rails in the greenhouse, so it could only move
from one place to another in a straight line. This meant that it could not travel in all directions.
With the aid of wheels that are not attached to the rails, the researcher was able to create a
supervised robot for citrus harvesting with the help of this review.
Whereas, (Tianhu Liu, 2022), developed a multi flexible-fingured roller pineapple harvesting
robot for batch harvesting in China to harvest pineapples with a diameter of 90 to 120 mm with a
length of 110-150mm.The mechanism of this robot wasp preprogrammed using MATLAB
R2020a in the lab such that the position of the pineapple fruit and the harvesting device and also
the designed basing on the pineapples ellipsoidal shape and hard waxy peel. During operation,
the system was flexible in that it could bend pineapples stalks as it separates the fruit from the
stalk using two pairs of tilted flexible rubber fingers which work like the manual picking method.
The speed of the system was 100r/min in order to prevent damages caused to the fruits by
collision force during harvesting. The harvesting rate was at 85% because the harvesting speed
was about 1sec/per fruit. Even though the system achieved a great success, the inclination angle
between the two rollers couldn’t be adjusted during operation hence made the system fail to pick
some fruits that are under the determined height.

2.2 Validation of robotic systems

According to (Ahmet Yazıcı, 2021), there are numerous techniques to employ when validating a
robotic system, such as simulation-based testing and user validation in studies. Although the
aforementioned simulation-based testing is frequently employed, it is insufficient to account for
all potential field scenarios. Because they may be applied to all stages of the system development
life cycle, including validations, formal approaches based on mathematical principles can
therefore significantly aid in the construction of reliable systems. They suggested using the
widely open-source robot operating systems to integrate the validation techniques and guarantee
a thorough certification procedure. Thus, in the validation of the developed system, based on the
formal method to determine the efficiency of its performance while collecting citrus fruits from
ground.

2.3 Research gap

Conclusively (Chunmao Li, 2023), claims that using manual citrus harvesting methods instead of
robotics technology method is the main problem to be solved in citrus harvesting. Thus
suggested the need to maximize the use of robotics technology to improve the anti-breaking
picking of citrus fruits during harvesting. As a result, they developed a target identification and
sequencing system for citrus harvesting in their study using machine learning technology and the
detection success was at 92.41% in densely vegetated conditions. And system developments
were carried out, for instance an autonomous citrus picking robot by Liu 2019, an integrated
loose fruit collector machine developed by Yusoff 2020, a continuum robot arm and gripper for
harvesting tomatoes developed by Yeshmukhametov 2019, and Mehta robot which was
developed by Jia 2020.Based on what has been done by other researchers as highlighted above,
different mechanisms were deployed in the different systems studied and several gaps were
identified and in their studies Citrus fruit storage unit during orchard harvesting was not covered
by the researchers' systems, had no capability to transport the harvested citrus fruits to the
desired stores because some had no storage units, they solely focused on detecting citrus fruits
during harvesting.
Therefore, with the identified research gaps in the studied literature, We managed to base on the
identified research gaps in order to develop a supervised robotic prototype for citrus harvesting
which uses a horizontal roller mechanism to collect citrus fruits from the ground with the help of
a single operator and also the use of corrugated plastic sheet for designing the storage unit which
is heat resistant and this makes them fit for this research because citrus fruits can be affected by
heat and it has been identified that based on the related work done, as shown above, none has
used a horizontal roller mechanism to collect citrus fruits from ground.

2.4 Conceptual model of a supervised robot for citrus harvesting

Basing on the conceptual models of different studies on robotics development by different
researchers like Wan Ishak W.I (2010), Jiacheng Rong (2022), Amish Patel (2016), Wenkang
Chen (2020), (Hamido & Morgan, 2018), (Chen et al., 2020), Yuki Onishi (2019), (Jia et al,
2020), (Didamony & Shal, 2020), (Peng et al., 2021), (Jawad et al., 2020), (Arad et al., 2020),
,(Xu Xiao, 2022), and (Jiacheng Rong, 2022), who have contributed much in agribots
developments globally, a conceptual framework for this study was developed as shown below:

Figure 1: A conceptual model of a supervised robot for citrus harvesting (source: researcher’s
concept)

For the developed robotic system to operate smoothly, the controller communicates with the
driver hub through driver hub by wired means while the drive hub sends instructions from the
operators to the microcontroller (Control Hub) through a wireless medium for instructions
initiation. After that the, microcontroller checks whether the instructions require the motors
connected on to it or the expansion hub. After that the instruction is executed basing on the
component holding the required motor hence initiating movement or picking mechanism as
illustrated above.

3 Materials and methodology

According to(Aburamadan, 2019), design science is a research paradigm mainly focusing on
developing and validating prescriptive knowledge by the researcher and they said that it started
in the 1960s, with an interest in showing productive knowledge based on the artifact so that the
problem has been easily solved in a context by coming up with a prototype. With this approach,
the solution to the problem faced by citrus farmers has been developed by developing a
supervised robotic system that is in a position to harvest citrus in the Soroti district. This has
been achieved by assembling and programming the system that is composed of different
hardware components like a control hub, expansion hub, DC motors, cables, and extrusions so
that the robot is developed. In addition, since the main goal is to come up with a harvesting
robot, a design science approach was considered as my development methodology. The phase of
the design is shown in figure 2.

Initial
Problem
Explicate
Problem Outline
artefacts and
define
requirements Design
and
develop
Artefact
Demonstrate
Artefact Evaluated
Evaluate Artifact
Artefact

Figure 2: Phases of Design Science Methodology(Johannesson & Perjons, 2017)

3.1 Conceptual design for a supervised robot

The system of a supervised robotic prototype is comprised of different components which are
connected such that the system can perform the required tasks as required by the user. These
components include the control hub which is responsible for holding the program (Instructions)
that runs the robot and also transmits instruction signals from the drive hub to the specific motors
basing on the instruction from the operator. This process is passed through the expansion hub
which is accommodating a number of actuators that are responsible for task execution. All this
process is powered by the battery which is one of the main components that power the entire
system.

Figure 3: Conceptual design of a supervised robotic prototype for citrus harvesting (source:
prototype diagram)

3.2 Components used to develop the prototype

Requirement gathering is an exploratory process that involves researching and documenting the
project requirements from start to finish point and it enables the developer to come up with an
intended system at the end of the day. Therefore, as this research is concerned, the following
hardware requirements were used during the development of a supervised robot for citrus
harvesting in Soroti district.

Expansion Hub
During the development of a supervised robot for citrus harvesting in Soroti district, Expansion
Hub was used such that more hardware parts can be loaded like driving motors (core hex motors)
and servo-motor for easy communication with other devices. The expansion hub helped in
mounting a number of motors that were to be used during development.

Figure 4: Expansion Hub

Control hub
Since the control hub runs, an android operating system was found fit for this development of a
supervised robotic prototype for citrus harvesting fruits in Soroti district becausethe android
operating system can easily be updated. In addition, not only the control hub provides space for
holding instructionsit also bridges the communication between the operator and the system via
the Wi-Fi signal it emits for easy communication.

Figure 5: Control hub

Core Hex Motor

During the development of the robotic prototype, two core hex motors were used to run the
collecting mechanism for the developed supervised robot for citrus harvesting in Soroti district.
These motors rotate the rotating shaft, which is attached to these motors by chained gears. These
motors are programmed in a way that once initiated they rotate in a forward direction hence
rotating the citrus collector in a forward motion while collecting the targeted objects. In the
diagram below is the core hex motor used during this research development.

Figure 6: Core Hex Motor

HD Hex Motor
In the time of developing a supervised robot for citrus harvesting in Soroti, We realized
that,there was need to use powerful motors for locomotive activities such that the robot can
manage to move while carrying the collected citrus fruits meaning these HD Hex motor are able
to withstand a certain weight basing on their power hence making it fit for this research.
Figure 7: HD Hex Motor

Cables
Different cables have been used during the development of this robot, for example, red and black
cables were used to transmit power all through the robot and mult-colored cables were used for
instruction transfer from the control hub to different devices of the robot.

Figure 8: Cables

Extrusions
In the course of development, metallic extrusions were used to come up with the entire frame of
the intended robot. These have been used to because they are strong and have a unique future
that enables screws to be tightened easily during assembling.

Figure 9: Extrusions

A wired Controller
During the design, a wired controller was given a high priority to be used in controlling the robot
by the operator. In addition, during the development we were able to program the robot basing
on the keys we intended to carry the instruction such that once triggered a certain action is taken
for the smooth running of the robot.
Figure 10: wired Controller

Surgical Tubing
In the course of development, we also realized that there was need to use, a soft material for the
collection mechanism such the citrus quality is highly maintained. Therefore, we had to use the
surgical tubing’s, which are soft tubes that we interlink together for smooth collecting of citrus
fruits from the ground without causing any damage.

Figure 11: Surgical Tubing

Corrugated Plastic Sheet

Corrugated plastic sheets have been used to design the storage unit for the supervised robot for
citrus harvesting in Soroti district such that when the fruits are safely stored after picking.

Figure 12: Corrugated Plastic Sheet

Nut Driver
During the development of a supervised robot for citrus harvesting in soroti district, a driver nut
was used to tighten the structure of the robot after putting together all components during
development.

Figure 13: Nut Driver

Omni Wheels
The Omni wheels have been used in this development of a supervised robot citrus harvesting in
Soroti district because these wheels are good for turning the robot because they can move in all
directions.

Figure 14: Omni Wheels

Tooth Plastic Gears

Tooth plastic gears have been used during the development of the robot, to come up with a
horizontal fruit collecting mechanism, with this mechanism which is interlinked with surgical
tubing that are powered by core hex motors.

Figure 15: Tooth Plastic Gears

12V Slim Battery

A twelve-volt battery was used to power the developed robotic prototype citrus harvesting
system. The opted battery is rechargeable hence becomes more advatengeuos during operating
since it can be replaced with the battery that has charge as the other one is being charged.

Figure 16: Slim Battery

Driver Hub
The device is used to pass the instructions from the gamepad to the control hub for a robot to
take actions based on the operator's instruction, this is all achieved by the wires technology
embedded in the control hub, which communicates with the driver hub through a wireless signal
generated by the control hub.

Figure 17: Driver Hub

3.3 Validation of the robotic prototype developed

The supervised robotic prototype was validated to check whether it fulfills the user requirements
like collecting capability, and Wi-Fi coverage distance between the operators and the developed
robotic prototype. For the collecting mechanism of the robot during collecting mode, a stopwatch
was used to determine the time taken to collect the first round, which was recorded in seconds.
And for the Wi-Fi coverage distance between the operator and the prototype during operation, a
tape measure was used to determine the actual distance the robot is able to pick the instruction
signals from the operator at a distance as discussed in chapter four of this research, where the
distance was measured in inches then after converted into meters.
4 Results and Discussions
4.1 Robotic prototype design
The design was done using the unified
modeling language; the prototype operation
was diagrammatically represented as a design
block, Flowchart, and use case diagram. The
proposed robotic system prototype is
developed using design science methodology
and it will perform tasks based on the
communication between the Control Hub,
which holds a set of instructions that controls
all the peripherals connected to the expansion
hub as motors such that all requests or instructions from the operator are executed as required
while in the field. This is achieved when the battery powers the control hub and then the control
hub powers the expansion hub and peripherals the system will now be in a position of receiving
instructions from the operator through the driver hub. In addition, when the operator initiates the
instructions, which will be passed through the control hub up to motors for operation actions,
which answers the specific objective of designing a robotic prototype for citrus harvesting.

Figure 18: Design block for the proposed robotic system

Control hub
This is the brain and heart of the developed supervised robot because it is the one that bridges the
communication between all the components mounted on the robot. In other words, it is the one
responsible for transmitting the signal to the expansion hub for the tasks to be executed. More to
that it is the one that holds the programmed instructions basing on user requirement. The control
hub features an Android operating system gives it the flexibility to during operation, and also
with this operating system the updating processes are easily done even if the robot is in the field.
Expansion hub
This is a hardware component that communicates with any computer and the control hub. The
expansion hub can be loaded with several components like driving motors, servos, and sensors
and can communicate with other devices via several protocol options for this research it is using
a wifi protocol to communicate with the control hub andother components, it is at this point
where the expansion sends signals to the motor that handles the task basing on the instruction
signal that has been sent from the control hub.
Drive Hub
This is a compact mobile computing device that was designed for interfacing or passing
instructions from the operator to the control hub such that activities or tasks are performed. It
features the android operating system that is incorporated with dual-band WiFi for easy signal
transmission during operation, this capability attracted me to use it such that the prototype can
run smoothly during operation.
Battery
The battery used in this system is a 12v rechargeable battery, which powers the control hub, and
then the control hub powers the expansion hub for the smooth running of the system. During the
development of this prototype, several batteries were studied but this type of battery challenged
all batteries in the category of lithium batteries because it takes a short time to get full charging
when it is being charged. And It takes a long to get out of charge. This was not only the fact that
was considered during development but also the lightweight which makes it easy to be mounted
on the system.
Switch
This component is mounted between the battery and the control hub so that the power supply can
be controlled off and on easily during the operation of the robot.
HD Hex Motors
HD Hex motors were used in the development of this system for the movements of the robot.
These motors have a high-speed ratio compared to core hex motors that are is why in this system
they are used for locomotion of the robot from one place to another. These have two cables
running from the expansion hub one for power supply and another for signal receiving from the
expansion hub during task execution.
Core Hex Motors
This type of motor was used as the end effector of the robot since they have a low speed when
compared to HD Hex motors. They also have two cables running from the expansion hub one for
power supply and another for signal receiving from the expansion hub during the collecting of
citrus fruits from the ground.

4.2 Use-case diagram for the robotic prototype

Below figure 19 is the use case for the developed robotic system, where three actors are playing
different roles. The actor on the left hand is the operator who interacts with different activities
like starting the prototype, initiating the movement of the prototype, controlling the movement of
the robot around the citrus trees, and initiating the collecting mechanism of the developed robot
such that it collects citrus fruits, moving the robot to the store, releasing of the collected citrus
fruits to the required stores, and finally this operator interacts with the stopping action of the
robot. All these processes are performed with the help of other actors on the top and the own
right side of the use case, where the top right actor is a citrus tree that plays the role of producing
citrus fruits to be harvested, whereas the down-right actor is a storekeeper who is at the storage
center to keep the released citrus fruits safe.

Figure 19: Use-case diagram for the robotic prototype

4.3 Flow chart of the robotic system
A flow chart was used to diagrammatically represent the processes that are taken by the
developed prototype to achieve the objective of the study. As shown above the prototype starts,
then the movement mode is initiated. Then the system goes into the mode of decision-making,
where it checks whether there are citrus fruits to be picked. If NO, the process goes back to

movement initiation. However, If YES, the prototype collecting mechanism and collecting of
citrus are initiated. After that, it starts to load the collected citrus into the storage unit of the
system, and when it is full. The system is moved to the releasing area or storage point where it
releases all that has been collected to the storage area. Thereafter, the process for one collecting
round is stopped, until the operator decides whether to go for another collecting round following
the above processes in the flow chart.

Figure 20: Flow chart of the robotic system

4.4 Prototype development

The citrus robotic prototype was developed to harvest citrus fruits from the gardens, and in this
system,several different components were used for example extrusions, nuts screws, cables,
corrugated plastic sheets, plastic brackets, surgical tubbing, Omni wheels, chain, shaft collar,
shaft nuts, control hub, expansion hub, battery, HD Hex motor, Core Hex motor, and a servo
motor. These components were integrated such that the system is developed during assembling
as seen in the figure below. After assembling, the prototype was programmed using Java-
programming language such that all the required instructions are well set for a smooth operation
of the robot. All this development is based on a Field-Oriented Control Algorithm(Xiao Tang,
2023), which was used to control the amount of voltage needed by the motors to generate the
required torque for each motor used in the system.
Embracing the usage of the Field-Oriented Control Algorithm (FOC Algorithm), surgical
tubbing, and corrugated plastic sheets during the development of this prototype makes it different
from the existing ones because none of the above has been used in the studied literature of this
research. Existing ones have used convolution neural networks, computer vision, and machine
learning. And the use of metallic storage makes a big difference because these metals have the
capability of absorbing heat from sunlight which affects the quality of citrus during harvesting,
that’s why during the development of the supervised robotic prototype for citrus harvesting in
Soroti district, a major concentration was put on the usage of corrugated plastic sheets to design
the storage unit of this prototype because the corrugated plastic sheets are heat resistant meaning
they can’t affect the citrus quality. Another important component that creates a difference
between the developed robotic prototype and those that have been developed before is the use of
Omni wheels in this development. Omni wheels are wheels that can move in all directions that
are forward, backward, left, and right directions which make it easy for the prototype to turn fast,
compared to other wheels. These wheels have not been used in the studied literature of this
research because the existing developments used wheels that only move forward and backward.
Hence answering the objective of developing a robotic prototype for citrus harvesting.
Figure 20: Developed Prototype

4.5 System operation

As discussed in the requirement gathering of this study was to test its operational capabilities
since several components were integrated such that it can fulfill the intended requirements of the
users. The following are the three functions of the supervised citrus harvesting robot in Soroti
district, where it was tried in the field such that its operational capabilities are tested as shown
below.
Collecting capability
The developed robot was deployed in the field to test the collecting mechanism of the robot,
which was designed by interconnecting different components like sprockets, surgical tubes,
shaft, chain, and collar nuts. It was identified that the robotic prototype was able to collect six (6)
citrus fruits, in 5 seconds. This was achieved by setting up a stop-clock to count the time taken
for the robot to collect citrus fruits from the ground
.

Storage capability

Figure 21: Collecting capability and performance

The robot developed is able to store the harvested citrus fruits collected from the ground with
help of the human supervision. The operator of the robot after initiating the collecting
mechanism collects the citrus fruits, which are conveyed, into the robot storage unit by the help
of stretched surgical tubes. This unit has the capability of storing approximately forty (40) citrus
fruits. The robot was able to move around collecting other citrus fruits from the ground without
releasing the collect fruits that are already stored in its storage.
Figure 22: Storage capability

Releasing Unit
The developed robot was designed to collect, store, and release the stored citrus fruits in its
storage units to the desired places or stores based on the operator’s wish. It is able to move from
one place to another carrying the collected citrus. This implies that the quality of citrus fruits is
highly maintained because some fruits are damaged once packed in sacks as is analyzed in
chapter four of this research.
Figure 23: Releasing Unit
4.5.1 Capacity validation for the robotic prototype developed
Now the developed supervised robotic prototype for citrus harvesting in Soroti district can
collect six (6) citrus fruits, in 5 seconds with a storage capacity of forty (40) fruits per collection
load. This means the robot takes 33.5 seconds to collect forty citrus fruits as shown below in the
calculation.
6 citrus fruits per collection in 5 seconds
40 citrus fruits total load in the storage unit
6 citrus fruits make 1kg
Therefore,
40/6 = 6.67kgs
This implies that the robotic prototype is capable of storing 6.67kgs per total load.
4.5.2 Time taken for the robotic prototype to collect forty citrus fruits from the ground
The developed supervised robotic prototype for citrus harvesting is in position to store a
maximum of 40 citrus fruits in its storage unit where it collects 6 fruits per intake in 5 seconds.
This means for the robot to be filled up to its maximum storage it takes 33.4 seconds as shown in
the calculations below.
6 citrus fruits per collection in 5 seconds
40 citrus fruits total load in the storage unit
6 citrus fruits make 1kg
40/6 = 6.67kgs
6.67 * 5 = 33.4 seconds
The results indicate that a supervised robotic prototype can fill up its storage unit with 40
citrus fruits in 33.4 seconds.
4.5.3 Number of loads taken for the robotic prototype to collect a sack of 100kgs of citrus
fruits
A supervised robotic prototype developed is capable of collecting or picking one hundred
kilograms in less than nine minutes, which is more efficient when compared to manual methods
used by citrus farmers during harvesting seasons. Therefore, the robot can collect a hundred
kilograms that fill up a sack. It has to load 14.993 times for the sack to be filled up as shown in
the results below.
One sack = 100kgs
Robotic storage maximum capacity = 6.67kgs
Therefore,
100 / 6.67 = 14.993
This means that the robotic prototype to collect 100kgs that fill a sack it loads 14.993 times.

4.5.4 Time taken for the robotic prototype to collect a sack of 100kgs of citrus fruits
The developed robot for citrus harvesting is able to collect a hundred kilograms of citrus fruits in
8.4 minutes as shown in the analysis below and it also answers the specific objective of
validating the developed robotic prototype, where the capability of the prototype to collect 100kg
is determined.
Number of loads to make 100kgs = 14.993
One load takes 33.4 seconds.
14.993*33.4 = 502.2655 seconds
Converting 502.2655 seconds into minutes
1 minute = 60 seconds
Therefore,
Number ofloads/seconds in a minute
502.2655 / 60 = 8.4 Minutes
This means that the supervised robotic prototype is in a position to collect or pick one
hundred kilograms of citrus fruits in 8.4 minutes
4.5.5The distance between the operator controlling the robotic prototype
The robotic prototype developed uses Wi-Fi technology to access instructions from theoperator
through the Driver Hub device, which transmits instructions to the control hub in form of a
network signal. During the robotic prototype operation, it was identified that the robot was able
to receive instructions between the operator and the robotic prototype in a distance range of
55.27 meters. This was determined by measuring the covered distance at a point when the robot
signal strength started to lower where a tape measure was used to measure the distance in inches
then after the inches were converted into meters as shown in the calculation below, which
answers the specific objective of validating the developed robotic prototype.The measured
distance between the operator and the robot = 2176 inches1 meter = 39.37 inches
Therefore, 2176 / 39.37 = 55.27 meters.
4.6 Benefits of the developed robotic prototype for citrus harvesting in Soroti district
The developed robotic prototype for citrus harvesting is in a position to collect citrus fruits from
ground, stores them in its storage units which are made up of heat resistant, and then lastly it is
able to transport the stored citrus in ts storage to the required stores. All these activities are
performed by a single system which is controlled by a single operator.
To the side of citrus farmers, the system helps them to save money in hiring many human labors
to do fruit collecting, packing and transportation because it is just one person required to operator
the system that performs tasks of around twenty people in a period of less than an hour.
The unique collecting mechanism of the system, makes it outstand a number of robots that have
been developed in the studied literature of this work because it is made of components that have
not been used before and these during operation, they are set in a way that they only collect
circular objects that is why the developed prototype cannot collect stones when it is collecting
citrus fruits because the mechanism used only collects circular objects into its storage unit. In
addition, the collecting mechanism is designed using a component like surgical tubes which are
soft and cannot damage the fruits while being collected from ground.
The system developed for citrus harvesting does not pollute the environment with carbon since it
uses rechargeable batteries as a source of power to the developed system unlike in other
development where fuel is a source of power to the machines which end up polluting the
environment during fuel combustion.
4.7 Challenges faced by the developed robotic prototype for citrus harvesting while in the
field
The developed robotic prototype faced some challenges of navigating on grounds with big stones
and citrus trunks, which consumed sometime during navigating. And this was due to the size of
omni wheels used. That’s why we proposed the use of bigger omni wheels in future works of this
research.
Another challenge was on the battery changing. The prototype required changing the battery in
case the charge gets low which consumed some seconds of time to replace the battery with a
fully charged battery. That’s one of the reasons why solar technology is suggested in future work
of this research.
4.8 Compulsion of the developed robotic prototype and those developed based on the
literature reviewed
A developed supervised robot for citrus harvesting in Soroti district is made up of different
components assembled such that it performs the required functions. This robot is capable of
collecting citrus fruits during harvesting, Stores the collected fruits in its storage unit that is made
up of corrugated plastic sheets that are heat resistant meaning the collected citrus fruits are not
affected by the heat that is always generated by the sun-rise into the storage units of some
developed robots which use plastic storage units, and metals which are not heat resistant. The
robot also has the capability of releasing the stored citrus fruits to the right stores based on the
instructions given by the operators. with these functions, it is identified that the robot designed
performs one similar function to those developed in the reviewed literature and this function is
the collecting function because it was identified that the developed robot in the literature can
collect citrus, eggplants, cabbages, and also some developed robotic machines in the literature
can store the collected fruits into buckets where people keep on picking up filled buckets when
the robot is moving on a trail. But the developed robot challenges most of the developed robots
in the literature with its unique function of being operated by a single person, and the capability
to transport the collected citrus to different stores while using Wi-Fi signals from the driver hub
at a distance of around fifty meters where the robot can receive instructions signals from the
operator wirelessly and the use of a field oriented algorithm in the development of this robot
makes it greater than others because non used this algorithm in the studied developments in the
literature in this research.

5 Conclusion
The creation of a supervised robotic prototype for citrus harvesting, especially in the citrus-
growing region of Soroti, signifies a substantial advancement in Eastern Ugandan agricultural
techniques. For the previous sixty-two years, picking ripe citrus fruits by hand has been the
standard. Hired workers perform the physically taxing process of selecting the fruits from trees
and the ground. There are drawbacks to this conventional approach, including as personnel
expenses, schedule conflicts, and the possibility of inconsistent fruit quality.
A potential answer to these problems is provided by the unveiling of the robotic prototype. The
prototype showed impressive speed and economy by applying a design science approach and
integrating a field-oriented algorithm (FOC) for power and torque regulation. Impressive
outcomes were obtained by using a stopwatch to verify the system's capabilities. The robot was
able to select six citrus fruits from the ground in a mere five seconds for every two picking
mechanism revolutions.
Improving harvesting efficiency is one of the created prototype's noteworthy benefits. According
to the results of the storage capacity analysis, the robot could hold forty citrus fruits at its
maximum capacity in just 33.4 seconds. The meticulous computation of the robot's loading
capacity, which considered the rate of fruit collecting and the established maximum storage
capacity, is credited with this efficiency. As a result, a system that maximizes the amount of
fruits harvested while reducing the amount of time needed to attain maximum capacity is
created.
This study has important ramifications for citrus growers in Soroti and elsewhere. Farmers who
implement the robotic harvesting prototype can lower their reliance on manual labor, which will
save costs and boost profitability. Furthermore, a more uniform and well-managed harvesting
procedure results in fruit of a higher caliber, satisfying consumer and market demands. The
robot's time efficiency guarantees that harvesting operations may be finished on time, enabling
farmers to follow schedules and possibly expand their crop fields.
Essentially, this study strongly suggests that the robotic prototype be incorporated into Soroti
citrus farming operations. The potential advantages, such as lower labor costs, better fruit
quality, and time efficiency, make it an important tool for raising the region's citrus cultivation's
overall productivity and sustainability. The use of cutting-edge technologies, like the supervised
robotic citrus harvester, can help citrus growers in Eastern Uganda have a more productive and
profitable future as the agricultural sector develops.

Competing interests
None

Funding
None

Ethical Approval
Not applicable

Consent to Participate
Not applicable

Availability of data and material

Data is provided within the manuscript or supplementary information files

Author Contributions
Conceptualization, J.N.; methodology, J.N.; investigation, J.N.; writing—original draft
preparation, J.N.; writing—review and editing, J.N., A.U., C.U., E.Y., B.G.; supervision, C.U.;
visualization, J.N., A.U., C.U., E.Y., B.G.

References
A. Gasparetto, L. S. (2019). A Brief History of Industrial Robotics in the 20th Century. Advances
in Historical Studies.
Aburamadan, R. &. (2019). Applying design science approach to architectural design
development. Frontiers of Architectural Research.
Ahmet Yazıcı, M. O. ( 2021). Verification and validation methods for industrial robots. Valu3s .
Akkineni Navya Sri, D. R. (2022). Post-harvest handling to reduce loss of fruits and vegetables.
The Pharma Innovation Journal .
Alatyrev, S. S. (2020). Technology and parameters of cabbage machine harvesting by careful
stacking of heads in containers. IOP Conference Series: Earth and Environmental
Science.
Arad, B. B.-S. (2020). Development of a sweet pepper harvesting robot. Journal of Field
Robotics.
Chen, W. L. (2020). Detecting citrus in orchard environment by using improved Yolov4.
Scientific Programming. https://doi.org/10.1155/2020/8859237.
Chunmao Li, F. L. (2023). Recognition of citrus fruit and planning the robotic picking. Research
square.
Das, R., Baishya, N. J., & Bhattacharya, B. (2023). A review on tele-manipulators for remote diagnostic
procedures and surgery. CSI Transactions on ICT, 11(1), 31-37.
Didamony, M. I. (2020). Fabrication and evaluation of a cabbage harvester prototype.
Agriculture (Switzerland), https://doi.org/10.3390/agriculture10120631.
Fouad Abobatta, W. (2019). Nutritional Benefits of Citrus Fruits. American Journal of
Biomedical Science & Research. https://doi.org/10.34297/ajbsr.2019.03.000681.
Fue, K. P. (2020). An Extensive Review of Mobile Agricultural Robotics for Field Operations:
Focus on Cotton Harvesting . AgriEngineering. https://doi.org/10.3390/agrieThe
harvesting2010010.
Hamido, S. A. (2018). Harvesting method affects water dynamics and yield of sweet orange with
Huanglongbing. Agriculture (Switzerland). https://doi.org/10.3390/agriculture8030038.
Jawad, M. F. (2020). Farmer Robot for Harvesting and Maintaining Plants. International Journal
of Electrical and Electronics Engineering.
Jhala, K. B. (2018). Design and Development of Battery Operated Mango Harvester.
International Journal of Current Microbiology and Applied Sciences.
https://doi.org/10.20546/ijcmas.2018.708.230, 7(08).
Jia, W. Z. (2020). Apple harvesting robot under information technology: A review. In
International Journal of Advanced Robotic Systems (Vol. 17, Issue 3).
https://doi.org/10.1177/1729881420925310 .
Jiacheng Rong, J. F. (2022). Development and Evaluation of a Watermelon-Harvesting Robot
Prototype: Vision System and End-Effector. MDPI.
Jingjuan Zhao, Y. Y. (2020). Global agricultural robotics research and development:Trend
Forecasts . Journal of Physics: Conference Series.
Johannesson, P. &. (2017). Untangling the Web of Practices: Designing Information Systems in
Context. Systems, Signs & Actions. An International JouWpractice stationtion
Technology, Action, Communication and Workpractices, 10(1).
Kondo, N. (2012). Fruit harvesting robot and grading robot. Semantic Scholar, 7.
Kongai, H. M. (2020). Smallholder orange farmer access to markets in Uganda. African Crop
Science Journal. https://doi.org/10.4314/acsj.v28i2.11, 28(2).
Kumar, S. K. (2020). Recent Advances and Future Prospects of Robotics in Agriculture. 3.
Liu, Y. P. (2019). A Visual System of Citrus Picking Robot Using Convolutional Neural
Networks. 2018 5th International Conference on Systems and Informatics, ICSAI .
https://doi.org/10.1109/ICSAI.2018.8, 8.
Mohamed, T. (2017). An Innovative System for Picking Mango Fruit. Journal of Soil Sciences
and Agricultural Engineering. https://doi.org/10.21608/jssae.2017.37572, 8(9).
Nehra, M. R. (2018). Role of Information Technology in Agriculture. ResearchGate, 29.
Oceger, Y. (2021). Fruit factory boosts citrus farming . Kampala: New Vision .
Patel, A. &. (2016). An Ample Study of Numerous Intelligent Systems used for Agriculture:
Review. International Journal of Computer Applications.
https://doi.org/10.5120/ijca2016909856, 143(1).
Peng, C. V. ( (2021). ). A strawberry harvest-aiding system with crop-transport co-robots:
Design, development, and field evaluation. http://arxiv.org/abs/2107.13063.
Reddy, N. V. (2016). A critical review on agricultural robots. International Journal of
Mechanical Engineering and Technology (Vol. 7, Issue 4).
sheikh, S. e. (2022). Climate implementation summit. cop27.eg.
Singh, J. S. ((2021)). Horticultural Classification of Citrus Cult,ivars In Citrus . Research,
Development and Biotechnology. https://doi.org/10.5772/intechopen.96243.
Stella, C. (2022). Farm and community practice carried out at Soroti District. Busitema.
Stewart Lockie, K. F. (2020). The future of agricultural technologies. Australian Council of
Learned Academies.
Taeyong Choi, J. P.-J.-S. (2022). Work Efficiency Analysis of Multiple Heterogeneous Robots
for Harvesting Crops in Smart Greenhouses. Agronomy, 13.
Tianhu Liu, W. L. (2022). A Multi-Flexible-Fingered Roller Pineapple Harvesting Mechanism.
Agriculture https://doi.org/10.3390/agriculture12081175.
Vrochidou, E., & Tziridis, K. A. (2021). An Autonomous Grape‐Harvester Robot: Integrated
System Architecture. Electronics ,https://doi.org/10.3390/electronics10091056.
Wan Ishak, W. I. (2010). Design and Development of Eggplant Harvester for Gantry System.
Pertanika Journal of Science and Technology., 18(2).
Xiao Tang, Z. Z. (2023). A Novel Field-Oriented Control Algorithm for Permanent Magnet
Synchronous Motors in 60◦ Coordinate Systems.
Actuators.https://doi.org/10.3390/act12020092, 12(2).
Xu Xiao, Y. W. (2022). End-Effectors Developed for Citrus and Other Spherical Crops. applied
sciences.
Yeshmukhametov, A. K. (2019). Development of Continuum Robot Arm and Gripper for
Harvesting Cherry Tomatoes. https://doi.org/10.20944/preprints201912.0237.v1.
Yusoff, M. Z. ((2020).). Development of integrated loose fruit collector machine for oil palm
plantations. Bulletin of Electrical Engineering and Informatics. https://doi.org/10.1159,
9(2).