Automated Robot for Warehouse using

Image Processing
Dhrauv Gupta
B.E Pre-final year student
Model Institute of Engineering & Technology, Jammu
Dept of Electronics & Communication
Email:[email protected]

Abstract

The use of robotics for industrial applications is common, but nowadays the use of robots in big
warehouses is gaining popularity as it saves huge manpower. In warehouses the challenges are
different from the manufacturing industry. In our proposed work we are trying to solve some
challenges related to warehouse maintenance. In this highly developing society, time and
manpower are critical constraints for completion of tasks on large scales. Automation is playing
an important role to save human efforts in most of the regular and frequently carried works.
Behind the scenes in the e-commerce industry lies a sophisticated system of automation and
robotics to help bring down costs and deliver goods to customers at the fastest speeds possible.
One of the major and most commonly performed work is picking and placing of objects from
source to destination. In a multi-robot automation system, the allocation of tasks has an
important influence on the efficiency of the whole system because of the large number of robots
and orders. The project proposes a method to solve the task allocation problem. This problem has
a transportation issue to find a collision-free task allocation and then improve the capability of
task processing. The task time window and the power consumption of multi-robot is the major
concern. The results of this project will show good performance and can improve efficiency of
the task execution.

1. Introduction

As the warehouse plays a crucial role in the supply chain between the manufacturers and
end-user or consumers, it is more important to adopt automation in large warehouses and
 industries. The large e-commerce companies like FLIPKART, AMAZON, SNAPDEAL
etc. deal with millions of goods and products from one place to another. In this
technically growing world, we need to reduce the manually done work by efficiently
using the automation technology. This paper presents a modified design of Autonomous
Inventory Management for Warehouses using automated mobile robots. This proposed
work mainly focuses on robotic operations in logistics and warehouses, especially on
collision avoidance and detection of the destination where it should halt (ie., location of
desired object). It will adopt the surroundings itself by using Centrally Controlled
Mechanism (Image Processing). Industries, military undertakings, and the medical sector
are some of the fields where these robots are now being used. Hence, Robots are
preferred for automation of tasks.

1.1 Necessity
This system is designed specially for transportation of product and avoiding the obstacle
to reach the destination in the warehouse. As here using image processing for detection of
path and obstacle, there is no use of sensors in circuit, hence circuit becomes less
complex. The accuracy of the system will be higher on the same hand we will require less
manpower and hence it will be more economical for the Industries & all the human errors
will be automatically reduced.

1.2 Methodology
I. Robot operations:
It will get instructions from raspberry pi (centrally controlling unit) and the
microcontroller of the robot will control the movement of the robot in a predefined path
for different operations (forward, left,right,stop).
II. Dropping system:
Drop Mechanism will be controlled by a centrally controlling unit (Raspberry pi) which
includes servo control Mechanical movement for dropping as well as circular
movements.
III. Centralized control system:
Raspberry Pi will get the images from the camera and process it using opencv and numpy
for fast and complex operations. The result from raspberry pi will be used to control both
robotic operations as per predefined instructions.

2. Literature Review

2.1 ‘Path Planning and Co-simulation Control of 8 DOF Anthropomorphic Robotic

Arm’ by Sudharsan, J.* & Karunamoorthy, L.
This paper was published to focus on the efficiency of the path planning with the help of
Matlab and ADAMS simulation software. The software was used to execute the control
algorithm in a real time case and see the functional behavior of the system. This showed
the results of the real time working of the manipulator. To control the motions of joints a
joint cubic path control algorithm is used with the help of MATLAB. The end effector of
the robotic arm was kept similar to human palm and fingers. The arm was modelled in a
Pro-E software tool. Through simulations they got the results on torque, position, velocity
and acceleration of the robotic arm.

2.2 Design and fabrication of pick and place robotic arm By Dr.T.Sunil kumar,
K.sarath, Sd.Family, A.V.S.Bhagyesh and Sk.Althaf.
In this project they have designed a manipulator of improved accuracy by using servos to
power the joints in the robotic arm.The robotic arm is designed using CATIA software. In
this project we are going to fabricate a robotic arm which performs ASRS function. The
project covers the procedure for selection of the servos used to power each joint of the
arm in detail. We are selecting Aluminum as our fabrication material. The torque exerted
at each of the joints is going to be calculated in this project and a servo with the required
torque rating is being selected for each joint. Selecting a suitable servo controller and
control software for the Robotic arm is developed using Microsoft’s programming
language. Kinematic Analysis The branch of mechanics which deals with the study of the
system of bodies without giving any importance to any other factor like force, mass is
called Kinematics. There are two categories of kinematics for robotic arm: • Forward
kinematics • Inverse kinematics.[19] 2.20 PLC based Robot Manipulator Control using
Position based and Image based Algorithm By Harshavardhan Reddy Kunchala & Jack
Toporovsky The author of this paper has used Programmable Logic Controller for
controlling the manipulator and also used two artificial intelligence algorithms that is
position based and image based algorithms. The manipulator used has 5 Degrees of
Freedom. The position based algorithm calculates the joint velocity by the data provided
by the algorithm based on images. The main purpose of this paper was to use the data
from the images captured by the camera and the space coordinates to the specific object
which at last controls the movement of the robot while performing a certain task. The
position determining algorithm outputs the X coordinates and Y coordinates in the image
frame captured by integrated camera along with a scaling factor. The measurements
related to the gripper are done according to the scaling factor and used to get an idea of
the height of the gripper from the datum. This makes the process easier for the machine
to detect the position of the object and come in the same plane of the object. [20] 2.21
Autonomous Object Detection and Grasping Using Deep Learning for Design of an
Intelligent Assistive Robot Manipulation System by Sanzhar Rakhimkul, Anton Kim,
Askarbek Pazylbekov and Almas Shintemirov This paper highlights the method of
designing a smarter human machine interface for a robotic arm which is controlled by a
joystick. This was done by integrating a set of methods based on machine learning for
automatic object detection, an estimation of its position which was done by RGB-D
sensor using processing of the collected data. A three fingered mechanical gripper was
used to grasp the target object. The movement of the robot towards the object was started
by selection of a desired element in the GUI.

2.3 Object Detection and Recognition for a Pick and Place Robot by Rahul Kumar,
Sunil Lal, Sanjesh Kumar and Praneel Chand.
In this paper, the authors presented the modelling and implementation of the feature
extraction algorithm and two classifiers for object recognition and detection. The major
challenge faced in developing this image processing algorithm was that of making the
test subjects in compliance with the classifier parameters, resizing of the images
conceded in the loss of pixel data. Therefore, a centered image approach method was
applied. The accuracy of the classifier developed in this paper was 99.33% and the
accuracy for the feature extraction algorithm was 83.6443%. The overall system
performance of the image processing algorithm developed by the authors after
experimentation was 82.7162%. The authors developed an IP technique which will
involve the FE (Feature Extraction) and classification algorithms suitable for object
sorting tasks. The system developed was tested on a real time basis. Further the
developed IP technique was used on SCORBOT ER-4U (robotic arm platform) which
was refurbished and utilized to sort electronic components. This paper also covers the
algorithms of feature extraction and classification. Further discusses on the determination
of object location and also portrays all the results carried out for the development of the
algorithms.

2.4 Task Allocation of Intelligent Warehouse Picking System based on Multi-robot

Coalition by Xue Fei, Li Tao, Tang Hengliang, and Su Qinghua
In the paper, a multi-robot task allocation model method of intelligent warehouse picking
system based on table working method is proposed. First, the problem statement of the
intelligent warehouse system based on multi-robot is given. Second, according to the
given Intelligent warehouse system environment information, an integer programming
formulation is constructed considering the time-bound tasks, the number of task
allocation on a robot and the agent power consumption. Then, a new method is proposed
to solve the problem of multi-robot task allocation. Finally, Simulation results show that
the model method in the paper can improve the efficiency of task allocation.

2.5 A Novel Warehouse Multi-Robot Automation System with Semi-Complete and

Computationally Efficient Path Planning and Adaptive Genetic Task Allocation
Algorithms by Kam Fai Elvis Tsang, Yuqing Ni, Cheuk Fung Raphael Wong and
Ling Shi
In this paper, they have presented a computationally efficient local path planning
algorithm, namely the RERAPF algorithm, and a genetic task allocation algorithm with
heuristic learning rules. The semi-completeness of the RERAPF algorithm was proven by
showing that it is impossible for any robot to be trapped in any fixed region that does not
contain the goal position. The performances of each subsystem, and the overall system
were shown in simulation. Future work may include consideration of additional practical
constraints, such as battery and robot capacities. Also, an equivalent RERAPF algorithm
of the continuous time or continuous space configuration may also be considered to
derive a more general algorithm.

2.6 A Robot Navigation Method Based on RFID and QR Code in the Warehouse by
Wang Xiao-Long, Wu Chun-Fu, Li Guo-Dong, Chen Qing-Xie College of
Mechanical and Electrical Engineering Longyan University Longyan, China
The paper uses Python to write the path planning background software to carry on the
simulation experiment. The simulation map uses the grip warehouse model. Experiment
results show that if there is no pre-set traffic rules, directly using the A* algorithm to find
the path, the robot planning path failure is very likely.This paper gives the layout of
automated warehouse model, the fusion of RFID and QR code information to provide
robot positioning data, and finally proposed a combination of traffic rules of the path
planning process. However the default robot in this paper is constant velocity, path
planning needs to further consider the actual acceleration and deceleration process.

2.7 Low Cost Arduino Based Voice Controlled Pick and Drop Service with Movable
Robotic Arm by Shaikh Khaled and Bishal Karmakar
This paper presents the designing and implementation of a low cost Arduino based
Robotic Arm with the ability of picking and placing the object as instructed by human
voice. The robot also contains a feature of detection of objects by itself through sonar
technique before proceeding voice commands. Such a system can be helpful in enabling
users to do work in industry or mining applications without risking their lives and
bringing more perfection. The user just has to own a Smartphone which is present in
almost every one's hand and the robot. The voice instruction can be transmitted to robots
via Bluetooth interface with Android Apps. Arduino Mega works as the brain that
controls the arm and makes decisions before grabbing any object.

2.8 Low Cost Obstacle Avoidance Robot Vivek Hanumante, Sahadev Roy, Santanu
Maity
Almost all navigation robots demand some sort of obstacle detection, hence obstacle
avoidance strategy is of utter importance. Obstacle avoidance robots have a vast field of
application. They can be used as service robots, for the purpose of household work and so
many other indoor applications. Equally they have great importance in scientific
exploration and emergency rescue, there may be places that are dangerous for humans or
even impossible for humans to reach directly, then we should use robots to help us. In
those challenging environments, the robots need to gather information about their
surroundings to avoid obstacles. Nowadays, even in ordinary environments, people also
require that robots can detect and avoid obstacles. For example, an industrial robot in a
factory is expected to avoid workers so that it won’t hurt them. In conclusion, obstacle
avoidance is widely researched and applied in the world, and it is probable that most
robots in the future should have an obstacle avoidance function.

2.9 Obstacle Avoidance with Ultrasonic Sensors JOHANN BORENSTEIN AND

YORAM KOREN
This paper discusses the obstacle avoidance algorithm used for a mobile robot. Since the
algorithm depends heavily on the performance of the ultrasonic range finders, these
sensors and the effect of their limitations on the obstacle avoidance algorithm were
discussed as well.

2.10 An RFID Warehouse Robot Loh Poh Chuan, Ayob Johari, Mohd Helmy Abd
Wahab, Danial Md. Nor, Nik Shahidah Afifi Md. Tajuddin, Mohd Erdi Ayob
The main purpose of this project is to develop an autonomous robot with RFID
application. The final version of the product is a prototype of an RFID robot. The design
of the robot is similar to forklift. The RFID robot’s locomotion system and navigation
system was implemented using a PIC microcontroller. RFID robot uses the line follower
module for navigation and locomotion. RFID robots are able to lift up the item at a
certain height by generating a PWM signal to the servo motor. Half duplex serial
communication is done between PIC microcontroller and RFID reader. The PIC
microcontroller retrieves data from the RFID reader and displays it at LCD.

2.11 Deep Learning-based mobile robot for warehouse keeping Akash Awasthi, A.
Madhu Vamsi , P. Deepalakshmi
This mobile robot mainly focuses only on small hold warehouses that cannot afford a lot
of money and whose warehouses are not well organized to use more sophisticated
systems like KIVA etc. We can add a manipulator to this mobile robot which will pick
the desired detected objects using YOLO or Faster RCNN so that the picking process will
be fully automated. In order to get an exact contour of the object we can use pixel wise
semantic segmentation techniques like PSPnet,Mask RCNN instead of faster RCNN, but
it would be more computationally expensive.

2.12 Pick-place of dynamic objects by robot manipulator based on deep learning

and easy user interface teaching systems by Delowar Hossain and Shin-ichiro
Kaneko
In this paper, they proposed an approach for object recognition using DBNN. Application
of DBNN makes it easy for adding other objects. In addition, the objects can be
recognized in different positions, orientations and lighting conditions by increasing the
number of images used during the training process. The performance of DBNN was also
compared with that of DBN-DNN. The results showed that DBNN outperformed DBN-
DNN in terms of training time and success recognition rate.

2.13 Perception and Motion Planning for Pick-and-Place of Dynamic Objects by

Anthony Cowley1 , Benjamin Cohen1 , William Marshall2 , Camillo J. Taylor 1 and
Maxim Likhachev3
They have demonstrated pick-and-place operations performed by a PR2 at a rate of 6.7 s
per object at a 91% success rate. Similar operations on a moving work surface yield an
87% success rate. Room for improvement remains in the areas of system integration, and
end-effector customization. The speed at which the PR2’s arms can move proved to be a
limiting factor in system performance, and the gripper design was not particularly suited
to absorbing the impact of moving objects. Despite these mechanical limitations, the PR2
proved to be capable of responsive, high throughput object manipulation.

2.14 COP-MAN — Perception for Mobile Pick-and-Place in Human Living

Environments by Michael Beetz, Nico Blodow, Ulrich Klank, Zoltan Csaba Marton,
Dejan Pangercic, Radu Bogdan Rusu
The presented system provides the means for solving interesting perception for
manipulation tasks through adaptation to the environment. COP-MAN can draw on the
background knowledge provided by the model of the static part of the environment, and
provides up-to-date information about the dynamic aspects necessary to perform pick-
and-place tasks successfully. As a knowledge base, COP-MAN can be used to answer
queries of the tasks executives about object locations, arrangements, state changes, which
trigger the execution of additional perception routine pipelines to fill in missing
information. While the presented pipelines work robustly for different kinds of sensors,
the problem of sensor capabilities still remains an open question. It might be necessary
for example, that in order to obtain a good model of an object, that object has to be
picked up based on a rougher model and moved closer to one of the sensors. This will
require even deeper integration between the different systems.

2.15 Autonomous Object Pick-and-Sort Procedure for Industrial Robotics

Application by Lianjun Li, Yizhe Zhang, Michael Ripperger* and Jorge Nicho†,
Malathi Veeraraghavan and Andrea Fumagalli
This paper demonstrated the feasibility of implementing a pick-and-sort application in
which computer vision is combined with sophisticated motion planning to enable an
industrial robot arm to position itself and its end e®ector to successfully pick up objects
from a moving conveyor belt, move to the correct bin (based on object type) and place
the object in the bin. Details are provided on how software modules from various ROS
and ROS-I packages were successfully integrated facilitating rapid application
development. Experiments were conducted on two testbeds. On the ¯rst testbed, we
found that the robot execution time for picking up, moving and placing an object in its
right bin, takes approximately the same amount of time as the processing time required
for object recognition. A second testbed experiment showed that the required object
recognition time can be decreased by (i) moving some of the ROS packages to be
executed on faster servers and (ii) by applying an iterative multi-attempt recognition
procedure. A technique was presented to help tune the multi-attempt recognition
procedure. These findings are important for the design of our second version of Gilbreth,
which is described in the next section.

2.16 Collision Avoidance for Industrial Robots with Arbitrary Motion* by E.

Freund and H. Hoyer
In this paper, a coordinated operation is characterized, for example, by a direct handing
over of work pieces, by combined working on one work piece or unit, or by the mutual
collision avoidance of the arms of the robots if both systems perform independent tasks.
In all these cases, the control strategy has to include the avoidance of collisions between
the industrial robots and any obstacle. So guided automatic collision avoidance is the
condition for realizing a coordinated operation of industrial robots. This guided automatic
collision avoidance must operate in real-time.

2.17 Motion Planning for an Automated Pick and Place Robot in a Retail
Warehouse by Sharath Jotawar, Manish Soni and Swagat Kumar
This paper provides the implementation details of motion planning for picking objects
placed in a rack. An algorithm for auto path generation to pick objects using any one of
three pre-dened orientations for the manipulator has been described. Moveit is used to
combine the path generation with collision avoidance to generate the trajectory in joint
space, and execute the trajectory commands on the robot manipulator. The experimental
results for the path generation have been provided for robots in simulation and real
robots. The use of octomap in the planning scene has enabled generation of path avoiding
collision with nearby objects. The future work will be in generation of trajectory for
picking objects in any orientation.

2.18 Multi-Robot Task Allocation for Real-Time Hospital Logistics Seohyun Jeon,
Jaeyeon Lee and Jaehong Kim
This paper considered the application of multiple mobile robots for hospital logistics. To
increase the efficiency of using multiple robots, fleet optimization is required to deal with
the frequent on-demand delivery requests occurring in a conventional hospital. Fleet
optimization is applied by finding the robot that consumes the least cost for conducting a
new task. This paper proposed an algorithm for finding the shortest path by generating
combinations. Using the method, it generates possible combinations with minimum
iterations of the functions, which reduces computational burden. When all combinations
of the routing paths are generated, the overall distance is calculated from the distance
matrix. Evaluating this process for all robots, an optimal robot with the minimum
distance is selected for the newly assigned task.

2.19 Designing pick and place robots using bluetooth and memes by Naraala Ravali,
Mohammad Gandhi Babu, Pippala Sohila and Vanaparthi Vijaya
The objective of this paper has been achieved which was developing the hardware and
software for a gesture based robotic hand. From observations that have been made, it
clearly shows that its movement is precise, accurate, and is easy to control and user
friendly to use. The robotic hand has been developed successfully as the movement of the
robot can be controlled precisely. This robotic hand control method is expected to
overcome the problem such as placing or picking objects that are away from the user,
pick and place hazardous objects in a very fast and easy manner or augmenting our
abilities to perform such tasks.

2.20 Collision Detection for Industrial Collaborative Robots: A Deep Learning

Approach by Young Jin Heo1,2 , Dayeon Kim2 , Woongyong Lee1,2 , Hyoungkyun
Kim2
They have demonstrated the design and implementation of a novel collision detection
framework. To the best of our knowledge, this is the first implementation of a real-time
robot collision detector that uses a DNN. The proposed approach eliminated tedious
heuristic decision processes. A model-based observer can adequately estimate external
joint torque, but the process is vulnerable to uncertain models and noisy signals. In
contrast, a DNN is very insensitive to model uncertainties and sensor noises.
Furthermore, it can automatically extract features from high-dimensional sequential joint
signals. These features enable the proposed CNN-based method to achieve both high
sensitivity to collisions and low susceptibility to false alarms. In quantitative evaluations,
the proposed collision detector showed reliable accuracy and wide applicability to
general industrial robots.

2.21 Depth camera based collision avoidance via active robot control by
BernardSchmidtaLihuiWanga
This paper presents an integrated and cost-effective approach for real-time active
collision avoidance in a human–robot collaborative work cell that enables all-time safety
protection. This approach connects a set of vision and motion sensors to virtual 3D
models for real-time monitoring and collision detection in augmented virtuality, aiming
to improve the overall manufacturing performance. Rather than emergency stops that
prevent human–robot coexistence, our approach links collision detection to active robot
control through three collision-avoidance strategies: warning an operator, stopping a
robot, or modifying the robot path to avoid a collision with the operator. The outcomes of
the human–robot collaborative assembly are much improved flexibility, absolute human
safety without a fence, and better overall productivity.
2.22 Real-Time Collision Avoidance in Teleoperated Whole-Sensitive Robot Arm
Manipulators by Vladimir J. Lumelsky, Senior Member, IEEE, and Edward
Cheung
In this paper, the robot arm avoids obstacles automatically and in real time and moves in
a collision-free manner although no prior knowledge of the objects in the environment is
available to the motion planning system; no constraints are imposed on the obstacles’
shapes. Using this system, the operator is relieved of the burden of providing safety for
the robot arm and the surrounding objects, and can concentrate on general strategy and
control. The increase in the overall performance is especially significant in cases when
the whole arm body or its large portion are subject to potential collisions. Since the
difficulties the operators have in reasoning about collision avoidance strategies tend to
increase with the complexity of the system at hand, the approach is expected to be even
more beneficial in systems with dual and/or redundant arm manipulators.

2.23 Mobile Robot for Object Detection Using Image Processing by Dr. Himanshu
Borse, Amol Dumbare, Rohit Gaikwad, Nikhil Lende
This paper describes a robotic application that tracks a moving object by utilizing a
mobile robot with sensors and image processing. The robotic platform uses a visual
camera to sense the movement of the desired object and a range sensor to help the robot
detect and then avoid obstacles in real time while continuing to detect and follow the
desired object. In terms of real-time obstacle avoidance capacity, this paper also presents
an algorithm for this robotic application specifically. Experimental results show that the
robotic and intelligent system can fulfill the requirements for detecting an object and
avoiding obstacles simultaneously.

2.24 An Autonomous Image-guided Robotic System Simulating Industrial

Applications by Raza Ul Islam, Jamshed Iqbal, Sarah Manzoor, Aayman Khalid,
Sana Khan
This work combines image modeling to simulate industrial processing and provides
autonomy to extraction of necessary features. Filtering techniques have been applied
Object’s coordinates from input to model. A complete solution to the DOF robotic arm,
ED7220C has been IK model always provides accurate sure that the end-effector points
to th orientation, provided the point of workspace. In case the object is model prompts the
user. Overall sys board web cam is shown in Figure also shown here. However use of
proposed system has eliminated the demonstration of the concept, a common automation
i.e. pick and place has testing and verifying individual subsystem has been already inte
undergoing in-lab trials. Multiple i will be considered in the future. Overall S Robotic
Arm Control Teaching pendant Cam Object (f) ple image: (a) Original ging filter (d)
Laplacian ter and corner points processing with robot al application. Imagine the system.
Prior to from image, various lies on captured image. of the developed robot IK of a
widely used. The derived joint angles to make the required position and of interest is in
robot’s out of workspace, the stem integrated. Teaching pendant is computer vision in the
role of pendant.
2.25 Development of autonomous lane following and collision avoiding robot using
image processing by Atif Fazili, M.M Rashid and Diyaan ul Imam
Numerous techniques and mechanisms, which can be implemented to develop a lane
following robot. In order to obtain the optimal results, one should be careful to choose a
particular type of hardware components. Followed by the efficiency of the software
programming, which will in return determine how smoothly the robot fulfills its
objectives and aims to get the desired output. There are a lot of good things that one can
learn by getting exposed to all these various techniques including to increase one’s ability
and to eliminate limitations. We will try our best in order to maintain the maximum robot
speed without compromising its efficiency. In near future practical implementation of the
system will increase the facility of vehicle automation.

2.26 On-line learning and planning in a pick-and-place task demonstrated through

body manipulation by Antoine de Rengerve, Julien Hirel, Pierre Andry, Mathias
QUoy and Philippe Gaussier
In this paper, they presented a bio-inspired model that enables a robot to build a
representation of a task demonstrated through passive manipulation. This representation
is based on proprioceptive states that are recruited and constitute the motor primitives of
the system. It can be used by the system to plan its actions depending on visually
categorized goals. Our system can continuously develop its model of the task through
several interactions with humans. It was tested with a real robot on a changing pick-and-
place task. Of course, if the planning and transition rearrange quickly, our model still
lacks generalization capabilities, due to the use of very elementary motor-state
categorization and recognition. In future works, we will link the transition and planning
part of our model with ongoing works on the learning of more robust visuo-motor
attractors allowing to adapt small parts of the trajectory to local changes and detect multi-
modal invariants.

2.27 A comprehensive taxonomy for multi-robot task allocation by G. Ayorkor

Korsah1,2, Anthony Stentz2 and M. Bernardine Dias2
They have presented a new task allocation taxonomy based on the degree of
interrelatedness between agent–task utilities. The new taxonomy, iTax, is significantly
more comprehensive than the current de facto standard MRTA taxonomy, and provides a
useful way of categorizing task allocation problems in a manner that is strongly related to
problem difficulty. It groups task allocation problems into four natural classes that relate
to the problem complexity. Problems in the ND class can generally be modeled by the
linear assignment problem, and solved in polynomial time. Problems in the other classes
are generally NP-hard. For problems in the ID class, the schedules of individual agents
can be optimized independently of each other. For problems in the XD class, schedule
optimization requires coordination between agents. Finally, the CD class requires task
decomposition and task allocation to be performed simultaneously.

2.28 Development of pick and place robot for Industrial applications by Vishakha
Borkar1, Prof G.K.Andurkar
The design and development of pick and place robots has been carried out. A prototype
confirmed functional working of the robot system. This system would make it easier for
human beings to pick and place the risk of handling suspicious objects, which could be
hazardous in its present environment and workplace. Complex and complicated duties
can be achieved faster and more accurately with this design. A robotic arm is
implemented using Atmega16 in pick and place objects more safely without incurring
much damage. The robotic arm used here contains a soft catching gripper, which safely
handles the object. In the modern era, time and manpower are major constraints for the
completion of a task. By the use of products, the industrial activities and hazardous
operations can be done easily and safely in a short span of time. The use of soft catching
grippers and low power wireless communication techniques like RF modules makes our
system more effective when compared to other systems. The developed system is capable
of lifting only small weights.

2.29 Mobile Gantry Robot for Pick & Place Application by Lokesh P Chandak,
Aniket Junghare, Tejas Naik, Neema Ukani and Saurabh Chakole
“Mobile Gantry Robot for Pick and Place Application” is an industry oriented project
aiming specifically on pick and place application in the industries which came out as a
solution looking at the problem in other pick and place robots as well as conventional
robots. This project will ease out the various operations of the industries as far as the pick
and place application is concerned.

2.30 Multirobot coordination in pick-and-place tasks on a moving conveyor by H.

Işıl Bozma NE.Kalalıoğlu
This paper considers the problem of multirobot coordination in pick-and-place tasks on a
conveyor band. The robot team is composed of identical robots with mutually exclusive,
but neighboring workspaces. The products are fed in at one end of the band, move
through each workspace sequentially until being picked up and are collected at the other
end—if not picked up interim. Each robot has the same task—that is picking up and
packaging as many products as possible. We propose an approach based on
noncooperative game theory where each robot uses local observations of the conveyor
band and their neighbors' actions in order to decide on its actions. The developed
algorithm has been implemented and tested in a simulated manufacturing environment
using Webots. Results obtained from the simulations are analyzed using a variety of
statistical performance measures.
3. Limitations

[1]. In one paper, robotic arms are used for the pick and drop mechanism. But Robotic
arm is stationary so the pick and drop tasks are limited upto certain distance(depends on
robotic arm distance).

[2]. There is no mechanism for collision avoidance in some papers.

[3]. Tasks are not fully automated, i.e they are controlled manually.

[4]. The size of the robot is large in some paper which will increase its cost.

[5]. Single robot is controlled at a time using RFID and QR code.

[6]. Earlier projects were using batteries which needed continuous charging. As for
operating time, it can be up to 15 hours, while charging takes two to three hours and this
is considered a lot of downtime in an industrial setting.

4. Proposed Project

We are proposing this project in which we will develop a multi-robot to maintain various
warehouse activities including picking, dropping, sorting and segregation of objects. In
this project, instead of using sensors we will be using a central control system for
navigation purposes. Raspberry Pi will be used as a central control unit on which a
camera of higher resolution will be mounted, which will be capturing the image through
the technology known as “image processing”. Also the project will include collision
avoidance features in it.

5. Framework
Analysis & Designing
Specification
(Oct-Nov)
(Aug-Sept)

Automated
Robot for
Warehouse
using Image
Processing

Testing Implementation

(Feb onwards) (Dec-Jan)

6. Conclusion

This project results in development of an autonomous robot which is centrally controlled

using Raspberry Pi and performs warehouse activities.

7.References

[1]. Sudharsan, J.* & Karunamoorthy, L. “Path Planning and Co-simulation Control of 8
DOF Anthropomorphic Robotic Arm”, Department of Mechanical Engineering, CEG
Campus, Anna University, Chennai, India.

[2]. Dr.T.Sunil kumar, K.sarath, Sd.Family, A.V.S.Bhagyesh and Sk.Althaf “Design and
fabrication of pick and place robotic arm”, 2nd National Conference on Recent Trends in
Mechanical Engineering, GIST, Nellore· August 2020.

[3]. Rahul Kumar, Sunil Lal, Sanjesh Kumar and Praneel Chand “Object Detection and
Recognition for a Pick and Place Robot”
[4]. Xue, Fei, et al. "Task allocation of intelligent warehouse picking systems based on
multi-robot coalition." KSII Transactions on Internet and Information Systems (TIIS)
13.7 (2019): 3566-3582.

[5]. K. F. E. Tsang, Y. Ni, C. F. R. Wong and L. Shi, "A Novel Warehouse Multi-Robot
Automation System with Semi-Complete and Computationally Efficient Path Planning
and Adaptive Genetic Task Allocation Algorithms," 2018 15th International Conference
on Control, Automation, Robotics and Vision (ICARCV), 2018, pp. 1671-1676, doi:
10.1109/ICARCV.2018.8581092.

[6]. Xiao-Long, Wang, et al. "A robot navigation method based on RFID and QR code in
the warehouse." 2017 Chinese Automation Congress (CAC). IEEE, 2017.

[7]. Mostaque, S.K. and Karmakar, B. 2018. Low Cost Arduino Based Voice Controlled
Pick and Drop Service with Movable Robotic Arm. European Journal of Engineering
and Technology Research. 1, 5 (Jul. 2018), 29-33.

[8]. Hanumante, Vivek, Sahadev Roy, and Santanu Maity. "Low cost obstacle avoidance
robot." International Journal of Soft Computing and Engineering 3.4 (2013): 52-55.

[9]. J. Borenstein and Y. Koren, "Obstacle avoidance with ultrasonic sensors," in IEEE
Journal on Robotics and Automation, vol. 4, no. 2, pp. 213-218, April 1988, doi:
10.1109/56.2085.

[10]. L. P. Chuan, A. Johari, M. H. A. Wahab, D. M. Nor, N. S. A. M. Tajuddin and M.

E. Ayob, "An RFID warehouse robot," 2007 International Conference on Intelligent and
Advanced Systems, 2007, pp. 451-456, doi: 10.1109/ICIAS.2007.4658428.

[11]. Awasthi, Akash, A. Madhu Vamsi, and P. Deepalakshmi. "Deep Learning-based

mobile robot for warehouse keeping."

[12]. Hossain, Delowar, et al. "Pick-place of dynamic objects by robot manipulator based
on deep learning and easy user interface teaching systems." Industrial Robot: An
International Journal (2017).

[13]. A. Cowley, B. Cohen, W. Marshall, C. J. Taylor and M. Likhachev, "Perception and

motion planning for pick-and-place of dynamic objects," 2013 IEEE/RSJ International
Conference on Intelligent Robots and Systems, 2013, pp. 816-823, doi:
10.1109/IROS.2013.6696445.
[14]. Beetz, Michael, et al. "COP-MAN—Perception for Mobile Pick-and-Place in
Human Living Environments." Workshop, IROS. 2009.

[15]. Li, Lianjun, et al. "Autonomous object pick-and-sort procedure for industrial
robotics application." International Journal of Semantic Computing 13.02 (2019): 161-
183.

[16]. Freund, Eckhard, and Helmut Hoyer. "Collision avoidance for industrial robots with
arbitrary motion." Journal of robotic systems 1.4 (1984): 317-329.

[17]. Jotawar, Sharath, Manish Soni, and Swagat Kumar. "Motion Planning for an
Automated Pick and Place Robot in a Retail Warehouse." Proceedings of the Advances in
Robotics. 2017. 1-6.

[18]. Jeon, Seohyun, Jaeyeon Lee, and Jaehong Kim. "Multi-robot task allocation for
real-time hospital logistics." 2017 IEEE International Conference on Systems, Man, and
Cybernetics (SMC). IEEE, 2017.

[19]. Part, Accelerometer, and Platform It is basically an Accelerometer. "DESIGNING

OF PICK AND PLACE ROBOT USING BLUETOOTH AND MEMS." Complexity
International 24.01 (2020).

[20]. Heo, Young Jin, et al. "Collision detection for industrial collaborative robots: A
deep learning approach." IEEE Robotics and Automation Letters 4.2 (2019): 740-746.

[21]. Schmidt, Bernard, and Lihui Wang. "Depth camera based collision avoidance via
active robot control." Journal of manufacturing systems 33.4 (2014): 711-718.

[22]. V. J. Lumelsky and E. Cheung, "Real-time collision avoidance in teleoperated

whole-sensitive robot arm manipulators," in IEEE Transactions on Systems, Man, and
Cybernetics, vol. 23, no. 1, pp. 194-203, Jan.-Feb. 1993, doi: 10.1109/21.214777.

[23]. Borse, Himanshu, et al. "Mobile robot for object detection using image processing."
Global Journal of Computer Science and Technology (2012).

[24]. R. U. Islam, J. Iqbal, S. Manzoor, A. Khalid and S. Khan, "An autonomous image-
guided robotic system simulating industrial applications," 2012 7th International
Conference on System of Systems Engineering (SoSE), 2012, pp. 344-349, doi:
10.1109/SYSoSE.2012.6384195.
[25]. A. Fazili, D. ul Imaan and M. M. Rashid, "Development of autonomous lane
following and collision avoiding robot using image processing," 2012 International
Conference on Computer and Communication Engineering (ICCCE), 2012, pp. 331-337,
doi: 10.1109/ICCCE.2012.6271206.

[26]. A. de Rengervé, J. Hirel, P. Andry, M. Quoy and P. Gaussier, "On-line learning and
planning in a pick-and-place task demonstrated through body manipulation," 2011 IEEE
International Conference on Development and Learning (ICDL), 2011, pp. 1-6, doi:
10.1109/DEVLRN.2011.6037336.

[27]. Korsah GA, Stentz A, Dias MB. A comprehensive taxonomy for multi-robot task
allocation. The International Journal of Robotics Research. 2013;32(12):1495-1512.
doi:10.1177/0278364913496484

[28]. Borkar, Vishakha, and G. K. Andurkar. "Development of Pick and Place Robot for
Industrial applications." International Research Journal of Engineering and Technology
(IRJET)-Volume 4 (2017).

[29]. L. P. Chandak, A. Junghare, T. Naik, N. Ukani and S. Chakole, "Mobile Gantry

Robot for Pick & Place Application," 2020 IEEE International Students' Conference on
Electrical,Electronics and Computer Science (SCEECS), 2020, pp. 1-5, doi:
10.1109/SCEECS48394.2020.171.

[30]. Bozma, H. Işıl, and M. E. Kalalıoğlu. "Multirobot coordination in pick-and-place

tasks on a moving conveyor." Robotics and Computer-Integrated Manufacturing 28.4
(2012): 530-538.