International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 07 Issue: 07 | July 2020 www.irjet.net p-ISSN: 2395-0072
© 2020, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 864
Development and Design of Autonomous Navigation Robot Using
Raspberry Pi and Computer Vision
Dhruv Patel
1
, Priyam Dalwadi
2
, Shubham Dhumal
3
, Abhishek Bhalodiya
4
1-4
U.G. Student Mechatronics Engineering, ITM Vocational University - Vadodara, Gujarat, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - Visual perception is the most important ability
for any robot, animal or social animal to learn, process and act
accordingly. Our paper represents the computer vision-based
approach programmed in python language - using various
libraries like OpenCV and NumPy on Raspberry Pi 3 model b+
to navigate robot autonomously. The high definition USB
camera mounted on a custom-built chassis (.approx. 23 X 20
cm) captures the surrounding at several field rates. An
algorithm based on simple geometric shape, area and color
detection is developed to control the behavior of wheels.
Digital image processing techniques like image thresholding
and contours detection are used. The robot will operate
autonomously without any manual control, it will act only on
the data input from the camera, processed by the mini-
computer as per the program written in it. Our experiment
illustrates the efficacy of technology to gain results with least
error.
Key Words: Computer Vision, Digital Image Processing,
Algorithm, Raspberry-Pi 3 model b+, Robot.
1. INTRODUCTION
Human exposure to an unsafe work environment and recent
COVID-19 pandemic has boosted the need to develop a
technological solution. Small steps can be taken to reduce
this vulnerability. 'Computer Vision' refers to the human
visual like the tendency for a computer such as extracting,
classifying and creating a model to act effectively. Lawrence
Roberts (PhD at MIT, the Year 1963) is known as "Father of
Computer Vision". "Machine
Perception of Three-Dimensional Solids" is his most admired
work. Robots are multipurpose programmable, versatile
manipulators forged to locomote material, parts, tools, or
specialized devices through various prearrange motions for
the execution of a variety of assignments. The cyber-physical
augmentation of computer vision on the robot or machine to
perform a task is known as robot vision.
Some work related to autonomous navigation has been done
previously. Unlike other computer vision-based autonomous
navigation techniques, to ease the process of way-finding use
of simple geometric shapes of a specific colour and area
calculation has been involved in this paper. This concept will
navigate the robot more effectively. It will transform and
advance more research based on driverless vehicle
technology.
The prototype involves a USB camera mounted on the
front portion of a custom-built Acrylic robot chassis,
Raspberry-pi 4 as robot brain, jumper wire as nerves to
carry electric impulse, 12V DC geared motor as Actuator, Li-
po battery. The USB camera will transmit the robot's view to
Raspberry-Pi 3 model b+. The brain of the robot Raspberry-
Pi processes raw input video using computer vision
technique. OpenCV is a computer vision library used to
perform digital image processing to control the vehicle in a
smarter and intelligent way. The output generated during
the various processes in raspberry-pi controls the behaviour
of the robot by generating signals in the form of pulses to the
motor driver circuit for wheel movement. Thus a complete
system is established. Such types of an autonomous robot
come to aid in a hazardous/non-hazardous environment to
explore, deliver material or to carry operations by adding
more functionality to it.
2. SOFTWARE IMPLEMENTATION
A. Modus Operandi
The environment which is captured using HD usb camera acts
as input for raspberry pi. It is further processed by the
algorithm based on the computer vision tool. The algorithm
enhances geometric shapes that produce commands to
perform the operation by the robot using a motor driver such
as moving forward, stopping; turning left and right of the
wheels. Computer vision techniques help us to understand
more of image information, mainly the area of contour and
counting edges of geometric shapes.