Monday, 19 August 2019

Design and Implementation of an Autonomous Surveillance Mobile Robot

Volume 14 Issue 2 November - January 2019
Research Paper
Vikas Kumar*, Arun Kumar Sah**, Prases Kumar Mohanty***
*-** B.Tech Graduate, Department of Mechanical Engineering, National Institute of Technology, Arunachal Pradesh, India.
***Assistant Professor, Department of Mechanical Engineering, National Institute of Technology, Arunachal Pradesh, India.
Kumar, V., Sah, A. K., Mohanty, P. K. (2019). Design and Implementation of an Autonomous Surveillance Mobile Robot with Obstacle Avoidance Capabilities using Raspberry PI. i-manager’s Journal on Future Engineering and Technology , 14 (2), 42-54.https://doi.org/10.26634/jfet.14.2.14856
Abstract
This paper focuses on exploring and analyzing the process of robot design and hardware implementation of the studies made on the autonomous mobile robot navigation reported in the paper, “Application of Deep Q-Learning for Wheel Mobile Robot Navigation” (Mohanty, Sah, Kumar, & Kundu, 2017). Incorporating autonomous robots into daily life for serving humanity has been a long-term goal for the robotics plethora. An autonomous mobile robot has tremendous application in various environments since they work without human intervention. The robot is defined as a device that is composed of the electronic, electrical, and mechanical systems with a brain imported from computer science. In this paper, a mobile robot is introduced which was fabricated using Raspberry Pi 3 B as a processing chip, range sensors, and camera, which are used for extracting raw sensory data from the environment and feeding it to the robot. The composed mobile robot can be remotely accessed from anywhere around the globe without being in the vicinity of the robot and can be controlled by the means of any gadget, regardless of whether a portable workstation, a versatile, or a tablet, which makes it perfectly suitable for surveillance, exploration, and military applications. For training the robot in the virtual environment, a simulation model was developed in python from scratch. The pre-trained model from the simulation was deployed for further training of the robot in the actual environment. Algorithms like obstacle detection and image recognition were merged together to equip the mobile robot with necessary controls. In the end, the progress of the robot was analyzed in different real environments and the performance accuracy of the obstacle avoidance ability of the mobile robot was calculated based on hit-rate matrices and tabulated.

No comments:

Post a Comment