Dynamic Obstacle Avoidance Using Mobile Robots

Abstract

Sensing and avoiding obstacles whether static or moving is a key requirement for mobile robots that function with minimum human supervision. This paper includes the methods studied and the observations recorded throughout the process of developing and testing algorithms on the mobile robot. The main objective of this project was to develop an algorithm that senses and avoids obstacles in its mapped environment while moving from one defined position to another. The project was divided into two parts i.e., software and hardware. The software part included developing and simulating the algorithms on a simulator, whereas the hardware part included testing the simulated algorithm on the actual robot. The main software used was ROS (Robot Operating System) which is a platform that provides unified approach for developing and testing algorithms, the algorithms were tested in a simulating environment called Gazebo which works in coordination with ROS. A single model of a mobile robot used to develop and test the algorithms is the PeopleBot which is equipped with IR sensors and LIDAR. The PeopleBot can also be equipped with a camera using which in addition to the sensors we mapped the working environment of the PeopleBot. The algorithms were tested in GAZEBO’s simulated environment and successful algorithms were then tested on the actual PeopleBot. For the hardware part, the algorithms were tested on the PeopleBot. PeopleBot is a two wheeled robot, it uses Red Hat Linux as its operating system, It has multiple sensors as well as a LIDAR already installed. Successfully simulated algorithms were tested on the PeopleBot to verify their simulated accuracy. The observed limitations were rigorously corrected and tested multiple times to maximize the efficiency of the algorithm. The environmental factors and mathematical calculations were considered to enhance the durability of the algorithms. The algorithms were successfully implemented in closed mapped environments.