Binocular Stereo Vision Based Obstacle Avoidance Algorithm for Autonomous Mobile Robots

Abstract

Binocular Stereo vision system has been actively used for real time obstacle avoidance in autonomous mobile robotics for the last century. The computation of free space is one of the essential tasks in this field. This paper describes algorithm for obstacle avoidance for mobile robots which can navigate through obstacle. While most of the paper based on stereo vision works on the disparity image but we are proposing a method based on reducing the 3D point cloud obtained from stereo camera after 3D reconstruction of the environment to build a stochastic representation of environment navigation map. The algorithm assigns each cell of the grid with a value (free or obstacle or unknown) which helps the robot avoid obstacles and navigate in real time. The algorithm has been successfully tested on "Lakshya" - an UGV <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Dagger</sup> platform in both outdoor and indoor condition.