PROBABILISTIC APPROACH FOR AUTONOMOUS NAVIGATION USING STEREO VISION

This paper concerns an outdoor and indoor autonomous navigation algorithm using stereo vision camera as the environment sensing device. The algorithm integrates binocular stereo vision system along with proposed probabilistic navigation approach to form a robust robotic system for navigation in unstructured environment. We use probabilistic clearance approach to avoid obstacles along with the introduction of speed control function for sharper turning and more time for the navigator to detect additional nearby obstacles. We show stereo vision as a remarkable alternative to active sensing devices such as sonar and laser range finder. Our approach considers 3D point data and thus it is able to avoid hanging obstacles with insufficient ground clearance and chair like obstacles also. We introduce the concept of low cost navigators which is very important for industrial and daily life applications. The algorithm has been successfully tested on an autonomous robot Lakshya- an Unmanned Ground Vehicle platform at Delhi College of Engineering, India.