Using fish-eye stereo vision system for autonomous vehicles

Abstract

Stereo cameras are widely used in robots as primary sensor for environmental perception due to its availability and low cost. Finding an effective method of obtaining the disparity images to generate a desirable local path for vehicles remains a difficult problem. In this article, we present a new method for autonomous navigation vehicles equipped with stereo, performing disparity calculation with the help of Fish-Eye lens. Disparity map is an effective way to describe a 3D Cartesian space, which makes path planning with better performance and safer. The proposed method takes advantage of a Fish Eye lens to acquire wide angle data of the environment and uses disparity map calculation to generate the 3D map of the field. The generated map is used in a Dynamic path planning algorithm to compute a safe path for the autonomous vehicle. Results show the proposed frame work successfully navigate the vehicle from point A to B without using any additional sensors.