Automatic path planning and navigation with stereo cameras

Abstract

Aiming at the scarcity of research of vehicle's navigation in non-signal environment, this paper prevents an approach of automatic path planning and navigation by two stereo cameras. Our approach is consist of three major steps. The first issue is stereo cameras calibration, to get parameters of camera. And then, SGM (Semi-Global Matching) algorithm, contribute in this part in order to get precise 3-Demotional landform. With the landform information, robot can relocate itself by comparing the landform got from satellite image in advance, in addition, avoiding unknown obstacles. Finally, vehicle followed path planned by A-star algorithm dynamically. The experiment shows that stereo cameras algorithm can accurately reconstruct 3-Dimentional landform, which directly make vehicle reorient itself accurately, therefore vehicle runs on the optimized path, navigation carry out successfully.