Automatic Navigation with Stereo Disparity

Abstract

Visual perception is a key source of information for autonomous navigation. Stereoscopicvision improves one-plane vision by adding the third dimension, depth, to attain a more accuratelocalization of objects within the sensed scene. This research used a compact stereo camera to finda path from crop rows in an agricultural field to support automatic vehicle guidance. Stereo disparitywas obtained from the segmentation of the obtained stereo images. Linear regression was appliedand a target point was selected from the central path determined by the combination of the fit lines. Abase rule to merge lines from different windows was implemented. The 3D coordinates of the targetpoint were computed by stereo analysis. This research has proven that stereovision has a greatpotential for autonomous navigation.