Binocular Vision Based Drivable Region Fast Detection for Indoor Mobile Robot

A fast method is proposed for mobile robot to detect drivable regions. The method bases on binocular vision without feature correspondence matching. First, the image is segmented into regions by searching contours. Then, part obstacle regions are determined by the vanishing line. Finally, the feature points extracted from the regions under the vanishing line are classified into ground points or obstacle points by various constraints, and the probabilities of drivable regions are computed to classify regions according to the feature classification. The results of indoor and outdoor experiments show that the method can detect drivable regions quickly and robustly.