Building of Occupancy Grid Map of an Autonomous Mobile Robot Based on Stereo Vision

스테레오 비전 방식을 이용한 자율 이동로봇의 격자지도 작성

This paper presents the way of building an occupancy grid map which a mobile robot needs to autonomously navigate in the unknown environment. A disparity map resulting from stereo matching can be converted into the 2D distance information. If the stereo matching has some errors, however, the subsequent map becomes unreliable. In this paper, a new morphological filter is proposed to reject 'spikes' of the disparity map due to stereo mismatch by considering the fact that these spikes occur locally. The new method has advantages that it is simpler and more easily realized than existing similar algorithms.