Obstacle detection based on Occupancy Grid Maps from virtual disparity image

We had reported on an obstacle detection method by a stereovision system up to now. The stereovision system can generate disparity image that express three-dimensional space. We could detect obstacles based on this disparity image, but there were the cases to mis-detect obstacles in the space that obstacle didn´t exist. We found this false detections appeared suddenly and appearing continuously in the same space was comparatively little. So in this paper, we propose the method that generates Occupancy Grid Maps based on measurement given by stereovision system and detects the obstacles robustly with detection and tracking of moving objects by this information.