An implementation of typical-obstacle detection and recognition with laser range finder

This paper presents an implementation of detecting and recognizing typical obstacles with laser range finders. Mounting the laser range finder with a pitch angle on top of the robot, we can obtain the height information of the obstacle. Techniques, such as clustering, line segment extraction, coordinate transformation, line segment matching, are then applied on the scan data analyzing to detect and recognize the platforms and ditches. Experiments are conducted to validate our implementation.