Collision-free path planning of stereo vision based mobile robots using power potential approach

This paper describes power potential approach for the collision-free path planning of stereo-vision based mobile robots. This approach relies on an obstacle detection procedure by stereo matching. We construct the potentials for the robot path planning using sources and sink. The proposed potential distributes sources discretely on the boundaries of obstacles and puts a sink at the goal point and then superpose the power potentials which is defined so that the potential for any one of the source points will have more influence on the robot than the potential for the sink point when the robot is near to source point and the total influence of the source potentials on the robot will be less than that of the sink potential when the robot is more than a defense distance away from the obstacle. We also propose the noble method of escaping from local minima by preserving the directional characteristic of the movement of the mobile robots and using the various potentials simultaneously.