Planning near-minimum-length collision-free paths for robots

This paper describes an algorithm for the automatic generation of near-minimum-length collision-free paths for robots. Obstacles are assumed to have polygonal cross sections. This algorithm requires very small data storage. Expansion of obstacles and shrinkage of robots to a point are used to simplify the analysis of the robot collision detection problem. The robot path is considered to be composed of straight line segments. Collision detection criteria between the robot path and the obstacles are discussed. A scheme to search for near-minimum length collision-free paths is presented. This scheme has the benefit of being concise and suitable for real-time implementation, especially for robots working in cluttered environments. Examples to show the above ideas are included.<
>