Map-link Method for Efficient Path Planning of Car-like Vehicle

Abstract

In this paper, we propose map-link algorithm for path planning of non-holonomic vehicle. This method searches a shortest path efficiently from its initial position to the goal position while satisfying the non-holonomic constraints. Previous path planning methods have to check collision between the vehicle and obstacles. However, collision checking takes large computational cost because many path candidates exist in the case of non-holonomic vehicle. Our method makes correspondences between each point on the map and all path candidates in advance. Then, by referring the correspondences and selecting collision-free paths, the feasible path that connects the initial and the goal can be obtained without collision checking of robot paths and obstacles. Experimental results show that our method is 11 times faster than the method which checks collision when searching feasible path in simulation environments.