A Reference Path Guided RRT* Method for the Local Path Planning of UGVs

This paper proposes a novel local path planning algorithm for Unmanned Ground Vehicles (UGVs). Rapidly-exploring Random Tree Star (RRT*) has recently gained immense popularity in the path planning as it provides a probabilistically complete and asymptotically optimal solution. But the uniform random sampling strategy of the traditional RRT* results in blind searching and massive computation. Our algorithm introduces a reference path guided non-uniform sampling approach to improve the planning performance. When the reference path is available, the RRT* based planner samples along the reference path. When there are obstacles on the reference path, guide points are generated to guide the sampling. Simulation results show that the algorithm can reduce the blindness of search, decrease the number of extended nodes, and generate smoother paths.