An Adaptive Local Path Planning Algorithm for Multi-robot Systems

Abstract

An improved dynamic window method (DWA) is proposed for the path planning of multi-robot systems for rescue in harsh environments. For a single robot unit, a cost function based on sensors' perception is constructed and added to the evaluation function to upgrade the performance of path evaluation. To develop the self-adaptability of the algorithm, fuzzy control is introduced to fine-tune the weights of the evaluation function to obtain effective path in different environments. For the multi-robot system, to enhance the rescue efficiency, the path reuse method is employed to reduce the planning time when the paths of multiple robots overlap. Finally, the proposed algorithm has been tested in indoor and corridor simulation environments to verify its feasibility and effectiveness in obstacle environments.