A Path Planning Algorithm for Mobile Robot Based on MR-QL in the Unknown Scene

Abstract The Q-Learning (QL) algorithm is widely used for path planning. As the scene in which mobile robot located becomes complex, suffers from limitations of low convergence speed and long exploration paths. Therefore, a Max Reward-Q-learning (MR-QL) planning based on maximum reward proposed complex unknown scenarios. original algorithm’s discrete function action selection strategy are improved, new designed to dynamically adjust mechanism heuristically guide motion. also optimized avoid invalid improve convergence. Finally, three experimental environments with different complexity constructed demonstrate feasibility algorithm. simulation results show that MR-QL about 50% terms step length training time, better than