Robot Path Planning Algorithm Based on Improved Q-Learning in Dynamic Environment
Liu Jiale, Wang Bing, Yuquan Chen, Yang Wu, Ren Chen · 2023
A Q-learning path planning algorithm is proposed for the path planning problem based on the artificial potential field method in an unknown environment. First of all, an algorithm that the value of exploration coefficient decreases with exploration step increasing is put forward. In the process of Q value initialization, the artificial potential field method is introduced. Besides, obstacle area potential energy value is zero. Secondly, after adding new obstacles in the original environment, the mobile robot uses the combination of the trained Q table and the repulsive field thinking to avoid obstacles in the changed environment. The simulation results show that the improved algorithm is more able to learn adaptation in the path planning than the traditional Q-learning algorithm, and the optimal path step convergence speed is faster, and at the same time, it also has a good obstacle avoidance ability for dynamic obstacles, which verifies the feasibility and efficiency of the method.