Research on path planning of mobile robot based on improved Deep Q Network
Meng Guan, Fu Xing Yang, Ji Chao Jiao, Xin Ping Chen
Abstract
Abstract In order to solve the problem of slow convergence and low learning efficiency when mobile robots use ordinary DQN algorithm to plan path in an unknown environment with insufficient prior knowledge, an improved DQN algorithm is proposed. A heuristic reward function is designed to realize continuous reward and solve the problem of slow convergence of DQN algorithm caused by reward sparsity. An adaptive exploration strategy is proposed to solve the problem of balance between exploration and utilization in reinforcement learning process, and improve the efficiency of exploration. Experiments show that the improved DQN algorithm not only has high learning efficiency and fast convergence, but also the planned path is optimal.