动态不确定环境下一种移动机器人路径规划方法.pdf

研究与开发
动态不确定环境下一种移动
机器人路径规划方法
王 帅
（煤炭科学研究总院沈阳研究院，辽宁 抚顺 113001）

摘要 提出了一种在动态不确定环境下基于信度分配 Sarsa 强化学习算法的移动机器人路
径规划方法。通过引入信度分配函数有效分配强化信号，来修正动作选择策略，提高了学习效
率和速度，能够很好地解决动态不确定环境下特别是存在随机运动障碍物的环境下移动机器人
路径规划问题。仿真实验说明该方法的有效性和可行性。
关键词：移动机器人；动态不确定环境；强化学习；信度分配；路径规划

A Method of Path Planning for Mobile Robot
in Dynamic and Unknown Environment
Wang Shuai
(Shenyang Branch of China Coal Research Institute, Fushun, Liaoning 113001)

Abstract A method of path planning for mobile robot based on credit assignment Sarsa
algorithm in dynamic and uncertain environment is proposed in this paper. Credit assignment function
is introduced to assign reinforcement signal and modify action selection strategy, and the learning
efficiency is improved. The problem of path planning can be solved effectively in dynamic and
uncertain environment especially having stochastic moving obstacles. Simulation results prove the
validity and feasibility of the method.
Key words：mobile robot；dy