基于改进Q学习的复杂环境下AGV路径规划研究  

Research on AGV Path Planning in Complex Environment Based on Improved Q-learning

作　　者：刘光印 钱东海[1] 王志国 肖子鸣 LIU Guangyin;QIAN Donghai;WANG Zhiguo;XIAO Ziming

机构地区：[1]上海大学机电工程与自动化学院

出　　处：《计量与测试技术》2025年第3期84-88,94,共6页Metrology & Measurement Technique

摘　　要：针对传统Q-learning算法中存在的搜索效率低、学习速度慢、收敛条件难以有效确定等问题,本文提出一种改进算法,并基于Python tkinter组件,在复杂环境下的栅格地图进行仿真实验。结果表明:该算法能在复杂环境的AGV路径规划条件下找到最优或次优路径,提高了学习效率和收敛速度。Aiming at the problems of traditional Q-learning algorithm,such as low search efficiency,slow learning speed and difficult to effectively determine convergence conditions,an improved algorithm is proposed in this pa-per.Based on Python tkinter component,simulation experiments are carried out on raster maps in complex environ-ments.The results show that the algorithm can find the optimal or sub-optimal path under the condition of AGV path planning in complex environment,and improve the learning efficiency and convergence speed.

关 键 词：Q-learning算法 路径规划 AGV 强化学习 栅格地图 

分 类 号：TP2[自动化与计算机技术—检测技术与自动化装置]