大型智能仓储多AGV无冲突路径规划研究  

Study on Large-Scale Intelligent Warehousing Multi-AGV Conflict-Free Path Planning

作　　者：丁海毅 佘世刚 强运哲 陆怡鹏 柯为 胡智喡 DING Haiyi;SHE Shigang;QIANG Yunzhe;LU Yipeng;KE Wei;HU Zhiwei(School of Mechanical Engineering and Rail Transit,Changzhou University,Changzhou Jiangsu 213164,China)

机构地区：[1]常州大学机械与轨道交通学院,江苏常州213164

出　　处：《机床与液压》2025年第3期74-80,共7页Machine Tool & Hydraulics

基　　金：江苏省产业前瞻与关键核心技术碳达峰碳中和专项项目(BE2022044)。

摘　　要：针对传统CBS算法在大型智能仓储环境中为大量AGV搜索路径时存在产生节点数量较多且搜索效率较低的问题,提出一种多地图融合搜索算法(MMFS)。通过在栅格地图上提取的拓扑地图中使用考虑道路拥挤度的Q-Learning算法进行初步路径搜索,并将其映射至栅格地图中,得到AGV行驶的粗略路径。在更精细的栅格地图中使用CBS算法搜索出AGV行驶的具体路径。仿真结果表明:在拓扑地图中进行预规划后,CBS算法产生的拓展节点减少了92.18%;MMFS算法的整体搜索时间仅为CBS算法的8.13%。In response to the issue of the traditional CBS algorithm generates a large number of nodes and has low search efficiency when searching paths for a large number of AGVs in a large-scale intelligent warehousing environment,a multi-map fusion search algorithm(MMFS)was proposed.The preliminary path search was carried out by using the Q-Learning algorithm considering road congestion in the topological map extracted from the grid map,and the rough path of the AGV was obtained by mapping it to the grid map.The identified paths were mapped back to the grid map to obtain the rough path for AGV navigation.The CBS algorithm was applied to the more detailed grid map to search for the specific paths for AGV movement.The simulation results indicate that expansion nodes generated by the CBS algorithm are reduced by 92.18%after pre-planning in the topological map;the overall search time of MMFS algorithm is only 8.13%that of CBS algorithm.

关 键 词：智能仓储 路径规划 多地图融合搜索算法 Q-Learning算法 基于冲突的搜索算法 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]