主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2025, Vol. 33 ›› Issue (3): 264-276.doi: 10.16381/j.cnki.issn1003-207x.2023.0429cstr: 32146.14/j.cnki.issn1003-207x.2023.0429

• • 上一篇    下一篇

智能仓库中多AGV在线任务指派与全局路径规划问题研究

李昆鹏(), 韩雪芳   

  1. 华中科技大学管理学院,湖北 武汉 430074
  • 收稿日期:2023-03-15 修回日期:2023-05-16 出版日期:2025-03-25 发布日期:2025-04-07
  • 作者简介:李昆鹏(1978-),男(汉族),湖北枝江人,华中科技大学管理学院,教授,博士生导师,博士,研究方向:物流与供应链管理、生产运作管理,E-mail:likp@mail.hust.edu.cn.
  • 基金资助:
    国家自然科学基金项目(72131009)

Research on Online Task Assignment and Global Path Planning Problem of Multi-AGV in Intelligent Warehouses

Kunpeng Li(), Xuefang Han   

  1. School of management,Huazhong University of Science and Technology,Wuhan 430074,China
  • Received:2023-03-15 Revised:2023-05-16 Online:2025-03-25 Published:2025-04-07

摘要:

人工智能的快速发展加速了仓库的智慧化转型,越来越多的仓库引入大量AGV替代人工作业。智能仓库中多AGV在线全局调度问题成为当前的研究热点和难点。此问题包含任务指派和全局路径规划,需要考虑双向导引道路、无碰撞、AGV电量约束和任务时间窗等特性。针对该问题,本文以最小化AGV的运行时间和任务延迟惩罚之和为目标,建立混合整数线性规划模型。同时,设计一种多AGV任务指派与路径规划协同优化算法。首先,基于任务与AGV双重优先规则设计指派算法。其次,考虑多种碰撞情况,引入五种策略改进A*算法并设置重调度机制,全局规划AGV路径。为验证算法效果,引入分支切割算法求解小规模问题。12组小规模算例测试结果表明,本文提出的分支切割算法可将CPLEX下界平均提高59.89%,启发式算法求解结果与分支切割算法下界平均Gap值为7.23%。144组大规模算例测试结果表明,5种改进策略均能提升算法求解质量。改进算法可将传统算法的结果平均改善27.69%,且求解时间缩短至2秒内,效率提高130.01%。研究成果不仅适用于智能仓库的AGV调度决策,还能拓展至生产车间、自动化码头等路网相对规则、自动化程度较高的封闭场景中,为全局集中管控的多车调度问题提供参考。

关键词: 智能仓库, 自动导引小车, 在线调度, 路径规划, 分支切割

Abstract:

The rapid development of artificial intelligence has accelerated the intelligent transformation of warehouses, more and more warehouses have introduced a large number of AGVs to replace manual operations. The online global scheduling problem of multi-AGV in intelligent warehouses is a hot and challenging topic. It integrates task assignment and conflict-free routing, and also needs to consider constraints such as bidirectional network, conflict-free, AGV battery limitation, task time windows, etc. The objective of this problem is to minimize the sum of AGV running time and penalty cost of AGVs waiting at grids. Exploiting the structure of the problem, a mixed-integer linear programming model is established first. Meanwhile, a multi-AGV task assignment and path planning collaborative optimization algorithm is designed: firstly, an assignment algorithm is designed based on the dual priority rules of task and AGV. Secondly, considering multiple collision situations, five strategies are introduced to improve the A* algorithm and a rescheduling mechanism is set to globally plan the AGV path. To verify the performance of our algorithm, a branch-and-cut algorithm is introduced to solve small-scale problems. The results of 12 small-scale instances show that the branch-and-cut algorithm can improve the lower bound of CPLEX by 59.89% on average. The average gap between the results of our heuristic algorithm and the lower bound of the branch-and-cut algorithm is 7.23%. The results of 96 large-scale instances show that all five strategies are valid. Compared to traditional algorithms, the results are improved by our algorithm by an average of 27.69%, the solution time is shortened to within 2s, and the solution efficiency is improved by 130.01% on average. The research is not only applicable to the scheduling decision of AGVs in intelligent warehouses, but also can be extended to closed scenarios with relatively regular networks and high automation degrees, such as production workshops and automated docks, providing a reference for multi-vehicle scheduling problems with centralized global control.

Key words: intelligent warehouse, automated guided vehicles, online scheduling, path planning, branch-and-cut

中图分类号: