电商物流智能仓库AGV路径规划研究

Study on AGV Path Planning in Intelligent Warehouses for E-Commerce Logistics

摘要: 电商物流作为电子商务运作的“大动脉”，其发展受到很多因素限制。其中，智能仓库的运转效率很大程度上决定了电商物流的效率。而如何高效规划AGV在智能仓库中的行驶路径又是智能仓库效率优化的关键问题，同时地图栅格化的精度也在一定程度上影响了路径规划的速度，为了有效提升智能仓库中AGV路径规划的能力，研究了基于改进冠豪猪优化算法的智能仓库AGV路径规划问题。首先，为了保证算法在初始阶段能拥有更加良好的全局搜索能力，通过引入改进Sine混沌映射算法，为冠豪猪优化算法提供混沌特性更加明显的初始可行解，为后续的优化过程提供良好的基础。接着，依照冠豪猪优化算法对AGV的路径规划问题进行求解。最后，仅靠冠豪猪优化算法的全局搜索可能无法在局部达成最优的路径规划效果。因此，通过引入A*算法进行局部搜索，将已有的路径进行随机分成数段小路径，并将分段采用A*算法进行求解。实验结果表明，在相同的参数设定与地图规模下，改进后的冠豪猪优化算法相较初始算法有着更强的寻优能力。同时，算法的求解能力与地图规模的大小密切相关，随着地图规模从小规模到中等规模再到大规模，寻优效果也从最初的1.9%的优化率逐步提升至25%的优化率。

Abstract: E-commerce logistics, as the “main artery” of e-commerce operations, faces many limitations in its development. Among them, the operational efficiency of intelligent warehouses largely determines the efficiency of e-commerce logistics. Efficiently planning the travel routes of AGVs (Automated Guided Vehicles) in intelligent warehouses is a key issue in optimizing warehouse efficiency. At the same time, the accuracy of map gridding also affects the speed of path planning to some extent. To effectively enhance the path planning capability of AGVs in intelligent warehouses, this study investigates AGV path planning in intelligent warehouses based on an improved Crested Porcupine Optimizer. Firstly, to ensure that the algorithm has better global search capabilities in the initial stage, an improved Sine chaotic mapping algorithm is introduced to provide the Crested Porcupine Optimizer with initial feasible solutions that have more pronounced chaotic characteristics, laying a solid foundation for subsequent optimization processes. Next, the Crested Porcupine Optimizer is used to solve the AGV path planning problem. Finally, relying solely on the global search of the Crested Porcupine Optimizer may not achieve optimal path planning locally. Therefore, the A* algorithm is introduced for local search by randomly dividing the existing path into several small segments and solving each segment using the A* algorithm. Experimental results show that, under the same parameter settings and map scale, the improved Crested Porcupine Optimizer has stronger optimization capabilities compared to the original algorithm. Furthermore, the algorithm’s performance is closely related to the map scale, with the optimization effect increasing from an initial 1.9% to 25% as the map scale grows from small to medium and then to large.