Abstract: For the path planning problem of the Automated Guided Vehicle（AGV）, many static algorithms can effectively solve. Due to the mutual influence and restriction of simultaneous preemption of system resources among AGVs, deadlock, collision and other problems may occur in collaborative operations of multiple AGV, and the static path planning algorithm cannot meet the system requirements of real-time dynamic operations. In the intelligent warehouse system, the core problem of multi-AGV dynamic path planning is not a single AGV quickly solving the optimal path, but multi-AGV conflict avoidance and resolution to achieve the overall coordination of the optimal. This paper adopts two solutions to the appeal issue. One solution is to improve the most effective static algorithm and introduce dynamic mechanism and conflict resolution strategy to meet the job requirements. Another scheme proposes an active obstacle avoidance algorithm with multi-step foresight, which optimizes the path and avoids the congested section in advance to reduce the possibility of conflict and the cost of rerouting. Experimental results show that both algorithms have good robustness and can effectively solve conflicts, and the latter algorithm can continuously expand the number of AGVs and has higher system efficiency.

Key words: Automated Guided Vehicle（AGV）, path planning, dynamic pathfinding, real-time obstacle avoidance, warehousing systems, forward prediction

摘要： 对于自动导引车（Automated Guided Vehicle，AGV）的单机路径规划问题，已存在很多静态算法可以有效求解。但由于AGV间抢占系统资源的相互影响和制约，多AGV的协同作业会出现死锁、碰撞冲突等问题，静态路径规划算法无法满足实时动态作业的系统需求。智能仓储系统中，多AGV动态路径规划的核心问题不再仅是单AGV快速求解最优路径，而在于多AGV的冲突避免或解决，达到整体协调最优。拟采用两种思路解决上诉问题：一种方案是对最有效的静态算法进行改进，并引入动态机制和冲突解决策略以满足作业需求；另一种方案提出一种具备多步前瞻性的主动避障算法，优化路径并提前避开交通拥堵路段，减少冲突可能性和重新寻路代价。实验结果表明两种算法都具有良好的鲁棒性，可有效解决冲突，且后者可持续扩展AGV数量，具有更高的系统效率。

关键词: 自动导引车, 路径规划, 动态寻路, 实时避障, 仓储系统, 前瞻预测