Tote delivery optimisation for multi-tote storage and retrieval autonomous mobile robot system with multiple workstations
研究了多料箱存取自主移动机器人系统中多工作站的料箱配送问题，提出两阶段启发式算法和混合遗传算法，平均缩短完工时间20.7%，为仓库布局和资源配置提供管理启示。
This study addresses the multi-workstation tote delivery optimisation problem (MTDOP) in multi-tote storage and retrieval autonomous mobile robot (MTSR AMR) systems, aiming to minimise the order fulfillment makespan under fixed order assignments. A mixed-integer programming model is formulated to simultaneously optimise three key decisions: tote set composition, delivery sequencing, and robot scheduling. We propose scenario-specific algorithms for the MTDOP. For high-order-volume scenarios with pending orders, a two-stage heuristic is developed: a total-distance-based adaptive large neighbourhood search (TD-ALNS) for tote set composition and delivery sequencing, followed by a genetic algorithm for robot scheduling. For low-order-volume scenarios with fully released orders, a hybrid genetic algorithm (HGA) with embedded iterated local search performs joint optimisation. Numerical experiments demonstrate that the proposed TD-ALNS/HGA approach not only significantly outperforms Gurobi and the only existing popularity-driven benchmark, achieving an average makespan reduction of 20.7% in large-scale instances, but also clearly surpasses three other metaheuristic algorithms, including local search, variable neighbourhood search, and simulated annealing. Managerial insights based on sensitivity analysis include clustering co-ordered SKUs, adopting moderate warehouse layouts, controlling batch sizes, and prioritise expanding robots buffer capacities and put-wall capacities over increasing their quantities.