Making better order fulfillment in multi-tote storage and retrieval autonomous mobile robot systems

The multi-tote storage and retrieval (MTSR) autonomous mobile robot (AMR) systems are increasingly prominent in e-commerce and third-party logistics. These systems feature robots capable of handling multiple totes per tour. The operational decisions of order fulfillment in MTSR AMR systems include assigning and sequencing orders and totes at various workstations and scheduling robots. The intricate interplay among order, tote, and robot significantly heightens the order fulfillment challenge in MTSR AMR systems. This study proposes a mixed-integer programming model that simultaneously determines the assignment and sequence of orders and totes, and the scheduling of robots in MTSR AMR systems with multiple workstations. We develop an item characteristic-driven adaptive large neighborhood search algorithm tailored to efficiently resolve this multifaceted problem. The numerical experiments demonstrate the effectiveness of the proposed algorithm, which swiftly yields optimal or near-optimal solutions for small-scale instances. For large-scale instances, the algorithm achieves a 50.2% reduction in makespan compared to the scheduling methods currently used in an actual warehouse. Keeping the number of robots fixed and increasing the buffer positions of the robots can lead to a substantial makespan reduction, up to 55.4%. Intriguingly, we find that augmenting the number of workstations does not proportionally decrease the makespan once the capacity of put wall at each workstation surpasses five order boxes. Furthermore, the experiments reveal that the optimal number of orders per wave is around 100, and a wider warehouse layout can reduce the makespan by 26.3% compared to a narrow layout.