Joint optimization of order allocation, rack selection, and robot scheduling under flexible storage location

Parts-to-picker intelligent warehousing systems, driven by the concept of “moving goods to stationary pickers”, significantly enhance storage flexibility and facilitate an on-demand goods movement paradigm. However, current practical applications and theoretical studies predominantly rely on the mechanized reset mechanism of “returning racks to their original locations”, thereby undermining the system’s inherent potential for storage location optimization. This study addresses the joint optimization of order allocation, rack selection, and robot scheduling for a wave of orders under a flexible storage location policy. A mixed-integer programming model is formulated to minimize the wave completion time. To solve large-scale instances efficiently, the classical harmony search algorithm is enhanced through the integration of Q-learning. The proposed model and algorithm are validated using real-world operational data from a leading parts-to-picker warehousing enterprise in China. Analytical results demonstrate that the proposed flexible reset mechanism significantly improves the operational efficiency of such warehousing systems.