Reinforcement learning for multi-item retrieval in the puzzle-based storage system

• A deep reinforcement learning algorithm is proposed for multi-item retrieval in the PBS system. compact integer programming model built to evaluate solution quality. conversion handle simultaneous movement. decomposition framework designed large-scale instances. The effect of several factors investigated deduce managerial insights. Nowadays, fast delivery services have created need high-density warehouses. puzzle-based storage system a practical way enhance density, however, facing difficulties process. In this work, algorithm, specifically Double&Dueling Deep Q Network, developed solve problem with general settings, where multiple desired items, escorts, and I/O points are placed randomly. Additionally, we propose Extensive numerical experiments demonstrate that approach can yield high-quality solutions outperforms three related state-of-the-art heuristic algorithms. Furthermore, movement instances respectively, thus improving applicability