Joint order batching and picker routing under human–robot collaboration: An efficient heuristic and practical insights

The growing reliance of warehouses on automation to improve productivity introduces new challenges, particularly in coordinating interactions between multiple picking entities. In this study, we address the problem of human–robot synchronisation in a picker-to-parts system, where Autonomous Mobile Robots (AMRs) transport bins across the warehouse, while human pickers move between locations to load them. This setting requires integrated order batching and routing of both pickers and AMRs. To tackle this, we propose mathematical models and an iterative decomposition-based heuristic to minimise makespan. The heuristic consists of two stages: (i) AMR batching and routing, and (ii) picker routing. In the first stage, we use a variable neighbourhood search to batch orders and route AMRs. In the second, we treat AMR arrival times at picking locations as due dates for human pickers and generate routes accordingly. Our computational experiments demonstrate the efficiency of the proposed approach and provide managerial insights. We show that collaborative picking under our setting decreases the makespan of a fully manual system by more than a half, and a fixed-resource-assigned system by 15% on average, with gaps reaching over 50% when a single picker is assigned to three or more AMRs. We also highlight the importance of resource balancing. Specifically, we find that increasing AMRs improves performance when picker capacity is sufficient, whereas adding pickers yields the greatest gains when AMRs outnumber them. Furthermore, fewer high-capacity resources outperform larger fleets of low-capacity ones. Finally, our findings are quite robust to varying speeds of picking entities.