Joint optimisation of order batching and collaborative picker routing in human-assisted robot-to-part order picking

With the rise of warehouse automation due to the efficiency benefits it offers, many facilities use a collaborative system involving Autonomous Mobile Robots (AMRs) and human pickers in order-picking. This study examines a picker-to-part system where AMRs, equipped with bins, navigate the warehouse while a fewer number of human pickers load them. Human pickers, unburdened by carrying items, follow a free-floating policy, moving freely between AMRs without returning to the depot. The key challenge is jointly batching orders for AMRs and routing all the pickers to minimise the makespan, defined as the return time of the last AMR. To tackle this, we propose an interactive two-stage decomposition-based heuristic approach wherein we first solve the AMR order batching and routing problem using a Variable Neighbourhood Search (VNS) algorithm. In the second stage, we determine the routes for human pickers by treating AMR arrival times at item locations as soft due times. By iteratively refining the interaction between these two stages, we enhance solution quality. We validate our approach through sensitivity analysis and demonstrate its advantages over traditional fixed-assignment and manual picking systems.