Picking the Best Bot: Collaboration Strategies for Humans and Bots in Order Pick Systems with Traveling Salesman Problem Routing.

This article analyzes collaboration policies between human pickers and autonomous mobile robots (AMRs) in e-commerce warehouse order picking, focusing on two main strategies: the swarm policy, where pickers dynamically switch between AMRs, and the system-directed policy, where a picker completes an order with a single AMR. Using a closed queuing network (CQN) model with a synchronization station and Monte Carlo simulations to estimate service rates, the study derives closed-form expressions for steady-state probabilities and throughput under stochastic conditions, validated against detailed discrete-event simulations with errors below 2%. Results indicate that the swarm policy generally achieves higher throughput, especially when the AMR-to-picker ratio and AMR speed exceed those of pickers, while the system-directed policy performs better with similar speeds, larger orders, or fewer AMRs. Additionally, cost analyses show AMR-assisted picking is more cost-effective for small or mixed orders and multi-shift operations, whereas manual picking remains competitive for large orders with moderate to high AMR costs; managerial guidelines emphasize tailoring policy choice to order size, resource ratios, speed, item allocation, and warehouse layout.