Distributed Multi-Robot Equitable Partitioning Algorithm for Allocation in Warehouse Picking Scenarios

Growth in e-commerce means that warehouses need to fulfil more orders in less time. Order fulfilment dominates operational cost (55% to 70 %), hence improvements can have substantial economic impacts. Warehouses can increase order picking throughput by using methods that account for the stochastic nature of real-time online order arrival. This paper introduces an improvement over traditional zone picking strategies by partitioning the warehouse into zones of equal work that account for spatio-temporal demand arrival. We then prescribe a service policy for the team of robots or human workers with fixed item-storage capacity to service the demands of a given zone. The policy and partitioning are designed to optimize steady state performance. Our method is not specific to a particular warehouse configuration and scales to large warehouses with many robots. We validate our algorithms' performance on simulated warehouse environments and show favourable performance compared to existing equitable partitioning methods and naive order to picker allocation. We show through simulation that a team of 5 robots with 5-item capacity collects 10-30% more items per day than in comparison methods.