Decision Rules for Robotic Mobile Fulfillment Systems

TL;DR

This paper evaluates decision rules in robotic mobile fulfillment systems through simulation, analyzing their impact on throughput and performance metrics in e-commerce warehouse settings.

Contribution

It introduces a detailed simulation framework to assess various decision rules and their effects on system performance in RMFS environments.

Findings

Unit throughput correlates strongly with other performance measures.

Decision rules for pick order assignment significantly affect throughput.

More robots are needed for high throughput in large SKU warehouses.

Abstract

The Robotic Mobile Fulfillment Systems (RMFS) is a new type of robotized, parts-to-picker material handling system, designed especially for e-commerce warehouses. Robots bring movable shelves, called pods, to workstations where inventory is put on or removed from the pods. This paper simulates both the pick and replenishment process and studies the order assignment, pod selection and pod storage assignment problems by evaluating multiple decision rules per problem. The discrete event simulation uses realistic robot movements and keeps track of every unit of inventory on every pod. We analyze seven performance measures, e.g. throughput capacity and order due time, and find that the unit throughput is strongly correlated with the other performance measures. We vary the number of robots, the number of pick stations, the number of SKUs (stock keeping units), the order size and whether returns need processing or not. The decision rules for pick order assignment have a strong impact on the unit throughput rate. This is not the case for replenishment order assignment, pod selection and pod storage. Furthermore, for warehouses with a large number of SKUs, more robots are needed for a high unit throughput rate, even if the number of pods and the dimensions of the storage area remain the same. Lastly, processing return orders only affects the unit throughput rate for warehouse with a large number of SKUs and large pick orders.