Perpustakaan Digital ITB

Robotic Mobile Fulfillment Systems (RMFSs) improve order fulfillment by using robots to transport storage pods to picking stations. However, intensive robot movement also increases energy use. Energy performance is jointly affected by SKU placement, pod selection, pod return, and robot charging. Because these decisions share the same robots, pods, and warehouse state, improving them separately may create operational trade-offs. This study therefore develops an integrated operational strategy for Company X to reduce energy use by at least 1.5% while maintaining order-fulfillment performance. Using product and order-history data, the storage-allocation design combined Business Category Nearest Neighbor for new products, Dynamic Time Warpingbased hierarchical clustering for temporal demand patterns, Jaccard similarity for product co-occurrence, and the Mayfly Algorithm for optimization. Pick-pod selection and pod return-to-storage were modeled using Proximal Policy Optimization-based deep reinforcement learning. Charging management combined placement screening, Response Surface Methodology, and Taguchi robust design. All modules were implemented and evaluated individually and jointly in a Python- NetLogo agent-based simulation. The integrated solution was compared with the baseline using ten replications and statistical testing. The integrated solution reduced energy per completed order by 18.66%, increased completed order lines by 8.49%, and increased order-line pile-on by 4.52%. These improvements were statistically significant, while throughput did not differ significantly from the baseline. Individual evaluations showed that storage allocation, pick-pod selection, and charging management improved their target indicators, whereas the proposed return-to-storage policy did not reduce robot cycle time. The integrated strategy therefore improved energy efficiency without significantly reducing throughput. However, stronger coordination among the four modules and validation in an actual warehouse remain necessary.