Robust Transfer Vehicle Swarm for Unreliable Production Facility

Abstract

We propose a dynamic rebalancing control algorithm for the swarm of transfer vehicles to improve the throughput of an unreliable production facility. The facility has unpredictable frequent changes in production capacity, where the optimal transfer task schedule of the swarm for a particular set of transfer requests may become obsolete within a short time period. Thus the proposed algorithm does not search for the optimal schedule but adapts the transfer behaviors of the swarm to arbitrary changes in production by rebalancing the stock levels. Through numerical experiments, we find that the throughput achieved by the proposed algorithm is 15% larger than that by the conventional manufacturing industry practice, and the necessary buffer capacity for the proposed algorithm is about 50% smaller than that for the conventional practice.