Distributed route planning and control of material handling vehicles in a computer-integrated manufacturing system

In this paper, we address the problem of real-time routing and scheduling of material handing vehicles, and describe a distributed solution approach in which the vehicles behave as passive entities whose movements are directed by intelligent, autonomous Traffic Control Nodes (TCNs) that reside at intersection points or load pick-up/drop-off points in the material handling network. The routing and schedule for each vehicle is constructed in an incremental and dynamic manner (through a process of stage-by-stage auctioning of the vehicle among the TCNs) as opposed to a complete source-to-destination manner. Using a parallel and distributed simulation model of the TCN framework built with the Parallel Virtual Machine software library, we have investigated two adaptive strategies, namely single-stage and look-ahead, and the static shortest path strategy. Our experiments show that the adaptive strategies result in a better distribution of vehicle traffic in the network and thereby reduce congestion-related delays. A counter-intuitive finding was that the performance of the single-stage strategy was comparable to (and sometimes better than) than that of the lookahead strategy.