Dynamic Task Assignment of Machine Repairing via Reinforcement Learning Method

Abstract

There are many situations in which human work is involved at manufacturing sites, especially repair work associated with the restoration of manufacturing machines, which is performed by workers. However, when there are multiple workers or multiple machines to be restored, decision-making on who should be assigned to which task may be based on on-site experience. In this study, a policy that decides appropriate task assignments based on the skills and status of each worker and information on restoration task was learned by reinforcement learning. As a result, we were able to obtain a task assignment method that can improve the total facility utilization rate more than the conventional method.