Fast Scheduling of Autonomous Mobile Robots Under Task Space Constraints With Priorities

Automation is becoming more and more important to achieve high efficiency and productivities in manufacturing facilities, and there has been a large increase in the use of autonomous mobile robots (AMRs) for factory automation. With the number of AMRs increasing, how to optimally schedule them in a timely manner such that a large school of AMRs can finish all the assigned tasks within the shortest time presents a significant challenge for control engineers. Exhaustive search can provide an optimal solution. However, its associated computational time is too long to render it feasible for real-time control. This paper introduces a novel two-step algorithm for fast scheduling of AMRs that perform prioritized tasks involving transportation of tools/materials from a pick-up location to a drop-off point on the factory floor. The proposed two-step algorithm first clusters these tasks such that one cluster of tasks is assigned to one single AMR, followed by scheduling of the tasks within a cluster using a model-based learning technique. For the purpose of clustering and scheduling, a task space is defined. The results from the clustering and scheduling algorithms are compared with other widely used heuristic techniques. Both the clustering and the scheduling algorithms are shown to perform better on task sets of relevant sizes and generate real-time solutions for the scheduling of multiple AMRs under task space constraints with priorities.