Should I stay or should I re-plan? A Monte-Carlo planning approach for flexible fleet control in Industry 4.0

Links:

Abstract

Robotic fleet for pick-and-delivery tasks is an important topic that has received significant attention, particularly in the context of Industry 4.0 scenarios. In this paper, we consider the problem of re-planning the robots’ paths in a complex and dynamic environment where robots may interact with dynamic (e.g human) or unexpected obstacles (e.g. fallen box). In particular, we propose an approach where a centralized fleet controller provides collision-free paths to the robots, and, if they encounter unexpected obstacles during their navigation (e.g., humans moving in the working area), a re-planning module aboard the robot decides whether it is better to wait or locally calculate another collision-free path exploiting other possible routes. The decision algorithm is formalized as a Partially Observable Markov Decision Process and solved using Monte-Carlo planning techniques. This approach has been tested and validated in simulation and in a research facility that reproduces an Industry 4.0 production line.