Assignment and Trajectory Planning for Two Robots with Sporadic Communication and Risk of Failure

Abstract

A common approach to distributing work to be completed simultaneously among different robots is called Multi-Robot Task Allocation. Particularly in a path planning scenario, two robots are assigned tasks to drive to two different destinations to perform some action, with emphasis on speediness. This paper tackles the real-world situation where some robots may fail, leading to any remaining robots reassigning their destinations to balance the differences in rewards and travel time between different locations. The catch is that when robots are restricted to sporadic communication due to environmental or resource constraints, surviving robots experience a delayed response to these failures - the uncertainty leading to many possible worlds where a late reassignment hinders making timely progress. We propose an approach where robots proactively plan their paths during these intervals of uncertainty in anticipation of possible robot failures. Specifically, robots will answer the question: to what location should I drive before my team’s next time of communication to best react on average to possible news of their failures? Observe the case when one robot will definitely fail – the other must drive directly to a particular destination. But when that same robot will definitely live, the other might drive to a different destination to balance the team-wise reward. For the case when that robot’s probability of failure is uncertain, we present that the surviving robot’s trajectory will incorporate a path that lies somewhere in between the prior two. We give examples where this is the best plan that handles delayed awareness of robot failure because, over multiple trials, it will ultimately travel less while collecting more reward.