Towards Replanning for Mobile Service Robots with Shared Information

Currently, multiple mobile robots, called CoBots, are deployed in an office building performing tasks for users such as delivering items and escorting visitors. Tasks for these robots are scheduled in parallel, but aside from the scheduling, the robots act entirely independently. In this work we explore how the robots can communicate dynamic changes in the environment, such as blocked corridors, closed doors at delivery locations, and robot failures, to accomplish their tasks more effectively. We introduce the concept of a “rationale graph” which documents the relationships between dynamic conditions and scheduling decisions. We propose to use the rationale graph to decide which information to share with which robots, and to replan for tasks which have rationales that are violated.