Dealing with Perception Errors in Multi-Robot System Coordination

A prerequisite to efficient behavior by a multi-robot team is the ability to accurately perceive the environment. In this paper, we present an approach to deal with sensing uncertainty at the coordination level. Specifically, robots attach information regarding features that caused the initiation of a course of action, to any coordination message for that activity. Further information regarding such features, acquired by the team, are then combined and the expected utility of the started action is re-evaluated accordingly. Experiments show that the approach allows to coordinate a large group of robots, addressing sensing uncertainty in a tractable way.