Principled communication for dynamic multi-robot task allocation

In the pursuit of an efficient cooperative multi-robot system, the researcher must eventually answer the question "how should robots communicate?"; a natural way to attack this question is to decompose it into three simpler corollaries: "what should robots communicate? when should they communicate?" and "with whom should they communicate?". In this paper, we propose answers to these questions in the form of a general framework for inter-robot communication and, more specifically, advocate its use in dynamic task allocation for teams of cooperative mobile robots. We base our communication model on publish/subscribe messaging and validate our system by using it in a tightly-coupled multi-robot manipulation task and a loosely-coupled long-term experiment involving many robots concurrently executing different tasks.