Coordinating Robot Teams for Disaster Relief

To perform complex tasks, a team of robots requires both reactive and deliberative planning. For reactive control, a restricted variant of Linear Temporal Logic called General Reactivity(1) can be used to synthesize correct-byconstruction controllers in polynomial time, but they often ignore time and resource constraints to maintain tractable synthesis. For deliberation, hierarchical planning can be used to reason about time and resources. However, the coordination of reactive control and deliberation remains a challenge, which we accomplish through a set of Coordination Variables. We integrate these two approaches in the Situated Decision Process (SDP), a system that we are developing. The SDP will allow an Operator to control a team of semi-autonomous vehicles performing information gathering tasks for Humanitarian Assistance / Disaster Relief operations. We demonstrate that the SDP responds to a dynamic, open world while ensuring that vehicles eventually perform their commanded actions.