A constraint optimization framework for fractured robot teams

In dangerous and uncertain environments initial plans must be revised. Communication failures hamper this replanning. We introduce fractured subteams as a novel formalism for modeling breakdowns in communication. We present a hybrid approach that employs distributed coordination mechanisms to provide robustness to these communication breakdowns and exploits opportunistic centralization. By modeling the problem as a mixed integer linear programming problem, we are able to apply constraint optimization techniques to efficiently find optimal or near optimal solutions to the difficult class of time critical tight coordination team planning problems. We then demonstrate that explicitly reasoning about communication failures through the incorporation of selective disruption minimization can improve team performance.