EgoPlan: A Framework for Multi-Agent Planning Using Single Agent Planners

Planning problems are, in general, PSPACE-complete; large problems, especially multi-agent with required co-ordination, can be intractable or impractical to solve. Factored planning and both address this by separating into tractable sub-problems, but there are limitations the expressivity of existing planners ability handle tightly coupled problems. This paper presents EGOPLAN, a framework which factors problem related sub-problems solved iteratively calling on single agent planner. EGOPLAN is evaluated multi-robot test domain durative actions, coordination, temporal constraints, comparing performance planner, OPTIC-CPLEX, without EGOPLAN. Our results show that for our domain, using allows OPTIC-CPLEX solve twice as complex it significantly faster.