Multiagent Planning with Partially Ordered Temporal Plans

This paper discusses the specifics of planning in multiagent environments. It presents the formal framework MAPL (“maple”) for describing multiagent planning domains. MAPL allows to describe both qualitative and quantitative temporal relations among events, thus subsuming the temporal models of both PDDL 2.1 and POP. Other features are different levels of control over actions, modeling of agents’ ignorance of facts, and plan synchronization with communicative actions. For single-agent planning in multiagent domains, we present a novel forward-search algorithm synthesizing MAPL’s partially ordered temporal plans. Finally, we present a general distributed algorithm scheme for solving MAPL problems with several coordinating planners. These different contributions are intended as as step towards a simple, yet expressive standard for the description of multiagent planning domains and algorithms. Such a standard could in the future allow cross-evaluation of Multiagent Planning algorithms on standardized benchmarks. In this paper, we discuss the specific properties of planning in Multiagent Systems (MAS). With the term Multiagent Planning (MAP), we denote any kind of planning being performed in multiagent environments, meaning on the one hand that the planning process itself may be distributed among several planning agents, but also that individual plans can (and possibly must) take into account concurrent actions by several executing agents.