Using Agent Commitments as Planning Contexts

A key challenge to planning in multiagent systems is to deal with uncertainty related to coordination, which means to manage the interdependencies between agent activities. Typically, a decentralized agent has only partial knowledge of the system and must deal with nondeterministic outcomes of both local and nonlocal actions. While there are many multiagent planning approaches, the issue of uncertainty in coordination has not been adequately addressed. Thus, in this paper we propose a framework that incorporates uncertainty into agent planning and coordination by using agent commitments as the multiagent context for agent local planning. A new, richer model of commitment is proposed to represent uncertainty and resolve interdependency. We then discuss the techniques to enhance planning and coordination under this framework, and demonstrate how to apply our framework to existing approaches and how it may improve problem solving.