In situations where it needs the actions of different agents to accomplish a task, the planning for the task involves selecting and organizing the actions belonging to multiple agents. Most existing multi-agent planning approaches to solving such a planning problem rely on the global knowledge of the capabilities of the agents, and are confronted with difficulties when the availability of the global knowledge is impossible or undesirable. Previously, we have proposed a distributed multi-agent planning algorithm Dis-graph planning to address the difficulties, which enables multiple agents to cooperate in generating a plan without relying on the global knowledge. In the work of this paper, we performed a theoretical study of Dis-graph planning approach, discussing its critical features such as completeness, soundness and terminability.
Proceedings of the 8th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2009), Budapest, Hungary, 10-15 May 2009 / Keith S. Decker, Jaime Simao Sichman, Carles Sierra and Cristiano Castelfranchi (eds.),
Vol. 2, pp. 1211-1212