Multi-Agent Planning

摘要

Multi-agent planning is an important issue in the multi-agent systems domain, that poses a number of challenges. As the agents are working in the same environment, they do not have only to plan their actions but also to coordinate their plans, so that they do not have any conflicts among them. It requires both an adequate plan representation and efficient interacting methods allowing agents to coordinate their plans. This chapter tries to answer the following questions: (1) What is multi-agent planning. (2) What are some of its important issues. (3) What are some of the important techniques of multi-agent planning and how can they be classified. We present two techniques in detail. The first one is a recursive model for the representation and the handling of plans by means of Recursive Petri Nets (RPN) which support the specification of concurrent activities, reasoning about simultaneous actions and continuous processes, a theory of verification and mechanisms of transformation (e.g. abstraction, refinement, merging). And in the second one, we describe how an agent can modify his temporal plan, when another agent comes up with a higher priority plan. As the latter has more priority, the former should modify his plan to remove any conflicts. We present a plan merging algorithm supported by a plan repairing algorithm to cope with this situation.