Planning with and for Multiagent Systems
Papers from the AAAI Workshop
Michael Brenner and Marie desJardins, Cochairs
Multiagent systems (MAS) have become an important subfield of AI, and several classical AI topics are now broadly studied in their MAS (i.e. distributed) variants. Multiagent planning (MAP) extends classical AI planning to domains where several agents can plan and act together. Application areas of MAP include multirobot environments, cooperating Internet agents, logistics, manufacturing, military tasks, etc. While related MAS disciplines (e.g. distributed constraint satisfaction) have benefited from standardized problem specifications and benchmarks, existing work on MAP is still very heterogeneous. Approaches differ for example in their emphasis on either the distributed planning or the distributed plan execution process, in the ways communication and perception are used, and in whether a global plan for all agents or a local plan for each agent is produced. Some of the underlying questions have been recently addressed in related fields, such as in extensions of classical planning to concurrent plan models or in distributed versions of heuristic search algorithms, but the diversity of MAP approaches makes it difficult for MAP research as a whole to benefit from these developments. Therefore, this workshop intends to bring together researchers working on any form of multiagent planning or in related fields to discuss their common and differing goals and research methods, and to identify potentials for collaboration and cross-fertilization.