An Intentional Framework for Communication in Multi-Agent Systems

Agent communication languages (ACLs) are designed to facilitate cooperation between agents by allowing them to collaborate, share information, and request help from other agents. The design of ACLs for autonomous underwater vehicles (AUVs) is constrained by features of the underwater environment and the inability of system designers to anticipate all of the messages underwater agents will need to construct during their missions and build them into the ACL. Thus, it is our position that ACLs used for underwater agent communication must be expressive enough to allow agents to construct novel messages during their missions. We introduce an analysis of natural language that reveals structures, which allow us to formalize communication in terms of minimal communicative act types. With this analysis, we provide a speech-act based framework for designing communication for multi-agent systems that allows for the construction of novel messages by agents allowing them to adapt their communicative needs to the uncertainties of their environment