Distributed clustering for group formation and task allocation in multiagent systems: A swarm intelligence approach

Most clustering methods rely on central data structures and/or cannot cope with dynamically changing settings. Besides, these methods need some hints about the target clustering. However, issues related to the current use of Internet resources (distribution of data, privacy, etc.) require new ways of dealing with data clustering. In multiagent systems this is also becoming an issue as one wishes to group agents according to some features of the environment in order to have agents accomplishing the available tasks in an efficient way. In this paper we discuss the application of a clustering algorithm that is inspired by swarm intelligence techniques such as organization of bee colonies and task allocation among social insects. This application involves a complex task allocation scenario, the RoboCup Rescue, where tasks with different characteristics must be allocated to agents with different capabilities. Our results have shown that clustering agents is effective in this scenario as agents act in a more coordinated way.

Figure optionsDownload as PowerPoint slideHighlight
► We discuss a swarm intelligence approach for clustering agents, which is based on the bee dancing metaphor for recruitment.
► The proposed method works in a decentralized fashion, with agents clustering based on local information.
► Clustering is used for group formation in the RoboCup Rescue (a dynamic, multiagent environment).
► Our results in the RoboCup Rescue show higher scores than those obtained by isolated agents acting greedily.