We present a teamwork coordination approach called the Rolegraph coordination strategy. It is used to dynamically coordinate agent teams that perform distributed collaborative tasks. Current agent coordination approaches recognise team activity by considering team member mental states using explicit communication, plan recognition, and observation. Such information is sometimes incomplete or unavailable. Our strategy recognises team activities by analysing role relationships in hierarchical teamwork structures. At runtime, graph representations of team structures, called Rolegraphs, are extracted, capturing collaborative team activity. Rolegraphs are approximately matched against predefined templates of known behaviour to infer agent mental states, such as intention, in order to recognise team activities. A case base is then referenced to retrieve suitable examples of coordinated team action.
|Title of host publication||Agents and Computational Autonomy|
|Subtitle of host publication||Potential, Risks, and Solutions|
|Editors||M Nickles, M Rovatsos, G Weiss|
|Place of Publication||Berlin, Germany|
|Number of pages||12|
|Publication status||Published - 2004|
|Name||Lecture Notes in Computer Science|