Multi-Agent Coordination is Arbitration from a Super-Holon Point of View

This paper proposes an approach for engineer self-organised Multi-Agents Systems. This approach is based upon the human immune system. The role of the immune systems, in this approach, is the arbitration of several behaviours. We extend the arbitration capabilities of the immune system to define an architecture for multi-agent systems coordination. The proposed architecture is composed of several levels. Each level corresponds to an immune system. The behaviour resulting from the arbitration of a level is used as input by the level immediately below. This architecture allows the coordination of interacting entities at several level of abstraction. We illustrate this approach with a MAS dedicated to robot soccers simulation.