Multi-agent planning as a coordination model for self-organized systems

Self-organized systems need sophisticated coordination mechanisms. This paper proposes a model based on multi-agent planning to coordinate such systems. In our application framework: tactical aircraft simulation, it is important to deal with environment changes. For that, we based our approach on the hybrid automata formalism. The main advantage of this formalism is to model different clocks evolving with different speeds, for example, the altitude, the kerosene, the time, etc. In fact, all the parameters evolving dynamically can be modeled by a clock. This paper also shows how to obtain coherent multi-agent plans from individual one´s in such a context. This coherence is guaranteed through consistent coordination mechanisms well-founded since they are based on hybrid automata and their synchronization. The result is a flexible multi-agent plan in the form of a synchronized network of individual hybrid automata. This paper carries on introducing a first implementation of this approach in a project named SCALA developed at Dassault Aviation.