Skip to Main content Skip to Navigation


The SMAC group is interested in modelling and solving complex problems using multi-agent technology. Complex systems are systems in which a great number of entities are interacting in a local manner and their evolution can only be predicted by experiment or simulation. In other words despite a perfect knowledge of the basic components of the system, no program able to predict its behaviour and shorter than the one describing the system in an exhaustive way is known. Controling the behaviour of such systems which contain non-linearities coming from feedback loops, as well as a coupling between levels (the system macro-level is produced by entities of the micro-level but it also constrains this micro-level) is the aim of our research works (illustrated by several PhD thesis, projects and applications)..

These works deal with six topics that are strongly connected by concepts such as adaptation, self-organisation and emergence.

  • Self-adaptation. The adaptive behaviour of a system enables it to react when interacting with a dynamic environment, either to get on with performing its task, or for improving its functioning [Robertson, 2000]. The challenge is then to conciliate the system autonomy with respect to unexpected situations and the required control a human being has to have on them.
  • Self-organisation, Cooperation and Emergence. Self-organisation is a mechanism or process that enables a system to change its organisation during its functioning without any explicit external control[A. Di Marzo, M.P. Gleizes, A. Karageorgos, TFGSO 2005]. In a multi-agent system, self-organisation is expressed by changing the organisation between agents. The challenge is therefore to find the local behaviours of agents (laws at the individual level), the rules of agents that implement self-organisation and that enable to design systems that are functionally adequate. The functional adequacy is the positive judgment adopted by final users about the system activity.
  • Design and Modelling. This topic aim is to improve and provide assistance for developing applications based on cooperative agents.
  • Simulation. Problematics of this topic is to study and develop tools and methods for exploring models. On the long view, the aim is to notably integrate tools required to study emergent properties into a platform for multi-agent simulation.
  • Collective Solving of Problems. The aim is to study how adaptive multi-agent systems provide a new solution for solving multi-objective, multi-discipline and multi-scale problems. This approach permits, in a dynamic environment, to offer a new configuration from a previous solution while minimising the endogenous modifications and, most of all, without starting again a new solving with every disruption.