Evolutionary Multi-agents System for Prediction of Social Behavior in Large Cities

Abstract

We deal with complex software architectures based on agents. More precisely,we are interested with systems involving a very large amount of software weak agents. For instance, such a system can be the sub-system of representation of the world for an autonomous robot walking and acting in its environment. Another example is provided by the system of interpretation of behavior of social organization. In both cases, the systems are build with an intelligent control of its inner information stream. They are composed of a lot of no fixed parts, evolving in run according to the external information. All the inner parts are expressed with variable groups of agents.