Genetic integration in a multiagent system for job-shop scheduling

The Job-Shop scheduling problem constitutes a typical NP-Difficult problem. Determining an optimal solution is almost impossible, but trying to improve an existent solution is the way to lead to a tasks repartition which is better. We use Multi-Agents Systems (M.A.S.). These simulate the behavior of entities that are going to collaborate to accomplish actions on a Gantt diagram with the intention to better resolve a given economic function. Communications between global and local agents, components of the MAS, due to their actions, manage the appearance of agents of intermediate granularity and the global optimization in production scheduling. To have micro and meta-agents, a Multi-Objective Genetic algorithm is used, and especially on account of an ideal solution of such a problem which is a point where each objective function corresponds to the best (minimum) possible value. The genetic autonomy and the notion of motivation for an agent may lead to a drastically new kind of emergence phenomenon (different social behavior, auto-referring evaluation process, ...) in self-organizing multiagent systems. It is certainly a difficult task but it may set the seeds of a prolific approach concerning artificial life to optimize a Job-Shop Scheduling Problem.