A society of agents for decision making in Supply Chain Management

In the current economical context (evolving markets, concurrence, etc.), the firms need to react decisively within their Supply Chain. This notion of reactivity comes from a new vision of the firm where actors are needed to be autonomous and flexible, and the decision processes are distributed to all the levels. This shared-out decision making process implies a cooperative way of working and involves the interaction of several different objectives to gather with heuristics and other organizing structures. In this context, we have set to work on the Multi-Agent Systems (MAS) studies, and we are proposing here a cognitive agent architecture based on a multi-module approach which enables an independent and modular description of each assignment, namely perception, communication and decision-making reasoning. This model permits to study the behavior of decisional agents, in particular important concepts such as negotiation, cooperation, conflicts, learning (the agents self-training process), emergence... and so on. The organization is described as the filling together on successive levels of partly autonomous integrated. modules. In these components behavioral orientation phenomena are modeled through the rationalities which motivate the decisions and the actions of the agents. This model will then be tested and validated on a supply chain management simulation example.