Implicit Cooperation and Antagonism in Multi-Agent Systems

Abstract

This paper fits in the framework of Distributed Artificial Intelligence and more specifically in the field of Multi-Agent Systems (MAS). We show why antagonism should be considered in MAS made up of autonomous agents and how these systems can even turn antagonism to advantage. We present a case study of a MAS aimed at exploring the possibilities of antagonism, oriented towards mobile robotics applications. Experimental results relative to a preliminary study of antagonism, namely implicit cooperation, illustrating the complexity and the power of this approach, are reported. Keywords: Multi-agent systems, autonomy, collective intelligence, cooperation, antagonism, distributed artificial intelligence. Introduction Multi-Agent Systems (MAS) are a recent area of Distributed Artificial Intelligence (DAI). An agent is a fuzzy notion which may designate either a physical entity (a computer, a robot, a human) or a formal entity (a process, a program) according to the target domain. An agent ...