COOPERATIVE SELF-ORGANIZATION TO DESIGN ROBUST AND ADAPTIVE COLLECTIVES

摘要

Abstract: This paper highlights the benefits of using cooperation as the engine of adaptation and robustness for multi-agent systems. Our work is based on the AMAS (Adaptive Multi-Agent System) approach which considers co-operation as a self-organization mechanism to obtain adequate emergent global behaviors for systems coupled with complex and dynamic environments. A multi-robot resource transportation task illustrates the instanti-ation of a cooperative agent model equipped with both reactive and anticipative cooperation rules. Various experiments underline the relevance of this approach in difficult static or dynamic environments. 1