Trueba, Pedro, Abraham Prieto, Francisco Bellas, Pilar Caamaño, and Richard J. Duro. “Self-organization and specialization in multiagent systems through open-ended natural evolution.” In European Conference on the Applications of Evolutionary Computation , pp. 93-102. Springer, Berlin, Heidelberg, 2012.
URL1
This paper deals with the problem of autonomously organizing the behavior of a multiagent system through a distributed approach based on open-ended natural evolution. We computationally simulate life-like dynamics and their evolution from the definition of local and low level interactions, as used in Artificial Life simulations, in a distributed evolutionary algorithm called ASiCo (Asynchronous Situated Coevolution). In this algorithm, the agents that make up the population are situated in the environment and interact in an open-ended fashion, leading to emergent states or solutions. The aim of this paper is to analyze the capabilities of ASiCo for obtaining specialization in the multiagent system if required by the task. Furthermore, we want to study such specialization under changing conditions to show the intrinsic self-organization of this type of algorithm. The particular task selected here is multi-robot collective gathering, due to the suitability of ASiCo for its application to real robotic systems.