Self-organizing robot teams using asynchronous situated co-evolution

Abstract

Self-organizing without a central controller in order to achieve collaboration towards an objective is one the main challenges in the design and operation of multi-robot systems. It is of great interest in the field to explore different approaches in order to achieve this end. Here we consider a distributed open-ended evolutionary approach called Asynchronous Situated Co-evolution (ASiCO) and introduce a series of biologically inspired concepts in order to address the solution of complex multi-robot problems with several objectives and which require the coordination of robots within distinct groups carrying out heterogeneous tasks. Different elements are explored in this paper, including how to efficiently implement a co-evolutionary approach that can operate in real time using only local information perceived by the real robots as they act on the environment and how these experiments can be tweaked in order to produce the desired behaviors from the teams and individual robots. © 2010 Springer-Verlag.