In this study we co-evolve CPN controllers using a multi-objective co-evolutionary algorithm for the purpose of a robotic soccer-like game. Such controllers are characterized by a first Kohonen layer, where the inputs are self-organized. This is followed by a Grossberg layer, in which the mapping, of the input classes to the control outputs, occurs. The Kohonen layer is taught ahead of the evolutionary process by either self-organizing (CPN-SO) or alternatively by k-means clustering (CPN-KM). Next, the Grossberg layer undergoes a co-evolutionary process. Both multi-objective and single objective evolutionary schemes are explored. Numerical simulations are carried out for six different approaches based on the use of the three aforementioned networks (CPN-SO, CPN-KM, FFN) and the two co-evolutionary schemes (single and multi objective). The simulation results from all the six cases are compared using equal-effort analysis and end-game analysis. The results of the analysis are discussed and future work is suggested.
