Evaluating the Effectiveness of Biologically-Inspired Robot Control Networks through Operational Analysis

Abstract

Biologically-inspired approaches to the development of intelligent adaptive systems can lead to radically new classes of solution. However, it is not always clear how to evaluate the e¿ectiveness of these classes. In particular, no framework exists within which to address the issue of what kind of solution class is appropriate for what kind of problem. In this paper, we develop a methodology based on operational analysis of successfully evolved solutions, allowing us to identify properties of network classes that are potentially useful over a wider class of problems than the original problem for which solutions were evolved