Robustness, evolvability and complexity in Boolean network robots

Abstract

In this work, we present the results of a thorough analysis of the BN-robot's dynamics. We analysed the trajectories followed by the BN-robot in the space of the BN states. We found that the best performing BNs show the capability of maintaining previous learned behaviours (robustness), while adapting to new tasks to perform (evolvability). This property seems also to be positively correlated with a measure of the complexity of the BN.