Analysing recurrent dynamical networks evolved for robot control

Abstract

This paper shows how a mixture of qualitative and quantitative analysis can be used to understand a particular brand of arbitrarily recurrent continuous dynamical neural network used to generate robust behaviours in autonomous mobile robots. These networks have been evolved in an open-ended way using an extended form of genetic algorithm. After brie y covering the background to our research, properties of special frequently occurring subnetworks are analysed mathematically. Networks evolved to control simple robots with low resolution sensing are then analysed using a combination of knowledge of these mathematical properties and careful interpretation of time plots of sensor, neuron and motor activities.