GasNets and Other Evolvable Neural Networks Applied to Bipedal Locomotion

Abstract

Evolutionary robotics relies upon techniques involving the evolution of artificial neural networks to synthesize sensorimotor control systems for actual or physically simulated robots. This paper is a comparative study of three principal types of artificial neural networks; the Continuous Time Recurrent Neural Network (CTRNN), the Plastic Neural Network (PNN) and the GasNet. An attempt is made to evolve networks capable of achieving locomotion with a physically simulated biped. Of the 14 distinct networks tested, GasNets were the only network to achieve cyclical locomotion, although CTRNNs were able to attain a higher level of average fitness.