Evolution of Neural Controllers for Locomotion and Obstacle Avoidance in a Six-legged Robot

Abstract

This article describes how the SGOCE paradigm has been used within the context of a `minim al simulation' strategy to evolve neural networks controlling locom otion and obstacle avoidance in a six-legged robot.A standard genetic algor ithm has been used to evolve developmental prog rams according to which recurrent networks of leaky-integ rator neurons were g rown in a user-provided developmental substrate and were connected to the robot's sensors and actuators.Speci c g rammars have been used to limit the complexity of the developmental prog rams and of the corresponding neural controllers.Such controllers were rst evolved through simulation and then successfully downloaded on the real robot.