Synchronization within Homogeneous Neural Modules Controlling a Simulated Hexapod Robot

摘要

In this paper we illustrate a system consisting of a collection of identical neural modules, communicating through signals that diffuse in space, that control a simulated hexapod robot with twelve DOF. The evolved neural controllers display an ability to generalize their ability to different environmental and body conditions. Synchronization and phase differentiation between joints and legs is achieved through simple acceleration/deceleration mechanisms based on local interactions.