A hierarchical connectionist CPG controller for controlling the snake-like robot's 3-dimensional gaits

Abstract

Connectionist Central Pattern Generator models (CCPG) are helpful to understand how the CPG neural mechanism functions, and have relatively small complexity which makes them suitable for controlling snake-like robots. However, there are few CCPG models are constructed to generate the snake-like robot's three-dimensional gaits, which are important for adapation, and their gaits generation ability is also very inadequate. According to the CPG mechanism, a hierarchical CCPG model (HCCPG) with small complexity is proposed to implement the three-dimensional gaits better. The HCCPG has a two-layers structure, namely the basic rhythmic signal generation layer and the output signal modulation layer. The HCCPG can generate three-dimensional gaits well and is extendable. Based on the HCCPG, a three-dimensional gait control method is proposed. The simulations and experiments validate this method.