A Hierarchical Connectionist Central Pattern Generator Model for Controlling Three-dimensional Gaits of Snake-like Robots

摘要

A key to promoting the snake-like robot s adaptability is improving its 3-D locomotion ability. Although the connectionist central pattern generator models(CCPG) have merits such as low complexity, appropriate for the hardware implementation, the current CCPG models have difculties in generating the phase-coordinated control signals for the multi-degrees-of-freedom motions. Consequently, the CCPG s ability to generate the 3-D gaits of snake-like robots are seriously restricted. According to the layered structure of the biological CPG mechanism and the functions of the motoneuron, a hierarchical CCPG(HCCPG) model is proposed. The HCCPG is composed of three layers, namely the basic rhythm generation(RG) layer, the pattern formation(PF) layer, and the motion modulation(MM) layer. The motoneurons of the MM layer can independently modulate the amplitude and phase of the PF layer s output so it overcomes the difculty faced by the current CCPG models. The HCCPG model has merits such as strong gaits adjustment ability,small complexity, and good expendability, which make it appropriate for generating the 3-D gaits. Based on the HCCPG model, a 3-D gait control method is proposed. Simulations have validated this gait control method.