The programmable CPG model based on Matsuoka oscillator and its application to robot locomotion

Abstract

The central pattern generator (CPG) is an important functional unit in the spinal cord which can produce rhythmic signals to control locomotion. Recently, there has been a growing interest in programmable central pattern generators (PCPG). In this paper, a new PCPG oscillator and a generic PCPG model based on the Matsuoka oscillator are presented. The perturbation method is used to determine the convergence of the generic PCPG model. The sine signal, the mix signal and chaotic signals are provided as inputs to the model, and the simulation results show that the generic PCPG can learn arbitrary periodic signals. In this paper, the generic PCPGs are allocated at each joint of the compass-like and the three-link robots and their outputs are chosen as joint position commands. The simulations show that the generic PCPG can be used to control robot locomotion effectively. The contributions of this paper are as follows: (1) A new PCPG oscillator based on the Matsuoka oscillator is presented as a beneficial enhancement to the PCPG oscillators. (2) A generic PCPG model is built comprising three PCPG oscillators. It can learn any periodic input signal. These findings are a significant contribution to generic PCPG research.

Dynamical State Forcing on Central Pattern Generators for Efficient Robot Locomotion Control

“…Others consider adjusting the amplitude of the oscillations [15]. The most related works are the ones that consider adjusting the shape of the signal [14,17,19,12].…”

“…This happens via the direct perturbation of the neural activities rather than some synaptic weights. Other CPG systems that can adapt the shape of their signals typically rely on additional CPG shaping networks with learning [14,17] or multiple CPG circuits [19,12]. Our system is simple and reactive.…”