Biological CPG (central pattern generator) is responsible for the generation of animals' rhythmic motion such as hopping and running. In this paper, a novel feedback-based CPG control method is proposed, in which output of the CPG network is totally activated and contained by sensory feedbacks. The network responses to environment changes quickly with multi-feedbacks included, and has strong adaption to disturbance and non-flat terrain. This control method is applied to the controlling of a single-legged hopping robot to show the effectiveness of the approach. The simulation results prove that this control mechanism closely couples environment and robot motion, and dynamically stable hopping emerges from feedbacks contained in the neural networks.
展开▼
