Bio-inspired Center Pattern Generator based control models

This paper presents an overview of the Central Pattern Generator (CPG) based control models as well as their applications to various multi-joint or multi-DOF robots. The motivation is to provide a relevant and practical guide to the existing literature for engineers with an interest in the emerging CPG-based biocybernetics approach. The CPGs function as a biological neural network with the great capability of producing stable rhythmic patterns that are robust against external perturbations. Some conventional models to emulate CPG network are firstly reviewed, and their merits and faults are analyzed. Combined the current stage of CPG-based locomotion control with its critical issues, a two-phase control architecture is formed to accomplish autonomous motions in dynamic environments. The developing prospect of the CPG-based control is finally outlined in the context of bio-robotics.