Central Pattern Generator Based Control of a Roller-Skating Robot

A bionic control method for a roller-skating robot locomotion is presented. The control system consisted of a high level controller and a central pattern generator(CPG) neural network. Each CPG is modeled as a modified Hopf nonlinear oscillator with frequency, amplitude and bias can be controlled individually. By changing the parameters of CPGs and connection weights between CPGs, the robot can realize multimodes locomotion of straight line skating and steering with different speed and direction. Furthermore, the control system can perform smooth transitions between different locomotion modes. The effectiveness of the proposed method is confirmed by simulation.