Designing CPG For Biped Locomotion Control Using Nonlinear Dynamical Theory

Generating biped locomotion patterns is a difficult task because maintaining balance on two legs is not easy. Nature has given humans a very stable and efficient locomotion. Therefore it is a good idea to take inspiration from biological world for developing natural and efficient walking patterns. In humans, Central Pattern Generator (CPG) is responsible for many rhythmic activities like locomotion, chewing, breathing, digestion etc.

In this work, a bipedal locomotion controller for simple walk is developed by drawing inspiration from biological CPG. The approach used for the work is nonlinear dynamic theory. The four body joints used for the modeling are two hip joints and two knee joints. The controller model is composed of four coupled Rayleigh oscillators, where each oscillator drives one joint. The four oscillators are coupled with each other. To search the values of CPG parameters genetic algorithm is proposed. Walking trajectories of adult humans are utilized to design the fitness function for genetic algorithm.