Research On Walking Control Of Humanoid Robot Based On CPG

Walking control for the humanoid robot has always been a difficult but important issue in the robotics filed.At present,the walking control method for humanoid robot mainly based on the dynamics model of the robot,and these gaits not only have a large gap with human natural gaits,but also can't adapt the environment well.The discovery of the CPG(Central Pattern Generator)theory brings a new idea to the walking control for the humanoid robot by simulating the circadian rhythm movement mechanism.This thesis aims at bringing the CPG control mechanism into the study of the walking control for humanoid robot,and designs a CPG network model to control the humanoid robot NAO walking on the flat terrain or slope terrain.The main researches are as follows:Firstly,according to the structure characteristic of humanoid robot and its walking pattern,a CPG network model is designed to control the robot walking in this thesis.By modifying the traditional Matsuoka oscillator,we propose a suitable mathematical model for the motion planning of the humanoid robot,and design a CPG network model for humanoid walking according to the structural characteristics of the humanoid robot.Meanwhile,in order to improve the adaptability of the robot to the environment,two feedback strategies are introduced to the CPG network.Secondly,aiming at the issue of network parameter allocation in CPG control method,this thesis proposed an improved genetic algorithm to optimize the CPG network parameter.By modifying the selection operator and the crossover operator of the traditional genetic algorithm,we design a higher efficiency CPG network parameters optimization method.In the simulation environment,we use the improved genetic algorithm to optimize the CPG network parameters of the robot walking on the flat terrain or slope terrain.The effectiveness of the proposed algorithm is demonstrated by simulations.Finally,using the CPG network model designed in this thesis,we control the humanoid robot to walk on the flat terrain and slope terrain.In the process of actual walking of the humanoid robot NAO,the robot uses the phase reset strategy and the torso posture feedback strategy proposed in this thesis to adjust the output of the oscillators and the robot pose in real-time,and complete the walking task successfully.Experiments show that the feasibility and superiority of the CPG control method proposed in this thesis.