Study On Gait Planning And Control Of Biped Robots

With the speeding up of industrialization process in Artificial Intelligence(AI),robots have become one of the most important branches in that field.Due to its unique features and anthropomorphism,biped robots have much better adaptation to environments than other type robots.Biped robots are a complex and inherently unstable system,so the study on gait planning and control of biped robots have a great significance to maintain robots' stability.In this thesis,a gait planning method is designed to let the humanoid robot NAO walk.Based on the analysis of Aldebaran Walk Engine,the gait planning algorithm of 3D linear inverted pendulum model is adopted and further improved.It improves the robot walking speed and dynamic ability.The main research contents are as follows:Firstly,kinematic analysis and simulation.The kinematics and the inverse kinematics equation of NAO are solved based on kinematics analysis by simplifying robot into planar kinematic chain model.The accuracy of the kinematic model which was developed in MATLAB is verified by experiment.Secondly,gait planning.According to the analysis of simulation of Aldebaran Walk Engine,the method of 3D linear inverted pendulum model is adopted to design the gait of the NAO robot.In addition,to solve the problems existing in the traditional gait planning method based on 3D linear inverted pendulum model,an improved method of maximizing the single-support phase in the gait cycle is proposed.In the end the effectiveness of the method is verified by platform experiment.Thirdly,balance control.In terms of the problem which the robot would occasionally fall,the model is further optimized.And the Kalman filter algorithm is adopted to estimate the trajectory of the center of mass position,so the parameters that define the motion of the inverted pendulum can be re-corrected.Experimental results show that the gait method are much better than Aldebaran Walk Engine on dynamic and anti-interference ability.