Study Of Gait Planning And Control Of The Humanoid Robot Based On Human Motion Capture Data

The premise of humanoid robots to achieve their activities is to have a stable gait.The gait research of humanoid robots includes two aspects of gait planning and gait control,in which the purpose of gait planning is to obtain a set of angle sequences for each joint of the lower limbs.The purpose of gait control is to modify the gait based on the relevant stability control strategy under the premise of gait planning.Human beings have the world's most perfect two-legged way of walking,the gait planning and control of humanoid robots based on human gait can not only reduce the difficulty of gait planning greatly,but also can obtain more human-like gait,and have broad application prospects.In this paper,the motion pattern of the robot is generated based on the human motion pattern,and the gait control is carried out based on the ZMP control strategy to realize the stable walking of the robot.The results include the following aspects:(1)Extracting human gait pattern based on human motion capture dataRealizing the extraction of human gait pattern.Pattern is the main way of the main behavior,the human gait pattern is the walking law of the human.Firstly,the D-H model of human skeleton was established,and the motion capture data of the.bvh format provided by the Carnegie Mellon University Graphics Laboratory was analyzed to complete attitude calculation of human body degree in the joint space,and the motion of the gait data in the joint space was generated which can be applied to the robot.And then the basic gait pattern is extracted based on the mathematical operations such as gait cycle segmentation and normalization,which lays the foundation for the generation of robot gait pattern.MotionBuilder and Matlab simulation results show the effectiveness of human gait pattern.(2)Humanoid gait pattern generation based on human gait patternRealizing the extraction of the robot gait pattern.Firstly,the simulation model of robot is built in Webots based on the DOF configuration of humanoid robot.There are structural differences between human body and robot walking model,and kinematics matching between human body and robot is needed.Therefore,it is analyzed between human body and robot.The kinematic matching between the human body and the robot based on the joint freedom degree constraint,the initial position adjustment of the humanoid robot and the key frame position constraint is completed.Finally,after the average filtering and Hermite interpolation,the gait pattern of the humanoid is extracted.Kinematic simulation experiments in Webots show the effectiveness of robot gait pattern.(3)Robot gait control based on ZMP stable performance index and gait patternRealizing the stability control of the humanoid robot.Firstly,the conditions of the humanoid robot's stable walking are analyzed,the ZMP position of the multi-particle and single-particle humanoid robot model is solved.Secondly,the robot real-time ZMP is measured based on the Forsing Sensor Registers which are under the humanoid robot's feet.Based on the ZMP control strategy,the hip joint of the robot is added with algorithm of searching optimization.The simulation results in Webots show that the robot's ZMP point is always in the stable region during the walking process,and the validity of the robot gait mode based on ZMP is proved.Finally,the humanoid evaluation of the robot walking process is completed.The similarity function shows that the gait similarity of robot and human is more than 70% in most of the time,which satisfies the human-like requirement of robot gait.