Research On Human Robot Motion Transfer Based On Semg Signals

With the rapid change of social production and consuming styles,the traditional robots can't meet the requirements of high flexibility and dexterity in production,and the complexity of people's living environment.Therefore,the new generation of robots must be equipped with friendly human-robot interaction interface,which can quickly and adroitly learn various human skills,sharing the same workspace with human.And thus such robots can improve the efficiency and quality of production etc.This desertation proposed a natural and flexible human-robot motion-transfer based on surface Electromyography(sEMG)signals.Compared with the traditional robot motions,human limb motion has good flexibility,environmental adaptability and dexterity.These characteristics benefit from the long-term evolution.Therefore,the transfer of human robot motion includes kinematic and impedance information.Firstly,The current research of human limb impedance adaptation and compliant human-robot interaction were reviewed.Then the disadvantages of robot in the new trend of production and life styles were summed up and indicated that the next generation robots require more compliant manipulations.The propose and significance of the research proposals in this dissertation were revealed thereafter.Then the mechanism of motion of upper limb,such as impedance adaptation and impedance learning optimization were summarized,especially in combination with robot technology,human-robot motion skills transfer and so on.Then the design and control methods of robot joint with imitation of human impedance in recent years were reviewed.The analogy,analysis and key techniques of human robot interaction system based on human limb impedance regulations and related methods were summarized,which laid a solid foundation for further research.Based on the biological principles of human limb impedance adaptation,human upper limb impedance modeling and identification were carried out.The second-order system model was employed to identify the human 3D-Cartesian impedance,and the human limb impedance adaptation characteristics were verified.For the estimation of the impedance of the upper extremity of the human body,this paper systematically expounded the spatial transformation of stiffness from joint space to Cartesian space.sEMG signal smoothing envelope algorithm and human upper limb stiffness identification algorithm were proposed.The kinematics of the human arm was modeled based on the D-H coordinate method.The kinematic measurement of the human arm was realized based on the vision and inertial measurement unit.On-line identification of stiffness of human upper limbs were implemented based on human limb motion and sEMG signals.In order to realize the transfer of human upper limb impedance and kinematics information to the robot,a deep study were implemented to reveal the principles of variable impedance robot control.A robot impedance controller model was proposed,and a human-robot motion transfer interface based on human impedance adaptation was established.The robot can quickly imitate human limb movements via the interface.In order to improve the performance of human robot interaction,tactile feedback was employed in the comparative tests.Human-robot motion transfer interface based on sEMG signal was proposed.human's upper limb impedance and kinematics were extracted as the input to the interface.Comparative tests were carried out to verify the interface performance,such as human body joint anti-disturbance and robot adaptive anti-disturbance tests,robot writing skill learning,human-robot-human skill training,teleoperation and so on,which validated the research proposals.