Research On Human Arm Motion Prediction And Humanoid Motion Planning Of Manipulator

With the development of the times and the progress of science and technology,the robot technology is becoming more and more mature.Manipulator don't just work in factories,more and more manipulators are used in human-robot cooperation scene to cooperate with people in the shared space to complete the specified tasks.In the process of human cooperation,when human can understand the intention of partner movement,the cooperation effect will be greatly improved.Similarly,for safe and efficient human-robot cooperation,it also requires that human and manipulator can understand each other's movement intention.Therefore,this paper proposes a arm motion prediction and humanoid motion planning method of manipulator to achieve the purpose of human and manipulator understanding each other's motion intention.Finally,experiments are designed to verify the human arm motion prediction method and humanoid motion planning of manipulator method.The main contents of this paper are as follows:1.Aiming at the problems of low accuracy and imperfection of hand trajectory prediction in human arm reaching point motion.The Cartesian components of the hand motion trajectory are fitted by fifth-order polynomial fitting method.The starting and ending points of Cartesian position components of hand motion are taken as input(6-D),and the fifth polynomial coefficient of each position component as output(6-D),and the prediction model of the Cartesian position components of the hand reach point based on Multiple-Output Gaussian Process Regression is established;The starting and ending points of Cartesian orient components of hand motion are taken as input(6-D),and the fifth polynomial coefficients of each Cartesian orient component are taken as output(6-D),and the prediction model of the Cartesian orient components of the hand reach point based on MOGP is established;Finally,the prediction model of the hand motion trajectory based on MOGP is obtained in hand reaching point motion.2.Aiming at the problems of low accuracy and poor universality of arm configuration prediction in human arm reach point motion.The arm 7 freedom(DOF)kinematics model is established,and the arm angle is selected as the redundant parameter.The influence of the upper and lower arm length on the arm configuration is analyzed,and the hand pose and the upper and lower arm length during the reaching point movement are taken as the input,and the arm angle corresponding to the pose is taken as the output,the prediction model of arm configuration based on Single-Output Gaussian Process Regression is established in human arm reaching point motion.3.Aiming at the problem of low degree of humanoid motion of humanoid manipulator,a humanoid planning method based on arm motion mapping is proposed.The model of the motion trajectory prediction of the hand based on MOGP and the prediction model of arm configuration based on SOGP are mapped to the Cartesian space and joint space of manipulator respectively.Then the analytical inverse kinematics equation of manipulator based on arm angle is established to obtain the joint angle of humanoid motion of manipulator and realize the humanoid motion of manipulator.4.The platform of virtual digital human and humanoid robot is built,and comparative experiments are designed to analyze the model of the motion trajectory prediction of the hand based on MOGP and the prediction model of arm configuration based on SOGP.The results show that the prediction performance of the proposed method is significantly better than other methods,and the arm motion of virtual digital human can be successfully reproduced.The human arm movement in handshake was compared with the humanoid movement of manipulator,the results show that the humanoid motion planning method can realize the humanoid motion of manipulator in Cartesian space and joint space.