Research On Prediction Of Human Arm Motion And Collision Avoidance For Surgical Robot Teleoperation

With the continuous improvement of the medical system,how to solve the uneven distribution of medical resources and improve the operation quality of grassroots community hospitals has become an urgent problem to be solved in the field of medical technology.The development of intelligent robot technology will provide a feasible way for it.Due to the complex operating environment,variable tasks,strong emergent,and the constraints of computer,artificial intelligence and other technologies,it is difficult to achieve fully autonomous robots to replace doctors to complete surgical tasks.Therefore,the solution of surgical robot teleoperation came into being.Supported by the project "Research on Nfotion Planning of 7-DOF(degree of freedom)Human-machine Cooperative Mechanical Arm for Complex Environment",this paper established the remote operation system of surgical robot and the prediction model of human arm motion of slave patients,learned the law of human arm motion,and predicted the current motion state of human arm according to the lagged data of human arm from slave end.Virtual reality technology is used to predict and display the 3D(three dimensions)graphics of the current motion of the human arm at the primary end to overcome the influence of time delay on the teleoperation system of the surgical robot.Firstly,this paper analyzes the motion mechanism of the human arm and establishes the 7-DOF kinematics model of the human arm.The forward kinematics of human arm is calculated and the closed solution of inverse kinematics of human arm is obtained by using arm angle parameters as redundant parameters.Using a recurrent neural netw ork(RNN)to study the relationship between human arm hand trajectory and time,hand trajectory prediction model is set up,usitig a multi-layer perceptron(MLP)learning the appearance of parts ot the hand and arm angle mapping relagionship,to establish personification arm configuration model,based on the arm Angle of inverse kinematics of the hand trajectory prediction model is combined with personification arm configuration model,Complete the human arm motion prediction model based on kinematics.Then,the recurrent neural network was used to learn the relationship between the position of shoulder,elbow and wrist joints and time,and the prediction model of human arm motion based on joint position was established.The advantages and disadvantages of the two prediction models were compared and analyzed.The comparison results show that the prediction model of human arm motion based on kinematics is more suitable for the prediction and display of the slave scene in the teleoperation system of surgical robot.Secondly,the teleoperation system of the surgical robot was built,and the master-slave control of the teleoperation of the surgical robot was realized by the incremental variable proportional velocity mapping method,and the force presence of the main end was established by the joint torque sensor of the manipulator.The teleoperation network communication link of surgical robot based on MQTT protocol was built,and the network delay in different time periods was tested.For network time delay of the effect of surgery robot teleoperation system security,combining with the actual video feedback with one arm motion prediction based on virtual reality show,a surgical robot teleoperation based on arm motion prediction method of collision avoidance,reduce time delay influence on vision telepresence teleoperation robot operation,increasing the safety of the teleoperation.Finally,the training data acquisition method of the human arm motion prediction model is introduced in detail,and the performance evaluation and simulation experiment of the human arm motion prediction model based on kinematics are carried out.The experimental platform of surgical robot teleoperation was built,the master-slave control experiment of surgical robot teleoperation system and the collision avoidance experiment of surgical robot teleoperation based on human arm motion prediction were carried out.Experimental results show the effectiveness of the proposed method for predicting human arm motion and avoiding collision by teleoperation of surgical robot.