Research On Motion Compensation Teleoperating Robot System For Dynamic Soft Environment

Since the development of robotics technology,robots provided significant assistance to human development in various fields as an extension of human senses and limbs.In the medical field,one class of surgeries,represented by heart surgery,is affected by the normal physiological activity of the patient in the surgical area.This physiological activity brings difficulties to the surgery and poses a significant challenge to the precision,physical strength,and mental state of the operating physician,resulting in significant surgical risks.Robot-assisted surgery technology provides a solution to such problems.The robot autonomously tracks the surgical site,creating a relatively static surgical environment,which greatly reduces surgical risks and lightens the burden on the operating physician.To compensate for the movement of the surgical site,calculating the desired position quickly and giving instructions to make the manipulator move in place.Introducing an advance by predicting to compensate for the delay introduced in each link of the entire process.The beating of the heart can be regarded as a quasi-periodic motion,which provides the possibility for predicting its position.But the interference introduced by other factors during the prediction process(such as chest deformation caused by respiratory motion)remains a significant challenge for motion prediction.Therefore,this thesis mainly studies the motion prediction problem under two sets of data: the heart beating data collected by the da Vinci surgical robot and the Phantom simulation heart beating data.Motion prediction is applied to teleoperated robots to achieve motion compensation and achieve good operating performance.The main contributions of this thesis are as follows:(1)A motion compensation teleoperating system under simulated medical background is proposed,and analyzes the reasons for tracking error in the system: spatial positioning,inverse kinematic solution,and robot arm response.If a motion compensation link is added,there is also a time delay in the prediction link.Using existing equipment to build a simulation control system,under the assumption of spatial positioning,the time lag introduced by each link is analyzed and statistically analyzed using the system’s operating results.Based on the predictive power of the autoregressive model,the performance of the long and short term memory network used for motion prediction and the time convolutional network not applied to related problems is compared.The application of the time convolutional network provides a new idea for motion compensation under this scenario.The above model is applied to the motion compensation problem.By analyzing and calculating,the number of steps that the model should predict is determined,and the model is used online to predict the position of the point of interest after motion compensation,demonstrating the effectiveness and accuracy of the model’s prediction in simulation experiments.(2)A complete motion compensation teleoperation system is built,using another robot arm to reproduce the motion process of the point of interest in the dataset,using a binocular camera to observe the motion of the point of interest,and performing spatial positioning to form a visual servoing system.The binocular camera can simultaneously observe the spatial relative position between the point of interest and the control point,providing feedback signals to enhance the accuracy of the controlled point’s motion.The effectiveness and accuracy of the system’s structure and active motion compensation are verified through physical experiments.