The Grip Force Perception And Motion Hysteresis Compensation For Cable-Driven Surgical Instrument

The emergence of endoscopic minimally invasive surgical robots solves the problems that appears in traditional minimally invasive surgery,such as unintuitive operation,fatigue of surgeons,and inflexibility of instruments.Robot can reduce the pain of patients,improve the quality and efficiency of surgery,and has been widely accepted by surgeons and patients worldwide.However,the introduction of robotics has deprived the surgeon of intraoperative force perception.A number of studies have shown that the gripping force perception in the master-slave operation has a great impact on the surgeon’s operation.However,there is no surgical robotic system that can provide gripping force perception and feedback on the market so far.This paper has important theoretical and practical significance to study the gripping force perception of endoscopic surgical robot.The grip force perception method based on learning algorithm possesses higher accuracy and stability than the model-based method,and also avoids the compact design requirements and disinfection and sterilization requirements brought about by additional integrated force sensors.First,from the perspective of the actual application scenario and the characteristics of surgical instruments,the design method of the three elements of the grip force perception learning algorithm,including training set,input features and learning model,is studied.An experimental platform of cable pulley mechanism is built.Through the analysis of the actual application scenarios and the characteristics of the cable pulley mechanism of surgical instruments,the design points of the components of the learning algorithm are studied and summarized,providing a basis for the following research.According to the research conclusion of the analysis of the characteristics of surgical instruments,the grip force perception learning algorithm is studied and designed.First,a data collection platform based on real surgical instruments.Eliminate the influence and interference of surgical instrument specificity and motion compensation control in actual application scenarios,this chapter focuses on the learning of the relationship between end clamping force and sensor data mapping,design training sets and input features,and propose a light depth network CAM-FoC as the learning model.Ablation experiments and comparative experiments are carried out to verify the progressiveness and effectiveness of the CAM-FoC based grip force perception learning algorithm.In order to simplify the identification process and reduce the identification cost,a motion compensation method considering the motion hysteresis specificity is studied to solve the problem of surgical instrument specificity in large scale application.Based on the characteristics of motor current curve when the actuate motor is rotating at a constant velocity and reciprocating,and combined with the semantic segmentation algorithm,a method for surgical instrument hysteresis specificity identification is proposed.Based on the identification results and the prior knowledge of the hysteresis curve,a compensation curve generation method is proposed,and finally a motion hysteresis feedforward compensation method is proposed.Finally,the intrinsic error sources of the method are analyzed,and experiments are carried out to study the compensation error range of motion compensation methods when applied in large quantities.Compensation errors are compared with other methods.Combined with the previous research conclusions,the grip force perception learning algorithm is developed considering both the actual application scenarios and the characteristics of surgical instruments.First of all,through the actual collected sensor data curve,the requirements for the design of learning algorithm components for the motion hysteresis specificity of surgical instruments,the specificity of zero-load friction,and the compensation control of surgical instruments are proposed.Furthermore,the training set design,input feature design and learning model design are carried out based on CAMFoC,and the grip force perception learning algorithm based on Augmented CAM-FoC is proposed to solve the grip force perception problem in large scale application with motion compensation control.Ablation experiments and comparative experiments were carried out to verify the progressiveness and effectiveness of the Augmented CAM-FoC based grip force perception learning algorithm,and to verify the error of grip force perception.The method studied in this paper is actually applied to the self-developed endoscopic minimally invasive surgical robot system,and the master-slave control performance evaluation experiment,the gripping force perception and feedback experiment,the animal surgery experiment,and the clinical trials are carried out to verify the gripping force perception in this paper.The feasibility of the grip force perception method and the motion hysteresis compensation method and the effectiveness of the improvement of the master-slave operation are verified.The grip force perception method proposed in this paper does not need to integrate additional sensors on the cable-pulley surgical instrument or consider the requirements of compact design and sterilization,and has high practical application value.