Research On Kinematics Simulation And Control Method Of Digestive Endoscopic Surgery Robot

There are many patients with digestive system diseases and high prevalence in our country.Digestive endoscopy is an important means of early screening and diagnosis and treatment of digestive system diseases.It is widely used in various foreign bodies or diseases of the digestive system.The endoscopy doctors are small in quantity and the workload is heavy.Traditional endoscopes do not conform to ergonomics,and it is difficult for doctors to perform terminal posture control.Environmental factors such as X-ray radiation and gastrointestinal fluid contamination of patients threaten the physical and mental health of doctors and affect work efficiency.In order to solve the above difficulties in reality,with the support of the National Key Research and Development Project of China—"Mixed Reality Assisted Digestive Endoscopy Robot Precise Minimally Invasive Treatment Technology Research",the research team proposed and constructed a digestive endoscopic surgery robot system.On the basis of the existing manipulator,this thesis research the kinematic modeling,image feature extraction technology,and image-based control method of the manipulator of the digestive endoscopic surgery robot during the intervention process.First,according to the structure and operation characteristics of the endoscope,the degrees of freedom of the endoscope are introduced,and the degrees of freedom allocation of the robot-assisted digestive endoscopy is completed.The kinematics modeling of the robot configuration was carried out using the D-H method,the workspaceof the manipulator was obtained,and the various joint parameters were further calculated through inverse kinematics.The MATLAB and ROS simulation verified the feasibility and flexibility of robotic interventional surgery for digestive endoscopic surgery.Secondly,the thesis explores the vision-based control technology of the digestive endoscopic surgery robot.The control problem in this research is attributed to the imagebased visual servo control problem of "eye on hand".By calculating the coordinate system transformation of the endoscopic camera and the image Jacobian matrix,a visual servo control scheme for the interventional surgery of the digestive endoscopic surgery robot is designed.On basis of the further research and elaboration on image feature extraction methods,the thesis proposes an improved oral anatomical structure recognition algorithm based on Mask R-CNN.Aiming at doctors check the procedure and the characteristics of the operation by digestive endoscopy in the stomach,an improved target tracking algorithm based on full convolutional neural network is proposed by merging detection framework and tracking framework.Finally,based on the existing experimental platform,the experimental scheme was designed.The thesis raised a remote image and command transmission system and constructed to obtain endoscopic images.Establishing the oral cavity data set and the gastrointestinal suspected lesion data set,the algorithm proposed in the thesis is trained and tested,and the comparative experiment proves the applicability and superiority of this algorithm to obtain the target features in the endoscopic intervention process.The existing surgical robot intervention process test was carried out and the robot arm condition of the control system was proved well.