Design And Interactive Control Of Colonoscopy-Assisted Robot

With the development of society,people ’s awareness of preventing colorectal cancer is increasing.Colonoscopy,as a standard for screening and diagnosis of colon cancer,has been increasing in demand in recent years.But at present,the quality of colonoscopy depends on the experience of doctors,who control the colonoscopy to observe the intestinal image.This method has problems such as high labor intensity of doctors,missed detection of polyps,and poor stability of long-term operation.In view of the above problems,this thesis introduces robot technology,AI assisted detection technology and human-computer interaction strategy to study a colonoscopy assisted inspection robot,which can effectively guarantee the quality of surgery and reduce the labor intensity of doctors.This thesis mainly carries out the following research.Firstly,the physiological characteristics of the colon and the principle of the use of colonoscopy are analyzed.According to the requirements of the use,the design scheme of the colonoscopy auxiliary inspection robot is formulated.Through the study of the driving mode of each part,a control device for controlling the colonoscopy auxiliary inspection is designed.Combined with the UR manipulator and the traction trolley,the overall structure of the colonoscopy auxiliary inspection robot is designed,the spatial coordinate frame of the propulsion part structure is established,and its working space is analyzed.Secondly,the difficulties encountered in the use of computer detection and recognition of colon polyps are analyzed,and the current target detection methods are analyzed and compared according to the requirements of polyp recognition in colon examination.A colon polyp recognition method based on YOLOv7 algorithm is studied and designed.The Py Torch deep learning framework is used to train the weight model for the preprocessed colon polyp data set,and its accuracy and real-time performance in polyp recognition are analyzed.It is verified that this method can help doctors identify polyps.Based on Py Qt5,the visual interface of colon polyp recognition is designed,which is embedded with colon polyp recognition algorithm and can display real-time images.The interface can be used in robot interaction system software.Thirdly,the interactive control of colonoscopy-assisted robot is studied.An interactive control method based on ROS system is designed to control the movement of colonoscopy-assisted robot by remote control handle.Firstly,the robot model under ROS system is established,and the control system of colonoscopy-assisted inspection robot is built.An interactive control mode based on Moveit＿Servo is proposed to realize the speed-based manipulator control logic.At the same time,the control algorithm is optimized,and a variable telecentric motion algorithm suitable for colonoscopy-assisted robot is designed.Finally,under the Ubuntu system,the interactive interface dedicated to the colonoscopy auxiliary inspection robot is designed and developed.The hierarchical processing mode is used,the function of the modular interactive interface is integrated,and the polyp recognition interface is integrated.The communication with the colonoscopy auxiliary inspection robot can be established through the interactive interface to realize the combination of the remote control handle and the interactive interface to complete the colon inspection process.Finally,the experimental platform of colonoscopy auxiliary inspection robot is built.The safety and rationality of the robot structure are verified by measuring the matching accuracy of the robot handle control end and the colonoscopy push end.In order to verify the real-time and accuracy of the polyp recognition method,the real colonoscopy video was used to complete the verification test of the colon polyp recognition method.The interactive control logic of colonoscopy-assisted robot is tested by remote control handle and interactive interface.The feasibility of this interactive method in colonoscopy and the effectiveness of variable telecentric motion algorithm are verified.