Design And Research Of A Robot Prototype For Oral Examination Based On Machine Vision

For the traditional diagnosis of oral diseases,although the artificial visual examination method has high accuracy,it is limited by professional doctors and equipment,and the regional differences are very obvious.In response to the above situation,the paper proposes a prototype design scheme for an oral examination robot based on machine vision.According to the traditional oral examination process and requirements,the mechanical system,control system,and visual system of the entire machine have been studied,designed,and experimentally analyzed,respectively,to verify the feasibility of using the oral examination robot in the paper for oral disease detection.The main research contents are as follows:(1)Aiming at the mechanical system of the oral examination robot,a tabletop four-degree of freedom manipulator is designed as the main structure.The standard D-H method is used to model the manipulator part,and the forward and inverse kinematics analysis,workspace analysis and trajectory planning are carried out with Matlab to verify the rationality of the design.(2)Aiming at the control system of the oral examination robot,the control idea of taking the computer as the upper computer and the Arduino single chip computer as the lower computer is designed,and other hardware can meet the needs of image acquisition,visual positioning and recognition detection.The basic motion of the manipulator is realized by the face tracking PID algorithm,and the status of the manipulator is monitored and controlled by the position sensor and the pressure sensor to ensure the integrity and safety of the oral examination robot when working.(3)Aiming at the mouth positioning of the oral examination robot,the coordinate system conversion relationship between the camera,the manipulator and the three-dimensional space is analyzed,and the camera calibration and hand-eye calibration are carried out with Matlab,and the system parameters are determined.Based on the Dlib model,the recognition and judgment method of face mouth state is designed,and the 3D coordinates of mouth center are extracted according to the characteristics of depth camera.Through the repeated positioning experiment of the manipulator,the positioning error in three directions in space is within ±2mm,which verifies the feasibility of the mouth positioning method in this paper.(4)Aiming at the disease recognition of oral examination robot,an oral disease detection model based on YOLOv7 target detection algorithm is constructed.Through the collection and enhancement of the oral disease image,the construction of the data set is completed,and the evaluation of the oral disease detection model after training is carried out.The Map＿0.5 is88.55%,and the FPS is 31.29.The accuracy rate is 90.69% after the validation set test,which verifies the feasibility of the oral disease detection method in this paper.(5)Aiming at the overall function of the oral examination robot,the human-computer interaction interface is designed,the online and offline examination modes of the oral examination robot are tested in real person,and the oral pathology model is used to supplement the simulation experiment.The experimental oral examination robot system can achieve the expected examination effect,and the reliability of the whole machine is verified.