Research On Assembly Technology Of Nuclear Rod Based On Uncalibrated Visual Servoing

As a basic energy source,nuclear power industry is relatively backward in the production level and equipment of fuel rod components.When being assembled,the task will be difficult to complete because of its slender structure characteristics.The important direction of automatic assembly is to optimize the control of the manipulator based on the vision.The robot visual servoing is an important part of robot research field.Its main core is the difficulty of processing the feedback information of the visual system is to establish the mapping relationship between the robot and the visual system.In this paper,the image Jacobi matrix is used to describe the mapping relationship between the two,and a fixed camera arrangement is used.The image Jacobi matrix is composed of points and angles as image features respectively.A robust information filtering recursive estimation algorithm is used to update the image Jacobi matrix online.The main work is as follows:In view of the structural characteristics of thin and long deformation of the fuel rod,the Ansys mechanics analysis is carried out to determine the best position of the clamp,and a pneumatic flexible and flexible special machine claw and pneumatic control loop are designed to achieve its grasping task and assembly operation,so as to avoid the failure or cause of the assembly task due to the position coordinates error and the error of the instruction,leading to huge stress causing damage to the robot.When extracting point and angle in image processing,considering adaptive defect of the Canny operator,the problem that the smoothing parameter and the high and low threshold of the Gauss filter can not be satisfied,the bilateral filtering is used instead to enhance edge preserving function,and then the edge information is enhanced with the wavelet transform.The optimized Canny operator improves the image quality by increasing the mean,variance,entropy and average gradient parameters of the image,and realizes the denoising and edge enhancement effect of the image,which lays the foundation for the online estimation of the subsequent Jacobian matrix.Six degree freedom visual servoing control method based on image uncalibrated fixed binocular manipulator is proposed.The image Jacobi is estimated online by the robust information filtering.Through the hand eye mapping provided by the image Jacobi,the desired image location is driven by image feature deviation,and the manipulator is guided to locate robot’s position in the task space.Based on the robot toolbox under the Matlab environment,a visual servoing simulation platform is built,comparison test based on image based uncalibrated manipulator is designed,and the advantages and disadvantages of different image Jacobian estimation algorithms for the task assembly are compared.The physical experiment is carried out.The simulation results show that the robust information filtering estimation image Jacobian has the advantage of speed and accuracy.At the same time,the feasibility and effectiveness of the uncalibrated servoing control based on the image are verified by the physical experiment.The software monitoring platform based on the MFC framework of VC++ displays the assembly progress and the error warning in real time.