In view of the problem that in the production process of the oil cooling tube,the roundness error of the nozzle is mainly measured by manual measurement,and has the problem of low detection efficiency and inability to detect in real time,a method for roundness inspection and sorting of oil cooling nozzles was proposed based on visual inspection.The kinematics model of the robot SR7CL was established,ADAMS and MATLAB were used to simulate the kinematics and dynamics to verify the correctness of the model to ensure the accuracy of grasping.The oil cooling pipe was taken as the object to be captured,through adaptive threshold segmentation to highlighted the characteristics of the nozzle;the edge information was extracted by using the Canny algorithm,the roundness error was detected by using the least square method,and the cooling pipes were sorted based on the roundness error.The results show that by using the proposed method,the quality and efficiency of detection can be improved and the problem of real-time performance can be solved.