Research On The Key Technology Of Visual Guidance In Robot Assembly Application

The vision-guided robot technology means that the robot is matched with the vision,which enables the robot to have a pair of "sharp eyes",strengthens its ability to identify things in the outside world,so that the robot's work efficiency,accuracy and reliability are improved.The vision system completes the inspection,tracking and positioning of the assembled parts,and transmits the acquired position information to the robot,so that it can complete the assembly task more automatically and flexibly.Therefore,how the two "sharp eyes" of the robot can better guide its "hand" to achieve high-precision assembly will be the core competitiveness of the future component assembly technology.This paper focuses on the automatic flexible assembly of incoming parts in industrial production lines,and carries out research on the key technologies of visual guidance of robots in assembly applications.The main results are as follows:(1)The comprehensive design of the overall scheme is realized.The parallel robot is selected and the selection and control methods of the core components in the vision subsystem and motion control subsystem are determined.(2)According to the classical assumption of Tsai radial uniform constraint,on the basis of improving the traditional ways of the two-step calibration method,we present the camera optical center calibration method which is based on the central undistorted region to solve the principal point coordinates and overcome the traditional hypothetical principal point coordinates.Then,a polynomial distortion model is established,which eliminates the limitation of the Tsai two-step method and considers only the radial distortion of the camera lens and completes the hand-eye calibration.(3)Based on the classical moving target detection method,a moving target detection method combining adjacent frame difference and background differential sequential processing is introduced.This method can complement the advantages and disadvantages of adjacent frame difference method and background difference method to improve the detection efficiency.Then,based on the filtering algorithm,the position of the moving target part and its velocity characteristic data are predicted.(4)The kinematics analysis of the robot is carried out,and the assembly motion control optimization is completed on this basis.Then,the assembly application experiment platform of visually integrated robot is built,and the joint debugging experiment of software and hardware is carried out to verify the accuracy of the vision-guided robot to detect,track,and assemble the pose in the assembly application.which has important reference significance in the practical engineering application.