Research On Optical Component Sorting System Based On Monocular Vision

Aiming at the phenomenon that traditional manual sorting not only has low efficiency and dangerous working environment,but also may cause secondary pollution to precision workpiece,a robot control technology based on machine vision is proposed for workpiece sorting.A further research on the position accuracy and efficiency of visual sorting system is carried out.The research involves camera calibration technology,kinematics calibration technology based on visual position detection and sorting path planning based on visual processing.Combining H-shaped parallel robot,A fast and accurate visual sorting of circular optical glass elements is realized,and a good visual sorting system is formed.Firstly,according to the overall functional composition and requirements of the visual sorting system for circular optical glass elements,the overall design scheme is formulated.the main hardware of the visual processing unit and the motion control unit is selected and designed,and the development scheme of the software system is determined by the mixed programming mode of Halcon,MATLAB and MFC.Secondly,aiming at the position accuracy of visual sorting system,a linear camera calibration model based on image principal points is constructed,and the model is optimized by Levenberg-Marquardt algorithm to form a complete camera model.Based on the results of camera calibration,position error detection of H-shaped parallel robot is carried out,and parameter identification and error compensation are combined with the error model of the robot to complete kinematic calibration of H-shaped parallel robot.The calibration results show that the camera model has good accuracy,and the kinematic calibration process is convenient and fast,which the effect of position accuracy improves obviously and conforms to the actual use process.Then,aiming at the efficiency of visual sorting system,the sorting path planning based on image processing technology is studied,and the visual sorting workpiece problem is transformed into a constrained TSP problem.A hybrid optimization algorithm combining ant colony algorithm and tabu search algorithm is proposed.The positive feedback mechanism of ant colony algorithm provides a good initial solution for the tabu search algorithm in the number of effective iterations.The optimal solution is obtained by the memory ability and the scorn principle of the tabu search algorithm.At the same time,according to the analysis of the problem,the hybrid optimization algorithm is improved adaptively,and the ant colony algorithm operator is optimized.On the one hand,2-opt local optimization and pheromone punishment-reward mechanism are introduced to improve the ant’s search ability,on the other hand,information volatile factor is improved adaptively to improve the ant’s adaptive ability.The simulation results show that the hybrid optimization algorithm can optimize the sorting path stably and effectively.Finally,the experimental platform of visual sorting system is built,and the testing software is developed.Platform comparison experiments are carried out in terms of position accuracy and efficiency.The experimental results show that the efficiency of the visual sorting system,which use hybrid optimization algorithm to control sorting path and guide the robot after kinematics calibration to complete corresponding actions,has been effectively improved and the position accuracy is high.