Research On Key Technologies Of Electronic Component Sorting Robot Based On Vision

In recent years,the updating and iteration speed of electronic products is getting faster and faster.The scale of electronic waste is increasing rapidly,and the global ecological environment is facing more and more challenges.At present,about 70% of the world's electronic waste is concentrated in China for processing,and manual dismantling is still the main method in actual production.Backward recycling technology seriously restricts the recovery rate of rare and expensive materials,resulting in a large amount of waste of resources and huge pressure on the ecological environment.To melt in formal recycling companies apart as the main way of dismantling,but do not generally classified after dismantling melt processing directly,this way of handling the elements of the pertinence is not strong,the waste of resources,so how to electronic components after dismantling classification,to make more efficient use of resources is one of the major problems facing recycling enterprises.This paper analyzes the economic benefits and recycling scale of the electronic components market.In order to improve the sorting efficiency,reduce the waste of resources and improve the working environment of employees,the key technologies of the visually-based electronic components sorting robot are studied in depth.The main research contents are as follows:The conversion process from pixel coordinate system to world coordinate system is derived in detail by analyzing the imaging principle of the camera and applying the nonlinear imaging principle which is more suitable in practice.The distortion model of the camera was established.By comparing the advantages and disadvantages of MATLAB calibration and Open CV calibration as well as the final calibration results,the MATLAB calibration method with lower calibration average projection error was finally selected.The internal parameters of the camera were obtained according to the calibration results,and the image distortion was corrected by input internal parameters.In order to enable the manipulator to grasp the electronic components in 3D space according to the position of the electronic components in the image,an Eye-to-Hand hand-eye calibration model was established and its transformation relationship was determined.In the aspect of target recognition and the calculation of the position and pose of electronic components,the network structure and detection principle of the YOLO target detection algorithm based on the convolutional neural network are analyzed.The images of some electronic components are used as the data set of the training neural network,and the YOLO algorithm is used for recognition and calibration.An improved algorithm based on Sobel operator was proposed to extract the edges of electronic components in the ROI,so as to obtain a more complete edge image of electronic components,so as to determine the pose information of electronic components,and to determine the pose of the end gripper for the manipulator to grasp electronic components.The characteristics of the sorting manipulator are analyzed.The forward and inverse kinematics of the manipulator are solved by D-H method.The motion trajectory of each joint is planned by the method of polynomial interpolation of the seventh order,so that it can complete a series of actions.In order to improve the motion efficiency of the manipulator,the related trajectory optimization is necessary.In order to improve the motion efficiency of the manipulator,an improved quantum genetic algorithm is proposed,The algorithm introduces the normal distribution probability density function to improve the quantum gate rotation angle step strategy,which satisfies the requirements of continuous smooth curve of each parameter of robot joint,and can get the optimal joint trajectory curve on the basis of satisfying the nonlinear constraints,effectively shortens the movement time of the manipulator and improves the efficiency of the manipulator.A sorting platform composed of manipulator,control system,conveyor belt,opencv and CMOS camera is built to verify the effectiveness of the system.Through 450 times of sorting experiments on 9 kinds of electronic components randomly placed on the platform,the comprehensive recognition success rate is 98%,and the comprehensive sorting success rate is 93%.The experimental results show that the sorting system can meet the sorting requirements,improve the efficiency and reduce the damage of toxic chemicals to human body.