Research On Key Technologies Of Visual Recognition And Guidance For Automatic Wiring Robots

With the development of intelligent and complex electrical equipment,the internal or external circuits of electrical appliances also become more and more complex.In these circuits extending in all directions,cables and connectors are the basic units that make up each circuit.The connection of traditional circuits and electrical components is basically done manually,by fixing the connectors and circuit boards,and plugging them according to the planned route.Such a large amount of repetitive work is time-consuming and cumbersome,and requires more manpower to complete.Therefore,in many electrical appliance manufacturers,it is urgent to develop a set of automatic wiring equipment that can be applied to production lines.In this paper,the scheme design is carried out based on the actual needs of manufacturers,and based on industrial robots and machine vision technology,the key algorithms for visual processing of cable terminal identification,cable grasping,terminal pose estimation and adjustment,and terminal vision guidance are carried out respectively.The design is verified on the built-in industrial robot automatic wiring system platform.The main research work of this paper is as follows:(1)Recognition of cable terminals and calculation of cable picking poses.First,the robot,the follow-up camera and the coordinate transformation relationship between them are established through calibration,and then the YOLOv5 model is improved to improve the detection ability of cable terminals.The stereo matching of the local key points of the cable is used to obtain the pickup pose of the cable through the principle of binocular triangulation;(2)Angle estimation and angle adjustment scheme for cable terminations.Aiming at the uncertain posture of the picked cable terminal,which cannot be adjusted directly at the end of the robot,a solution for the angle adjustment of the cable terminal is designed,including the design of the corresponding rotation auxiliary mechanism,and a time-consistency-based method is proposed.The multi-view regression discriminant model can predict the deviation value of the cable terminal from the target state by taking the local image features of the cable terminal in the multi-view as input,and can adjust the terminal angle by rotating the auxiliary mechanism,so that the terminal has The attitude of entering the hole;(3)Wire insertion guidance system based on end visual servo.During the wiring process,since the specific pose of the cable terminal held by the robot end cannot be accurately obtained,a visual guidance system based on SVR without calibration is designed.According to the geometric characteristics of the target image and the requirements of the wiring task,the The distance between the cable end and the peripheral contour of the connector is the feature value as the image feature information.The non-object contour suppression method based on the local color probability histogram is studied to process the image,and the extraction of the connector based on the single-line scan histogram is proposed.The method of the peripheral contour completes the calculation scheme of the image eigenvalues.Secondly,the theoretical basis of support vector regression machine is studied and analyzed,and a visual mapping model based on SVR is designed by studying the mapping model representing the hand-eye relationship in the visual servo system;(4)The construction of the experimental platform and the testing and inspection of related algorithms.According to the design plan,an experimental platform is built,and the control program of the plug-in system is compiled.Then,based on the experimental platform,the cable terminal detection algorithm,cable picking,cable terminal angle adjustment and closed-loop vision guidance algorithm for wire insertion are experimentally verified.Through the above work,this paper completes the design of the automatic wiring system for industrial robots,realizes the design of key algorithms for visual detection and guidance of multiple wiring steps,and validates it by building a prototype.The work of this paper is expected to provide a solution for automated wiring for production lines.