Research On Automatic Welding Technology Based On Binocular Vision

With the proposal of "made in China 2025",the automation level of industrial welding has been continuously improved,and the development of welding automation,flexibility and intelligence has become an inevitable trend.Compared with traditional welding technology,automatic welding technology based on robots is being widely used in industrial production.Existing industrial welding robot has welding track deviation,which affects the quality of welding.The reason is that the welding robot lacks feedback adjustment when the welding parts have production and assembly errors or the real-time welding environment changes.In order to solve this problem,an automatic welding robot with visual feedback was developed to make it capable of automatic track tracking and real-time adjustment during welding.Based on MCU technology,a robot control system is designed.The image of welding seam is processed by Matlab software in PC,and the 3d coordinates of target point are calculated.Then the target point 3d coordinates are obtained and the welding motion is performed.In the study of the binocular vision feedback system,the improved Tsai two-step method was used to obtain the internal and external parameters of the camera,so as to complete the calibration of the camera.Moreover,only position relation exists between hand and eye of the robot,which simplifies hand and eye matrix transformation and speeds up hand and eye calibration process.According to the arc wavelength and intensity of tig welding,the wavelength and power of line structured light were determined,and then suitable filter and attenuator lens are used to deal with arc interference,avoiding high intensity arc radiation to blind the robot vision.Filter algorithm is used to eliminate noise interference for the collected images,and combined with threshold processing,edge detection and other algorithms to obtain the target point of the weld irradiated by the line structure light and the target point of the welding pool.The 3d coordinates of the target points are obtained by fastly matching the target points in the left and right eye images,and the deviation of the solder joint coordinates is calculated and analyzed to guide the welding robot to realize real-time adjustment.The experiment proves that the welding robot with visual feedback control has real-time adjustment function and improves the stability and accuracy of the automatic welding robot.