Study On Seam Recognition And Tracking With Robot

Modern industrial production has a new requirement of the welding technology from the cost and quality point of view,thus manual welding is gradually unable to meet the demand of production process.At the same time,the work environment of manual welding is bad and harmful to human health.In this context,the automation and intelligence of welding become a trend.Aiming at the problem that the weld seam of the car intercooler is deformed due to the thermal expansion and contraction during the welding process,this paper presents a seam recognition and tracking system based on visual teaching and correction,which mainly consists of visual sensor,image processing system,robot control system,and so on.On the basis of analyzing the principle of using the line laser to detect the seam,a visual sensor is designed,which is formed by a camera with a line laser installed at an particular angle,to obtain a clear image of seam and light band.Using the vision sensor and the robot to construct a machine vision system,which is to be calibrated to obtain the parameter matrix for use.After obtaining the weld image,the ROI is needed to be extracted firstly to reduce the amount of data processing.And then a series of preprocessing of the image like Gauss filter,adaptive binary,morphological expansion is needed for obtaining a good image of the laser band.Finally,using Hilditch thinning algorithm to extract the center line of light stripe and using Hof transform to detect straight lines in an image.The curved part of the straight line is the weld point,so the welding seam can be obtained by getting a serious of weld points from multiple images.After that,rectifying the deviation of teaching trick and sending the deviation to robot controller,the correction to the weld can be realized.The experimental system was established to apply seam recognition and tracking experiment to test weldment,and the results and precision of the tracking were recorded and analyzed.The experimental results show that the system has a good recognition and tracking ability to meet the actual demand,and prove the effectiveness and feasibility of the method this paper presents.