Welding Robot System Based On Visual Guidance And Arc Tracking

The widely use of welding robot has played a significant role inpromoting the development of welding automation. In order to increase theintelligentize and flexibility of the welding robot, it should be equipped withsensors to obtain external information. Visual sensing has advantages ofgetting a large amount of information and high precision of recognition.Therefore, it can be used in the recognition of key points of welding seam.However, for the curvilinear seam which has variation in the height, thedifficult problem of stereo matching has to be solved while using visualsensing to obtain the information of height. In contrast, arc sensing is to usethe relationship of arc signal in welding circuit and the torch height to realizethe seam tracking in height direction, which has good real-time property andhigh stability. Thus, in this article an intelligentized welding robot systemwas developed combing the advantages of visual sensing and arc sensing. Itcould realize the preliminary path planning and the guidance of initialwelding position through visual sensing, and complete the seam tracking inheight direction by arc sensing. Hardware system was developed based on modular design includingmodule of welding robot, visual sensing and arc sensing. A special calibrationtarget full of circular dot and MALAB calibration program were used tofinish the calibration of camera and hand-eye quickly. Proper Imageprocessing algorithm was used to find the initial welding point and end pointin the image and get the pixel coordinate values. The’one eye-doubleposition’ mode was applied to complete the stereo reconstruction and achievethe preliminary path planning and guidance of initial welding point.Butterworth filters of different cut-off frequency were designed to do thewelding current filtering and the best cut-off frequency was determined. Analgorithm of getting the characteristic value of welding current was proposedand the relationship between the welding current and the torch height wasestablished. Furthermore, a PID controller was designed.A software system was developed based the hardware system includingcomputer program, robot program and DSP program, which had a friendlyinterface and simple operation. Based on this system, the experiment ofguidance and arc tracking was conducted. As a result, the precision ofguidance was less than0.5mm,0.4mm and1.0mm respectively in x, y and zdirection. The precision of arc tracking was less than1.1mm. Finally, awelding experiment of self-guidance and arc tracking on an arced workpiece was conducted. The result shows that the requirement of precision was met,which proved the reliability of the visual sensing and arc sensing.