| At present, teach-and-playback arc welding robots most used in practical production lack the ability to adapt to variations and must have stable working conditions. But the real welding process is a complicated process full of uncertain factors. So the study on the seam tracking based on visual sensor has a significant influence on improving the automation and intelligence level of welding robots.Based on the homemade"XINSONG"robot, this paper has designed and improved the passive vision-sensor system based on CCD. The pulsed GTAW welding experiment is done using the 4mm thickness LD10 aluminum alloys. Through lots of experiments and analysis, real-time and steady seam pictures are obtained by adopting an appropriate composted filter system combined with suitable sample time etc. which can overcome the strong arc disturbance to the images.The accuracy and reliability of the image processing algorithm is very significant since accurate acquisition of the center of the seam determines the realization of seam tracking. Based on seam images obtained, a complete image processing algorithm is developed, which contains image filtering, edge detection, removing false lines, picking up the edge, fining processing, edge fitting. This algorithm can acquire the seam center effectively and in real time.The mapping between the tungsten point position in the work piece and the angle of the 7th axis is described by a BP neural network. So the real time tungsten point position is calculated when the angle of the 7th axis is read through the main interface of the seam tracking system software.Based on the information of the seam center and the tungsten point position, the seam tracking controller is used to realize automatic seam tracking process controlled by the master computer by experiments. Since this kind of experiment is still on the exploratory stage, there is much improvement and research work to be done for the reliability and accuracy of the tracking.
|
PDF Full Text Request