The Research Of Welding-seam Tracking By Robot Sense Of Vision For Oil Derrick

This task derives from a research of "Intelligent Welding-Robot for oil derrick" project cooperated with NanYan Second Oil Equipment Company, in order to realize completing the huge welding workload, also improving efficiency and quality by welding-robot instead of manual. Because of the complex structure of the oil derrick, it maybe appear some mistake about weld trace when the robot working, the Intelligent Robot is required to track the seam timely, and correct the weld trace independently. So the machine sense of vision is an important component of automatic welding by Intelligent Welding-Robot.In this paper, we have discussed the framework and working theory of the machine sense of vision system, designed a steady and reliable mechanical structure with the same precision at low price, designed the control system of the Welding-Robot, and also realized the teach-redo programming system and off-line programming system for the robot.Seam tracking system includes three parts: video input, image analysis and weld track adjusting. The video images captured by CCD camera are transmitted into the central control computer through USB interface. In this paper, we introduce DirectX technique to the machine sense of vision, which is a new way to capture video images. Because of high speed and high quality of the video stream in images capture and convenient COM condition, the DirectX technique will paly an importent part in machine sense of vision system.We make a research and compare between various image disposal and identification arithmetic, and take a seam image which has much noise as an example to analyze. In this research, we choose a component of image disposal and identification arithmetic with fastest speed to realize and pick-up the characteristic of weld seam out of the image, it is the necessary condition for seam tracking timely.The final purpose of image disposal is to get warp information of seam, and then correct the moving track of the welding torch. In order to get the appropriate control value of track control, we take a research of the contour of the weld zone, and use curve-fitting method to get the information of the true seam track, so as to get the value of seam warp truly. Because this arithmetic was very simple, this method could improve the disposal speed of seam track processing, and fit for timely seam tracking system.