A real-time weld line detection for derusting wall-climbing robot using dual cameras

The real-time tracking of a weld line for the removal of rust can significantly improve the derusting efficiency and autonomy of a wall-climbing robot, wherein, weld line detection and identification is very crucial in order to realize weld line tracking. To solve this problem, the present study first established a seam tracking system based on dual cameras with auxiliary LED lights to present real-time image processing algorithms. A continuous recursive algorithm was then proposed to remove the burr of the edge of the weld line following image thinning. Furthermore, the identification rules of the weld line and the decision strategies of the final deviation were investigated. Finally, verification experiments were performed, wherein the experimental results indicate that the average processing time of a single weld image is less than 60 ms such that the success rate of weld line identification based on dual cameras reached 100%, thereby validating the reliability and effectiveness of the proposed detection methods for the cleaning process of the wall-climbing robot.