Precise initial weld position identification of a fillet weld seam using laser vision technology

Full autonomy is the ultimate goal in robotic welding. Heretofore, to weld a fillet weld seam formed by two steel plates, the initial weld position had to be found by procedures that involved some manual steps. In this paper, the authors propose a completely autonomous method to find the initial weld position for a fillet weld seam formed by two steel plates. This method employs an automatic dynamic programming-based laser light inflection point extraction algorithm. This algorithm needs no information provided about the region of interest (ROI), even when the inflection point is near the border of the image. The algorithm for this method can overcome unstable factors induced by natural light (e.g., a strong light reflection, a light intensity difference), which may be present during the processing of laser vision images. An auxiliary algorithm to compensate for a time delay is also presented. The algorithm proposed in this paper has been validated in actual industrial environments and can satisfy any real industrial application requirements in terms of efficiency and reliability. With this algorithm for the initial weld position, the entire robotic welding process for a fillet weld seam formed by two steel plates can now be fully autonomous.