Laser scanning-based weld seam parameter extraction for slotted V-groove welds in shipbuilding

To address the requirements for efficient and high-precision extraction of weld-seam parameters for slotted V-grooves in automated welding of small and medium-sized shipbuilding subassemblies, a laser point cloud-based parameter extraction algorithm is developed for both V-groove welds with a root face and butt-joint V-groove welds. First, outliers are removed using a percentile-based method, and the point cloud is separated into front- and side-view subsets by exploiting the neighborhood range of the Z-coordinate. Subsequently, a rectangular-plane recognition algorithm and a boundary detection algorithm are formulated according to the geometric characteristics of the two weld types, enabling accurate groove point-cloud segmentation and precise boundary localization. Finally, adaptive extraction of the groove angle is achieved by integrating RANSAC-based line fitting with an endpoint method. Experimental results demonstrate that, for V-groove welds with a root face, the mean absolute errors (MAEs) of the groove angle, groove length, root gap, and plate thickness are 1.52°, 2.76 mm, 0.22 mm, and 0.75 mm, respectively; for butt-joint V-groove welds, the MAEs of the groove angle, groove length, and plate thickness are 0.41°, 0.96 mm, and 0.76 mm, respectively. Compared with two representative existing weld-parameter extraction algorithms, the detection accuracy is improved by 79.2% and 74.5% for V-groove welds with a root face and by 78.5% and 44.1% for butt-joint V-groove welds, respectively, demonstrating the engineering applicability of the proposed algorithm to shipbuilding automated welding.