Enhanced precision in seam tracking sensors via adaptive refinement and fixed-geometry laser triangulation

Precise and reliable seam tracking is essential for ensuring weld quality and productivity in robotic welding, especially under challenging conditions such as spatter, smoke, and arc glare. This study presents a seam tracking sensor that combines an adaptive refinement algorithm with fixed-geometry laser triangulation to achieve high-precision weld groove detection on matte surfaces (Ra ≈ 12.5 µm). The adaptive refinement algorithm effectively suppresses distortions caused by welding noise, while the fixed-geometry triangulation framework computes three-dimensional weld coordinates (X, Y, Z) using pre-defined, invariant hardware parameters, without requiring any online or runtime parameter re-adjustment, A Douglas–Peucker-based weld point extraction method, applied for the first time in seam tracking, was optimized for V-groove geometries. Real-time experiments using a Cartesian robotic platform demonstrated mean tracking errors of 0.10–0.18 mm in Y and 0.13–0.22 mm in Z across six workpiece orientations, with 95–98% of points within ± 0.2 mm of the reference trajectory across all tested orientations. The system achieved an average processing latency of 12.7 ms, enabling integration into high-speed welding operations. These results demonstrate the proposed sensor’s robustness to welding-induced disturbances and its potential to enhance seam tracking accuracy in robotic welding applications under realistic operating conditions.