An accurate and robust alignment method for wheel profile wear measurement with 2D laser sensors

Non-contact measurement using laser sensors for wheel profile inspection is commonly employed in railway infrastructure. However, two primary challenges arise: first, the deviation of the laser beam from the sensor’s axis, which leads to affine distortion in the measured wheel profile; and second, the interference from ambient light and contaminants, such as soil, on the wheel surface, which introduces ineffective points into the measurement data. To address these challenges, a robust scaling iterative alignment (RoSIA) method is proposed for wheel profile inspection. Initially, the Majorize-Minimization method is employed to construct a surrogate function for the registration objective. Second, the Welsch error metric is incorporated to enhance robustness against nonlinearities and outlier points. Third, a gradient-based optimization method is introduced to correct the affine distortion of the measured profile. The proposed method is comprehensively evaluated against conventional Iterative Closest Point (ICP) and Scaled Iterative Closest Point (SICP) algorithms in terms of accuracy, stability, and noise resistance. Field experiments confirm that the proposed RoSIA method performs well in both wheel shape calibration and noise resistance, demonstrating its potential for practical application in measuring wheel profile wear.