Using LiDAR technology for bridge structural health assessment in smart city infrastructure

Amid the growing demand for smart city infrastructure management, bridge structural health assessment urgently requires high-precision, automated technical support. LiDAR technology, with its advantages of high resolution and non-contact data acquisition, has emerged as an important tool for multi-temporal three-dimensional monitoring of bridges. Addressing the issue of insufficient robustness of traditional point cloud registration algorithms in complex conditions such as uneven density and non-rigid deformations, this study proposes a Feature-Guided Multi-Scale ICP algorithm (FGMS-ICP). This algorithm introduces FPFH local geometric features, a multi-layer pyramid structure, and local rigidity constraints to enhance registration accuracy and deformation recognition capabilities. Experimental results demonstrate that the proposed algorithm outperforms methods such as GICP and NDT in terms of registration error (RMSE). For Bridge A under loaded deformation conditions, the RMSE is reduced to 3.21mm, with the average error of deformation detection controlled within 2.16mm and a deformation recognition rate exceeding 98%, with a stable runtime of less than ten seconds. FGMS-ICP exhibits significant advantages in accuracy, stability, and adaptability, providing a more efficient and reliable technical pathway for bridge structural health monitoring.