Bridge Deck Restoration Using UAV and 3D Photogrammetric Modelling

Abstract

Purpose. The study aims to develop a three-sortie methodology for employing the DJI Air 2S unmanned aerial vehicle (UAV) for comprehensive remote inspection of damaged railway track on bridge structures, verification of three-dimensional spatial model accuracy, and formation of an algorithm for quantitative damage assessment and calculation of the materials, equipment and personnel required to carry out restoration works. Methodology. Based on an analysis of relevant sources, a five-stage remote inspection methodology is proposed whose key element is a three-sortie aerial photography scheme: a photogrammetric sortie (altitude 25 m, overlap 85/80%), a close-range inspection sortie (1.5–2.0 m above the surface, GSD 0.57 mm/pixel), and an independent verification sortie (altitude 35 m). The acquired imagery was processed using the Structure-from-Motion/Multi-View Stereo method in Agisoft Metashape Professional, generating a dense point cloud, orthophoto and digital surface model. Damage analysis was performed in CloudCompare and Agisoft Metashape using the Difference of DEMs (DoD) approach and YOLOv8 algorithms for automatic defect classification. Findings. For a notional single-track railway bridge with a 45 m span, nine categories of damage were identified, including ballast layer destruction (26.4 m²), cracks in beams (83 units), concrete spalling with exposed reinforcement (8.6 m²), sleeper damage and rail deformation. Spatial model accuracy was RMSE = 1.7 cm in plan and 2.4 cm in height. Total flight time for the inspection was 63 minutes; data processing time was 4–5 hours. Requirements were calculated for 10 material items, 7 equipment units and 22 personnel; the normative restoration period is 36–48 hours in a two-shift regime. Originality. For the first time, a three-sortie aerial photography scheme with an independent verification sortie is proposed, providing self-monitoring of model accuracy without ground-based laser scanning; a closed management cycle “detection → metric assessment → resource planning” for railway track on bridge structures has been developed. Practical value. The proposed methodology can be introduced as a standard for primary inspection of damaged bridge crossings in technical regulations of railway and engineering troops, while the calculated material and equipment requirement tables can be directly used for compiling supply requests for repair-and-restoration units.