Fast deformation assessment of riveted steel railway bridges during load testing using unmanned aerial vehicles (UAVs)

Static load tests are commonly performed as part of bridge maintenance plans, but require considerable human and financial resources, placing a substantial burden on national budgets. In this context, this work introduces a novel methodology for remote displacement assessment of riveted steel bridges using Unmanned Aerial Vehicles (UAVs), aimed at reducing costs and expediting structural assessments during load tests without on-site instrumentation. By employing UAV photogrammetry to generate digital models at different test phases and using machine learning to identify rivet positions, the method estimates vertical deflections along the bridge’s length by comparing rivet locations before and after deformation. The proposed methodology was tested on a real bridge in Toledo, Spain, achieving millimetre-level resolution in the point clouds and demonstrating close agreement with deflections recorded by on-site potentiometers. These results represent the first documented application in the literature of UAVs without external instrumentation for assessing the vertical deflection of riveted steel railway bridges. • Deflection assessment is conducted by a single drone pilot with minimal risk. • A proper flight planning achieved millimetre resolution in the obtained point clouds. • A machine learning approach has been developed for automated rivet identification. • The proposed approach yields deflection values comparable to on-site potentiometers. • The proposed methodology can operate without graduated scales and on-site markers.