Uncertainty mitigation in drone-based 3D scanning of defects in concrete structures

Abstract

The measurement of defects in a concrete structures is highly relevant to determine how maintenance interventions should be performed. However, it could be difficult and potentially dangerous to inspect a certain structure by bringing trained operators, to places that are difficult to access. This issue could be overcome by framing the parts of interest of a building with a drone equipped with cameras. Nonetheless, a quantitative measure of a defect cannot be obtained with 2D cameras, since the pixel to millimeters scale and the estimation of depth are missing. To obtain a 3D shape measurement of a defect, 3D scanners, joined with 3D reconstruction, could be applied. In this article, we present a metrological evaluation of low-cost Time-Of-Flight (ToF) sensors for defects in concrete structures measurement. The defects of interest for this class of 3D scanners are mainly related to concrete spalling. This type of scanners was assembled on a drone with an onboard acquisition system. The testing benchmark for this study is based on a real structure with concrete spalling defects. A ground truth 3D model was obtained with a high-precision 3D scanner, used with a scaffolding. The effect of disturbances on measures were investigated, as well. The results of drone tests show that the systematic error of the 3D reconstruction with the selected sensors is about 0.5-2 mm, with a dispersion of raw data around the 3D reconstruction of about 2-4.5 mm, at a distance from the target of about 1.8-2 m.