PSI - Issue 62

Matteo Castellani et al. / Procedia Structural Integrity 62 (2024) 193–200 Matteo Castellani / Structural Integrity Procedia 00 (2019) 000 – 000

199

7

Fig. 7. Point cloud obtained from TLS process.

Before comparison procedures, the two clouds were cleaned by manually eliminating superfluous parts, such as vegetation and objects, and then by applying a SOR filter (Statistical Outlier Removal), that computes first the average distance of each point to its neighbors (considering k nearest neighbors for each - k is the first parameter) and then it rejects the points that are farther than the average distance plus a number of times the standard deviation. Hence, the UAV cloud was compared to the TLS one with the Cloud-to-Cloud tool, that computes the distances between two clouds, taking into account two central spans of the structure. As shown in Figure 8, it is clearly visible the absence of inconsistencies area depicted in red below the structure. The histogram of the differences (Figure 9) shows how the differences are shaped as a single-headed Gaussian function with an average value of 2 cm and standard deviation σ of 1.6 cm. To better interpret the results obtained, we consider a distance value of 3 cm between the two different clouds, noting that 77.50% of the points of the UAV cloud fall within this tolerance. In fact, the points that deviate the most from the reference cloud are found at the base of the first stack, where vegetation is present, and along the objects close to the structure, which do not affect the survey of the work. 6. Conclusion The prevalence of deteriorating bridges worldwide underscores the necessity for cost-effective and systematic inspections, along with health assessments, to address infrastructure concerns. Understanding the bridge's geometry, construction details, and structural defects is crucial, yet not always assured due to the absence of original design documents for the infrastructure. The paper proposed a new framework for integrating point clouds obtained from UAV photogrammetry and TLS surveys for reinforced concrete box-girder bridges. Methodologies were suggested to approach these specific structures, aiming to expedite survey timelines and optimize result quality. Specifically, the proposed methodology was tested on a structure located along a vehicular artery in central Italy. The comparison of the two different techniques demonstrates how the point clouds obtained from UAV photogrammetry exhibit significantly less distortion compared to those obtained from TLS, to achieve an average value of 2 cm and a standard deviation of 1.6 cm. The obtained results underscore the efficacy of the suggested approach in systematically gathering crucial data for asset understanding. They also indicate that UAV photogrammetry has the potential to largely substitute traditional and costly laser scanning in the structural risk assessment process for bridges. This can be advantageous for infrastructure management entities as it drastically reduces acquisition and processing times, leading to economic savings and increased safety for operators.