Record Details

Data for "CAIT-UTC-REG65: Development of a geometric extraction framework as part of a pilot digital twin framework for open-deck rail bridges"

Harvard Dataverse (Africa Rice Center, Bioversity International, CCAFS, CIAT, IFPRI, IRRI and WorldFish)

View Archive Info
 
 
Field Value
 
Title Data for "CAIT-UTC-REG65: Development of a geometric extraction framework as part of a pilot digital twin framework for open-deck rail bridges"
 
Identifier https://doi.org/10.7910/DVN/9BMR20
 
Creator Najafi, Amirali
 
Publisher Harvard Dataverse
 
Description Open-deck railway bridges require expensive and customized timber sleepers. When these sleepers are due for replacement, a manual process for geometry measurement is necessary which can be time consuming, inaccurate, and expensive. In addition, significant downtime is required for the safety of the inspectors that measure and assess the conditions of open-deck bridges.

In this report, an alternative approach is proposed for geometry extraction of timber sleepers using unmanned aerial vehicle (UAV) inspections and use of artificial intelligence. First, a photogrammetric procedure for development of three-dimensional (3D) bridge models from UAV-based images is provided. Next, a deep learning-based algorithm for segmentation of 3D bridge model into recognizable components is described. Finally, a geometric primitive fitting algorithm is outlined for identifying the geometry of individual components. The aim for this development with 3D scans and automation is to reduce the maintenance and sleeper replacement procedure costs and challenges for open-deck bridges.
 
Subject Engineering
 
Contributor Stiesi, Ryan