���ϳԹ���

Drones are increasingly being used for bridge inspections and to create 3D digital twins of the structures, but guidelines to replace in-person measurements of bridge geometry needed for structural assessment are still a work in progress. That’s where research by Akram Jawdhari and Ahmad Faqiri comes in.

They are part of the Department of Civil and Environmental Engineering at South Dakota State University. Jawdhari is an assistant professor in his fourth year at SDSU. Faqiri is a doctoral student who arrived in January after completing his second master’s degree at the University of Utah.

They have taken thousands of images of four concrete bridges in Brookings County and are working on data analysis. The goal is to have turned their research into a paper to submit to a peer-reviewed journal by the end of December.

Creating a precisely accurate 3D model from 2D images is filled with complications, said Jawdhari, who has studied bridge structures since his doctoral research at the University of Kentucky beginning in 2012.

Faqiri, who already had degrees from his native Afghanistan, Japan and Utah before coming to SDSU, also needed to learn how to fly a drone. For a man who speaks five languages, this was a fun challenge of a different sort. “I enjoyed watching drones being flown, so I’m glad I was able to learn that for myself,” he said.

After dedicating many hours to studying drone flight controls, safe operating procedures and federal aviation regulations, he passed the required online examination in May to become an FAA-certified drone pilot. This achievement reflects his commitment to developing his technical skills and operating drones safely and responsibly, Jawdhari said.

Creating a 3D model requires the computer to understand and be able to accurately measure each element of the bridge, Jawdhari said. The process is called photogrammetry, which means measuring with light. The basic idea is to photograph the same object from enough angles to allow the software to estimate where its surfaces sit in 3D space.

For the software to produce accurate results, a number of factors need to be considered. For example, wind affects the stability of the camera and clouds will affect shadows on the photo, Jawdhari said.

“We’re trying to find the best techniques to automate the process,” he said.

Faqiri said the drones take two passes over the top of the bridge, two passes along the sides, two passes from underneath and one pass by each pier to get a very accurate digital construction of the bridge. Jawdhari added that the photos must overlap on two sides so they can be stitched together to create an accurate model.

Mission planning needs to account for the number of passes needed as well as the number of cameras, Jawdhari said.

Two larger drones used in the research are equipped with RGB and thermal cameras. A smaller drone, which has a built-in camera, is used if Jawdhari and Faqiri want to send a drone to tights spaces with limited overhead, they said.

Jawdhari said he already is considering a follow-up project involving steel and wooden bridges, which are more complex than concrete bridges. “That’s where machine learning comes in,” he said.