New AI-Enabled Technology Quickly Identifies Bridge Defects - Sachse Construction
Topics: Architecture & Design, Construction, Industry News

New AI-Enabled Technology Quickly Identifies Bridge Defects

The standard approach to identifying structural deficiencies in bridge decks is decidedly low tech. The time-intensive “sounding” process requires ground crews to drag a chain across the surface of the road to listen for hollow-sounding fluctuations and watch for vibrations. Often requiring road closures for days or weeks at a time, the technique creates headaches and hazards for motorists, road workers and inspectors.

Two companies aim to bring the bridge inspection process into the 21st century with a new system that employs drones, artificial intelligence and algorithms to quickly spot defects and problems. The creators of the patent-pending methodology estimate it can help transportation projects slash up to 70% in man hours compared to traditional methods, in less than half the time.

The program, a joint project of Kansas-based AEC firm GBA and AI provider Dynam.AI, records overhead images shot from an unmanned aircraft hovering about 200 feet above the bridge. Equipped with FLIR infrared thermography cameras, the DJI-made drones evaluate the condition of concrete decks, overlayments and bridge joints.

The camera’s infrared sensors can detect electromagnetic radiation and heat fluctuations that indicate areas of stress, said Ben Lindner, advanced robotics and remote sensing group leader at GBA. The aim of the surveillance, which requires a trained operator and a spotter who keeps an eye on surroundings, is to detect compromised areas before potholes develop and further erode the structure.

Once the images are captured, Dynam.AI’s Auguste algorithm then converts the findings into data streams that detect areas of decay with greater than 85% accuracy. The combined technology can detect problems unseen by the human eye, said Lindner, and can also convert images into 3D models to help visualize anomalies.

“A human can detect a pattern of temperature changes, but by using a computer vision system you can get down to the pixel level,” he said. “It’s kind of like an X-ray vision that helps identify portions of the bridge that are at risk of delaminating.”

Impact assessment

The need for timely and accurate inspection of bridges and other structures is growing as America’s infrastructure ages. According to a 2019 study from the American Road & Transportation Builders Association, the U.S. has over 47,000 structurally deficient bridges.

“The economic value of quicker bridge inspections is kind of tremendous, when you look at the financial need for infrastructure repairs today,” David Ferrell, managing director of Dynam.AI, told Construction Dive. “It also can help save lives and create a better work environment for civil engineers who don’t have to strap on a hard hat and step out where traffic is whizzing by at 70 miles per hour.”

GBA has used the new system on about 35 projects over the past year, including along the I-135 Canal Route highway through Wichita, Kansas. For that project, the company inspected five linear miles of bridges for a total of 1.6 million square feet of bridge deck and 594 bearing assemblies, Lindner said.

That stretch of highway carries an average of 92,500 vehicles a day, according to the Kansas Department of Transportation, so minimizing closures was a big priority.

“To shut that down and do a traditional chain drag would have meant six to eight weeks of traffic disruption and put traffic engineers in harm’s way,” Lindner said. Instead, the drone-based data collection took five days.

The project team estimates that the process saved $250,000 on direct costs alone. In the future, the tool will allow KDOT to track areas of damage as repairs are made or additional assessments are gathered.

Future applications

The system has improved since it was launched nine months ago, Lindner said. For instance, the drones can now pinpoint issues to within a very precise area of roadway. Earlier versions were accurate to within 5 to 15 feet, but the program is now accurate within just a few feet.

“Getting within less than 5 feet may not seem like much, but it is a big difference,” he said, “it can mean not having to shut down an entire lane of traffic to make repairs.”

While the system is designed for inspection of existing bridges, the techniques it employs could be a stepping stone to technologies that can help avoid a situation like that of the Florida International University pedestrian bridge that collapsed just before opening last year, said Ferrell.

In addition, the team sees the potential for infrared-equipped drones helping with analysis of new construction such as parking lots, power plants and evaluating the curing temperature of concrete, said Lindner.

The goal for now, though, is to develop an off-the-shelf bridge inspection solution that is available to traffic engineers and builders nationwide. “We’re trying to wrap all of this technology up in a neat piece of software,” Lindner said.

Recommended Articles