Developing Reliable Data for Robotic Bridge and Tunnel Inspection

David Lattanzi (on left), Assistant Professor, and Ali Khaloo, graduate student, work with Occulus hardware. Photo by Evan Cantwell.

Anyone who has ever tried to navigate a busy street or drive on an interstate highway when roadwork is going on knows that traditional repair methods can be a slow and cumbersome process. These methods use heavy equipment and often place workers in life-threatening situations.

That’s where George Mason University engineering professor David Lattanzi believes robots can help.

“If we can develop safe, cost-effective and accurate robotic inspection tools, we can eliminate many of the dangers of this hazardous work,” he says.

Lattanzi and his research team at the Volgenau School of Engineering are studying how to convert robotic inspection information into virtual computer “worlds” that can be explored by a human inspector. To accomplish this, they adapted techniques from computer vision, such as Structure From Motion, and virtual reality equipment like the Oculus Rift.

Wearing the virtual reality headsets allows the researchers/inspectors to see what the unmanned aerial vehicles inspecting the site sees, and then take measurements and create a scale model to examine, discover and fix structural problems. With this combination of technology and accurate data interpretation, the team hopes to save time and money while reducing the dangers for inspectors and their crews.

Prior to completing his doctoral degree, Lattanzi was a structural engineer at the firm Gannett Fleming, where he inspected and was responsible for the rehabilitation design of automobile and railroad bridges, tunnels and other structures.

“Most of the nation’s highways, bridges and tunnels were built to last about 75 years,” says Lattanzi, who teaches in the Dewberry Department of Civil, Environmental and Infrastructure Engineering. “The big postwar building boom was in the 1950s so these structures are nearing the end of their life span. The cost and difficulty of inspecting and repairing them will demand that we think of new ways to attack this problem.”

But as robotics become an accepted tool for infrastructure inspection, it becomes important to consider how humans will interact with the information that those same robots capture. If the data are not clearly represented or interpreted, tunnel walls could be mistaken for floors­ or the top of a bridge deck could be mistaken for the bottom.

“Sometimes, when you watch the camera footage from one of these unmanned aerial vehicles, you almost need a dose of Dramamine,” says Lattanzi.