Flying Robots Inspect Infrastructure Damage

CEE Professor & Chair Jerome Hajjar and colleagues from Carnegie Mellon University are working to create autonomous flying robots that will be able to inspect, analyze, and assess any damages to bridges and buildings.

Source: News @ Northeastern

Next month, the con­tractor for the Mass­a­chu­setts Depart­ment of Trans­porta­tion expects to begin Stage 4 of con­struc­tion on the his­toric Longfellow Bridge, which con­nects Boston and Cam­bridge, its iconic “salt and pepper” towers guiding the pas­sage. Will it happen?

Per­haps not now, but soon. Northeastern’s Jerome Hajjar and col­leagues at Carnegie Mellon Uni­ver­sity are devel­oping air­borne robots to ensure that in the future such timeta­bles are kept and improve­ments are razor-​​sharp.

The team is devel­oping a sophis­ti­cated system called the Aerial Robotic Infra­struc­ture Ana­lyst, or ARIA. It uses small low-​​flying robots known as micro air vehi­cles, or MAVs, cou­pled with 3-​​D imaging and state-​​of-​​the art plan­ning, mod­eling, and analysis to inspect struc­tures such as bridges and build­ings and to auto­mat­i­cally iden­tify prob­lems, track their progress, and assess the need for follow-​​up.


An overview of a 3-​​D model of a bridge cre­ated by the autonomous ARIA robot using its unique sen­sors. Image by Varun Kasireddy, Carnegie Mellon University

Hajjar is cre­ating the algo­rithms that will help engi­neers assess the nature and sig­nif­i­cance of the damage using the data cap­tured by the MAVs and deter­mine the best course of action.

The MAVs, which are pro­grammed to fly and nav­i­gate autonomously, can safely and effi­ciently address places that are dif­fi­cult or dan­gerous to reach as well as those that have to be inspected repeat­edly.”
—Jerome Hajjar

Con­sider bridge inspec­tions,” says Hajjar, CDM Pro­fessor and chair of the Depart­ment of Civil and Envi­ron­mental Engi­neering. “Until now, they were pre­dom­i­nantly done by people with cam­eras crawling on stretches of scaf­folding that would be built for the pur­pose, taken down, and then recon­structed a little fur­ther along the bridge—an expen­sive process. The MAVs, which are pro­grammed to fly and nav­i­gate autonomously, can safely and effi­ciently address places that are dif­fi­cult or dan­gerous to reach as well as those that have to be inspected repeatedly.”

The research is being funded by the National Sci­ence Foundation’s National Robotics Initiative.

Taking flight

With their eight rotors splayed, the MAVs look like spi­ders whirring in the air.

Each MAV mea­sures about 3 or 4 feet across, and is equipped with a light­weight rotating laser scanner that pro­vides 3-​​D mea­sure­ments, three video cam­eras for high-​​resolution imaging, and a GPS to mon­itor its posi­tion. Onboard wire­less com­mu­ni­ca­tion tech­nology will send data—including 3-​​D models of struc­tures and their components—back and forth between a tablet oper­ated by an inspector on the ground.

Should the inspector see a problem, he can direct the MAV to overlay its characteristics—location, size, volume, and so on—on the 3-​​D model and then launch a sim­u­la­tion of how the struc­ture will behave under dif­ferent con­di­tions to assess the problem’s severity based on DOT criteria.

May 29, 2012 - Jerome Hajjar, Professor and Chair of the Department of Civil and Environmental Engineering at Northeastern University.  Photo by Brooks Canaday

Jerome Hajjar, CDM Pro­fessor and chair of the Depart­ment of Civil and Envi­ron­mental Engi­neering Photo by Brooks Canaday/​Northeastern University

The algo­rithms being devel­oped by Hajjar deter­mine the nature of the damage—spalls and cracks in con­crete, rup­tures or cor­ro­sion in steel, bends or buckles in members—on a refined level, doc­u­ment that damage so it can be com­pared with older and future images, and pro­vide a “con­di­tion rating,” from “excel­lent” to “failed,” to guide the mon­i­toring and sched­uling of repairs.

We want to know not simply whether the damage is there but if it mat­ters,” says Hajjar, who spe­cial­izes in com­pu­ta­tional sim­u­la­tions, exper­i­mental testing, and field inves­ti­ga­tions of struc­tures. That’s why the sim­u­la­tion is so impor­tant. “In the case of a bridge, for example, the system would sim­u­late the behavior of a bridge sub­jected to loads,” he says. “How much does the bridge deflect under par­tic­ular loads? How much stress on the mem­bers is too much? How will the damage the MAV has detected affect that?”

From inspec­tions to repairs

Next steps include making some fixes as auto­matic as the inspec­tions. “I think modest repairs are actu­ally quite fea­sible,” says Hajjar. He cites a steel crack in a bridge as an example. “The most common repair is to drill a hole in the crack tip, which changes what we call the ‘stress con­cen­tra­tion’ from infi­nite, meaning that it’s going to keep prop­a­gating, to a much more dulled-​​out lower-​​stress ver­sion,” he says. “It enables the bridge to handle the fatigue stresses for a much longer time.”

The trick is keeping the MAVs small, so they can inspect every nook and cranny of a struc­ture, while ensuring they are light enough to fly autonomously and still carry the laser scan­ners, high-​​resolution cam­eras, and other equip­ment required to per­form their jobs.

Bridges have proven optimum for field studies because they are public and acces­sible, says Hajjar. “Every bridge engi­neer I’ve talked to is excited about the project.”

Related Faculty: Jerome F. Hajjar

Related Departments:Civil & Environmental Engineering