Robot bridge inspectors more reliable than people, UW researchers say

Sriram Narasimhan and his team of researchers at the University of Waterloo have developed robotic technology to automate infrastructure inspections.

There's more to bridge inspections that meets the eye

Researchers at the University of Waterloo developed a new robotics system that can inspect bridges and other infrastructure more reliably.

A team of researchers from Waterloo University have designed a robot that they say performs bridge inspections faster, more reliably and at a lower cost than ever before.

The robot, developed by engineering professor Sriram Narasimhan and his team, drives around the structure and creates a 3D map. It creates a profile for the specific bridge, and records the progress of cracks over time.

"There are two different types of cameras on onboard," Narasimhan explains. 

"One is an infrared camera that would detect defects that are sub-surface that are hard to see with the naked eye. And the other types are basically what your eye would see."

Narasimhan says the process takes about 10 to 15 minutes, a fraction of what is required for a typical bridge inspection. The costs saved on labour and renting equipment such as platform lifts also make it the less expensive option.

Eliminating human error

For now, bridges and other infrastructure are checked by an inspector, who looks for and records any cracks or defects. But lead researcher Sriram Narasimhan says that process is subjective and prone to human error.

"First of all we want to alleviate the access issue. The other things we want to make our inspections are more reliable and more repeatable," Narasimhan said.

Some places are difficult to reach, and require inspectors to use binoculars. There is a possibility of misidentifying a crack from a previous report, or the size of the defect appearing different based on the inspector's vantage point.

"It's hard to actually measure the defects by hand," said Nicholas Charron, a PhD student working on the project.

"A lot of it is done by just visually estimating the size of it. Whereas our sensors can measure size up to plus or minus one or two centimeters depending on the type of defect and where it's located."

Looking ahead

The next challenge for Narasimhan is developing software that will analyze the data collected by the robot and predict when a bridge needs repair.

Currently, the robot is designed as a vehicle with wheels. But the team is also creating a water-borne robot as well as a drone. The technology is being tested at the shuttle train track system at Toronto Pearson Airport, and on Ontario Power Generation infrastructure.