You are here

UW researchers develop artificial intelligence software to detect road problems

Thursday, August 2, 2018

WATERLOO — Potholes and deteriorating roadways are the bane of motorists and a big task for municipalities to manage.

An artificial intelligence software system developed by University of Waterloo researchers can simplify road maintenance to flag defects early, saving taxpayers money and drivers headaches.

"It would make everybody's lives a lot better," said John Zelek, an engineering professor at Waterloo.

The system automatically analyzes photographs snapped by cameras mounted on vehicles for potholes, cracks and other issues on the roadways. It's efficient and effective compared to approaches currently used by governments.

A municipality can employ vehicles with specialized equipment to capture images of pavement for a team of analysts to look at.

"It's time consuming and very expensive," Zelek said. "You don't get consistency because there is some subjectivity."

Many small jurisdictions can't afford that, and instead have workers simply drive around looking at roads and noting those spots needing repair.

The automated system could save money on monitoring while allowing more timely repairs. Data could be loaded onto street maps to help officials plan repairs and identify areas that need more urgent attention.

"It's more accountability for the public where their tax dollars are going," Zelek said. 

Rather than relying on expensive technology, low-cost cameras or even cellphone cameras can do just as good a job. They could be mounted on city vehicles, such as buses and garbage trucks, to monitor the roads while going about their daily work.

This article was originally posted on