Meet Rover, the AI app that's helping sniff out potholes in Markham
Rover uses smartphone camera instead of expensive scanning equipment
Potholes might look a a cheap and simple problem for cities to solve — after all, they're just holes in the road that need to be filled in with a bit of asphalt, right?
Well, actually, no.
Potholes are expensive to track down and fix, which is why the City of Markham is piloting a new smartphone-based artificial intelligence system that detects them more efficiently and at less cost — so they can be repaired before they become bigger, and more costly, problems down the road.
Called Rover, the app uses the phone's camera to spot damage that human eyes may miss, helping municipalities save money by making more timely road repairs.
After a quick set-up, the phone is mounted on a cradle on a vehicle's windscreen, with the camera pointed at the road.
While city inspectors drive their routes checking on the city's infrastructure, such as signs, painted lines and sidewalks, Rover is keeping a close eye on the condition of the road.
Alice Lam, the City of Markham's senior manager of roads, survey and utility, says initially two city vehicles will be equipped with Rover, helping staff detect problems faster and more efficiently.
"Having this device capturing the pothole location can save them time from stopping and having to mark down the location. It also eliminates human error."
Potholes are marked by a blue mark on the screen and the GPS coordinates are automatically logged so a road crew can be scheduled to return and fix them.
"I have a separate crew to help us to do the pothole repair. That crew has a list of all the pothole locations. Boom, boom boom they can go out and fix all of them and one day," said Lam.
While there are other AI pothole detection systems, this one doesn't require thermal cameras, laser scanners and other sensors mounted to a vehicle, which makes it an affordable solution for municipalities. It's also easy to train employees to use.
But like a new puppy, the Rover artificial intelligence program must also be trained.
Roy Tal, the chief technology officer at Visual Defence Inc., based in Richmond Hill, Ont., says Rover relies on machine learning, so the program requires a lot of diverse data to be able to tell the difference between a pothole and a shadow, for example.
"You'd be surprised by how many different types of potholes there are," said Tal.
The company hopes this pilot will show that AI can easily and cheaply detect problems with roads, bridges and other infrastructure.
It can be taught to spot cracks or water damage underneath the asphalt layer that will become problematic, or other warning signs, like water pooling on pavement.
"I would say that Rover is like a second set of eyes," said Tal.
"It gives the municipality's operator patrols that ability to have a more real time picture about what's going on on the road."
The City of Markham budgets $8 million annually to maintain 1,100 kilometres of road.
Mayor Frank Scarpitti says the city prides itself as a technology hub.
He says repairing potholes in a timely way not only saves a bigger expense later, but protects the city against legal claims by motorists for vehicle damage if they hit one.
"I think it's just cool technology. We're living in some pretty exciting times and I think it's really cool when technology allows us to cover more ground with less costs and in a more effective way."
As for any privacy concerns, the city says Rover is only collecting pothole images and locations. Any licence plates or anything captured that may identify an individual are blurred out.
But Scarpitti says Rover won't replace calls from the public reporting nuisance potholes.
"Any member of the public can take a picture of anything they see that needs our attention, so we haven't eliminated what I call that neighbourhood watch type mentality."