$data.PageTitle

image.jpg

Driving in the winter is more challenging than on sunny summer days, even for self-driving cars.

Researchers at University of Waterloo are conducting studies to help train future self-driving cars to handle wintry conditions.

They say it's the first publically-available set of date in the world for self-driving cars in winter.

"The main issue is that publicly released data sets are currently sunny weather, night, cloudy and at the most rain, so there's nothing with snow right now," explains graduate student Matthew Pitropov.

Of course in Canada, there's no way to avoid winter on the roads.

Falling snow, black ice and darker days make for a more challenging set of circumstances.

"We need to address this problem," says lead researcher Krzysztof Czarnecki.

"We don't want Canada to be behind, in Ontario, particularly."

Over the past two winters, the team joined forces with researchers at University of Toronto and drove Canadians roads to scan harsh, snowy conditions using cameras, scanners and a GPS tracker.

They then processed that data into the vehicle's computer algorithms that allow cars to self-drive.

"When the vehicle detects something that isn't normal asphalt, whether its snow covered or ice covered, what ends up happening is the vehicle asks a few further questions: what's the date, what's the weather been like the last 36, 48 hours," explains Automotive Research Centre Manager Ross McKenzie.

Based on that information, the self-driving car can make decisions right away.

The team says they've made progress in teaching the self-driving cars the intricacies of winter driving but that there's still a lot of work that needs to be done.

They say that, just as it is for people, winter driving for a self-driving car is a lot more difficult, but that their information is important if self-driving cars are to become available everywhere.