While Google's making great headway when it comes to developing self-driving cars, it's awfully hard to test your feat of autonomy in the snow when you're exclusively testing around Silicon Valley. In and around Detroit, the weather is a little more variable -- particularly this time of year. Ford has taken advantage of this to become the first automaker to test its autonomous cars in the snow.
This, as it turns out, is a ridiculously difficult problem. Most autonomous systems rely, or at least heavily utilize, painted markings on the road. Once the snow starts to fall, those are often all but invisible. Even the shape of the road can be obscured by piles of snow or ruts from other passing cars.
To solve the problem, Ford's engineers started scanning far more than just the road itself. By expanding the scope of the car's laser scanners, the company generated 3-D maps of the test area's surroundings, including things like buildings, trees and signs. Using these objects, the car can accurately position itself regardless of road conditions.
The sensors are also refreshing quickly enough, and scanning accurately enough, to still see effectively in heavy rain or snow. And, the car is not only always calculating the safe way forward, it's also calculating a safe escape. So, should a chunk of ice or rock unexpectedly take out a sensor and the driver is not able to take over, it can autonomously move itself to safety -- despite being effectively blind.
Ford announced last week at CES 2016 that it's tripling the size of its autonomous test fleet, and with all-season testing like this taking place, it's good to know that those cars are being designed with those of us living in the north in mind.