Yahoo Autos 
Waymo is teaching its self-driving cars how to handle Michigan's snow
It’s beginning to look a lot like Christmas...in Michigan, at least. That’s part of the reason is why Waymo is set to begin testing its self-driving vehicles in the in the Detroit area.
Waymo’s CEO, John Krafcik, announced the news in a blog post on Thursday. In it, he said that the blustery and snow-filled Michigan winters offer an ideal environment for just training self-driving cars how to successfully remain in control in treacherous conditions. Just as importantly, the winter testing will give Waymo’s engineers ample data points with which they can assess how the various sensors perform in less than ideal conditions.
MORE: The five levels of self-driving cars
“And,” he said, “it will also build on the advanced driving skills we’ve developed over the last eight years by teaching our cars how to handle things like skidding on icy, unplowed roads.”
The capability of the sensors in conditions where visibility is hampered is particularly relevant. While many automakers utilize cameras to varying extents in their semi-autonomous systems, like lane keep assist and adaptive cruise control, they shut off completely when vision is obstructed. Even when precautions are taken, such as placing the camera behind the windshield, where wipers can help keep things cleaner, no system has managed to perfect the problem of system visibility.
DON'T MISS: Waymo has its own fake city for self-driving car testing
Testing in Michigan, then, with its notoriously snowy-salty-sludge-filled winters, represents a serious challenge for any systems that rely on visual data. As Thrillist says, “You can't even tell if it is drizzling rain or snow, because the brown salt sludge that sprays up off the road coats your windshield more completely than anything that falls from the sky.”
The testing will center around Waymo’s existing facility in Novi, Michigan, where Waymo’s engineers can work side-by-side with its auto industry partners to refine its proprietary suite of sensors and equipment.
