Ford has announced at the Detroit Auto Show this week that it's now testing its autonomous research vehicles in snowy conditions. Inclement weather — conditions like heavy rain, snow, hail, and the like — have long been viewed as one of the final unsolved technical challenges to bringing self-driving cars to market.
The snow testing is, to some degree, a side effect of Ford's home base in Michigan, where wintry conditions are unavoidable for several months out of the year. The company notes that traditional autonomous driving sensors like LiDAR — those spinning things you see atop many research vehicles — can't see through snow, which renders them useless for building the high-resolution maps of a car's surroundings that are necessary for safe driving. Instead, Ford is using LiDAR to detect landmarks above the road, then switching to high-resolution maps of the road that are already stored onboard the vehicle to actually drive.
LiDAR isn't great in the winter
Those high-res maps, of course, don't always exist. Volkswagen and GM both announced this week that they've partnered with self-driving sensor company Mobileye to crowdsource better mapping using cellular data connections and the sensors that are already on board many vehicles. Tesla, which already has its Autopilot system in public beta, is doing something similar.
In Google's most recent self-driving report for the month of December, the company noted that it is using the rare rainy weather in California to test inclement weather in its own autonomous test fleet, but the company's two current testing markets — Silicon Valley and Austin, Texas — don't do much good for trying out snow. Perhaps a rumored tie-up between Google and Ford (which was expected to be announced at CES, but didn't happen) could help everyone involved.
