So, how does an autonomous vehicle successfully navigate snow-covered streets? I mean, such road surfaces cause vehicles to handle far differently compared to what’s experienced in “good weather” for starters. Also, visual reference points are – obviously – obscured to a high degree in wintry conditions.
What to do?
Well Ford Motor Co. thinks it has worked out a solution of sorts based on recent testing at Mcity (click here for more information on THAT interesting place) that relies on five key steps:
- High-resolution 3D roadway maps created in good weather: To operate in snow, Ford said the Fusion Hybrid autonomous vehicles it tested first needed to create high-resolution 3D digital maps of the roadways with its four light detection and ranging (LiDAR) sensor system. By driving routes first in “ideal weather,” the self-driving systems are able to create highly accurate digital models of the road and surrounding infrastructure, the OEM said – mapped with sensors generating a total of 2.8 million laser points a second. The resulting map then serves as a “baseline” used to identify the car’s position when driving in autonomous mode, Ford said, so the car can locate itself within a “mapped” area even when roads are covered in snow.
- GPS isn’t accurate enough: Ford noted current GPS systems are accurate to a little over more than 10 yards; good enough for human drivers, but not for the far more precise needs of autonomous vehicles. By scanning constantly for landmarks, then comparing that information to previously created 3D digital maps stored in their databanks, Ford said its autonomous vehicles can locate themselves to within a centimeter.
- Detailed data collection makes for better maps: Lots of data is needed to make such “high-resolution” 3D maps. Indeed, Ford said its autonomous vehicles collect and process more mapping data in an hour than the average person’s mobile-phone data plan allows for over a decade. (Hate to the see the phone charges for THAT!) Such humongous amounts of data are needed because Ford said its self-driving vehicles collect and process information about both the road and surrounding landmarks simultaneously – signs, buildings, trees and other features. All told, the car collects up to 600 gigabytes per hour, which it uses to create a high-resolution 3D map of the landscape. FYI: in the U.S., the average cellular data plan subscriber uses about 21.6 gigabytes per year, for a 10-year total of 216 gigabytes. Gads!
- Filtering out sensor data to eliminate “false impression” images: Ford said the laser points within the four LiDAR sensor arrays used in its autonomous vehicles are so fine they even bounce off falling snowflakes or raindrops; creating a “false impression” that there’s an object in the way. Thus Ford – along with and University of Michigan researchers –needed to create an algorithm that “recognizes” snow and rain in order to “filter” both out of the car’s vision so it can continue driving normally.
- Sensor fusion – the combination of data from multiple sensors – keeps self-driving vehicles from going blind: In addition to LiDAR sensors, Ford said its autonomous vehicles use cameras and radar to monitor the environment around the vehicle, with the data generated from all those diverse technologies “fused together” in a process known as sensor fusion. That creates 360-degree situational awareness for the vehicle and allows it to overcome failures caused by ice, snow, grime, or debris build-up without hindering autonomous driving. Eventually, Ford thinks self-driving might be able to handle ice and grime buildup themselves through self-cleaning or defogging measures.
At the end of the day, though, it seems like an awful lot of technology needs to remain up and running so a car can successfully pilot itself over snowy and icy roads. Will it hold up under the wintry strain? We’ll see.