Autonomous Driving

Ford tries out self-driving cars in winter

Snowy-day testing completed at University of Michigan's Mcity

Ford winter autonomous driving test

One of the biggest concerns with autonomous driving, especially in northern climates, is how the technology can cope in sloppy, slippery conditions.

One of the things drivers are told about some driving aids is not to use them in snowy and icy, and even sometimes rainy, conditions because they will slow down the car’s programmed emergency response times — braking takes longer, for example, though with collision mitigation the car’s programming will maintain the “optimum” following distance before reacting to a perceived emergency. Plus, there’s also the chance of the occasional white-out, which means some of the camera-based systems could be rendered blind and road markings essential to staying within lanes could be obscured.

So if self-driving cars are to become reality, they have to be able to handle adverse conditions, and Ford wants to show that it is up to the challenge by being the first automaker to test an autonomous vehicle in wintry conditions, and also to have the technology to make self-driving in snowy conditions as easy as it is in the dry.

One of the Ford technologies toward its autonomous vehicle(s) is LiDAR (an amalgamation of Light and Radar), a camera-based technology that maps routes in 3D, storing them for future use when drivers input their autonomous trips. LiDAR is able to map route characteristics out in dry conditions, to the tune of 2.8 million laser points per second, which can be used as a baseline when snowy conditions would otherwise prevent technology from knowing where the car is and what it should be doing.

The Ford autonomous vehicle (a Fusion Hybrid, for testing purposes) collects hundreds of gigabytes per hour as its driving around — signs, buildings, even trees and streetlight posts — which it uses to build its 3D renderings that can then accurately depict the landscape through which the vehicle is driving, even if it’s under a blanket of snow.

The downside to this rendering technology is that LiDAR can also map the occasional snowflake or raindrop, which theoretically could lead the collision avoidance system to attempt to steer around it or attempt to brake to a stop. So, the company has teamed with the University of Michigan to create an algorithm to filter out snowflakes and raindrops as potential obstacles to be avoided.

And there are also sensors to scan the sensors to make sure all cameras and radar units are running up to standard, in the event they get crusted over with snow or road grime.

Most of the snowy-day testing has been completed at Mcity, a 32-acre, real-world driving environment at the University of Michigan.

Autonomous Driving | Auto Technology


<p>2017 Nissan Sentra SR Turbo</p>
FIRST DRIVE: 2017 Nissan Sentra SR Turbo

Nissan’s compact Sentra gets a turbo boost and a big dose of sportiness

<p>2014 Jeep Wrangler</p>
Should you buy a used Jeep Wrangler?

The Wrangler is not just a vehicle: it’s a lifestyle statement

<p>2017 Lincoln MKZ</p>
What you should know about the 2017 Lincoln MKZ

The MKZ, introduced just last year, has been heavily revised for 2017