Further steps are being taken by Ford in autonomous vehicle development by testing cars in conditions where road markings are obscured by snow.
Taking place on roads in Michigan in the cold northern regions of the USA and specifically at the MCity proving grounds, the winter testing regime seeks to find answers to autonomous driving in conditions that are far less than ideal and which will be encountered by motorists in the real world rather than the laboratory.
According to Ford’s autonomous vehicle expert, Jim McBride, around 70% of Americans live in regions that are affected by snowy weather during some periods of the year, which is why the company believes it’s important to test in such conditions.
“It’s one thing for a car to drive itself in perfect weather but it’s quite another to do so when the car’s sensors can’t see the road because it’s covered in snow,” he commented.
A key aspect of autonomy is positional accuracy, something that can’t be achieved with GPS navigation alone, which can only pinpoint locations to within a few yards or metres. This is enough to know a location within a town but not enough to identify the precise position on a road, including positional accuracy within driving lanes.
To achieve this precision, Ford uses LiDAR technology, a laser based variant of Radar, which gives centimetre precision in identifying road position based on high definition images created in the vehicle’s computer system of the surrounding topography.
LiDAR represents the best technology currently available for gathering information about the road and surroundings in ideal conditions, but it fails to meet the right requirements when the road can’t be seen or if the sensor itself is covered in film, grime or other debris which is common in winter conditions.
Achieving autonomy in snow
Ford has long predicted that autonomous driving would be a problem in poor weather conditions and so it is now reaping the benefits of this foresight by using highly accurate high resolution 3D maps, which the company created using data gathering vehicles operating in ideal conditions. Such features as traffic signs, buildings and trees were automatically annotated on the map to create a signature profile of every location along the test routes.
When the road is covered in snow and the road markings are no longer visible, the car can still determine its precise location on the roadway by comparing the 3D features gathered from its surroundings to the mapping information. According to Ryan Eustice of the University of Michigan, home to MCity, “The maps we created with Ford contain useful information about the 3D environment around the car, allowing the vehicle to localise even with a blanket of snow covering the ground.”
Explaining further how the autonomous driving software works in such conditions alongside normal safety systems, Jim McBride continued, “These safety systems are the ones which are of most use on slippery winter roads so we want our autonomous vehicles to detect deteriorating conditions, decide whether it’s safe to keep driving, and if so, for how long.”
More work in the pipeline
The technical difficulties of autonomous driving on winter roads represent an enormous challenge and there is still a long way to go in its development but the onset of testing in these conditions is an important achievement in driverless technology made by ford.