A Google self-driving vehicle hit a city bus in broad daylight this week. Meanwhile, Ford says it has snowstorm situations covered.
We’re not sure why the Google-modified Lexus pulled out from a stop and hit the side of the city bus on a sunny California day, but it was a wake-up call that on the path to autonomous cars there will literally be some bumps along the way.
If a self-driving vehicle can hit a bus, it can hit a bicyclist or a pedestrian or a train.
Despite an overall good track record so far, many people have their doubts about self-driving vehicles. Part of the doubt stems from the fact that much of the technology manufacturers use now for lane keeping and forward collision prevention is optical.
For example, Subaru’s excellent EyeSight forward collision prevention system uses two cameras to do its job. It is reasonable to ask how well such a system might work in a snow storm in which not just forward visibility is reduced, but also when all lane markers and center lines are covered in snow.
Ford says that its coming autonomous vehicle technology is not limited by optical cameras. Nor is it limited by GPS, which is only accurate to within a few feet at best. Ford says its system is accurate to within a centimeter.
Rather, Ford uses LiDAR, which uses lasers to estimate distances and locations.
The LiDAR Ford uses captures 2.8 million laser points a second. Ford says that the amount of data acquisition the system generates each hour is roughly equivalent to the amount of data your cell phone uses in ten years.
Ford’s trick is that it is not just using the LiDAR in real time to map the area you will be driving. Ford will first map out the area ahead of time, and this super-accurate map will be in your vehicle’s memory.
The map, along with other inputs, will determine your route in combination with the data the vehicle is getting as you drive along.
Interestingly, early military cruise missiles used a similar method. They had pre-loaded maps of the area the device might travel and would scan for landmarks along the way to fine-tune the route.
Ford adds radar and cameras to the LiDAR system. The combined input of all these devices is called sensor fusion. The company says that one sensor covered in road salt or ice may not limit the system’s operation, and future systems will be self-clearing.
None of this really explains why the Google vehicle pulled out into a huge, slow moving vehicle in broad daylight, but in fairness, the vehicles that Google is currently testing on public roads are just beta units.