You’re sitting at your chair, clicking around, when you come across something you want to comment on. You click into the text field, spend a minute thinking of something funny, erase it, spend another two minutes trying other things out, erase those, take a bathroom break, go back to your original joke, tighten it up a bit, and hit post.
A box pops up to prove you’re not a spambot. It might be a few letters or a picture of a cloud, or a series of pictures. The point is that there are things that a human can recognize by looking at which a computer cannot. Give a computer a password to break and it will figure it out. There are only so many combinations of numbers and letters, after all; it will brute force its way in. But a picture? Pictures are different.
Or so we thought.
Audi is working with chipmaker Nvidia to develop the artificial intelligence that will allow cars to drive themselves. Nvidia’s vision is for a car that can see—both literally and metaphorically—the world around them. While automakers typically use computers in terms of brute force that requires huge amounts of processing, Nvidia sees a better way.
It’s called deep learning, and it’s the future of automotive computational processing. Audi deep learning relies on cameras and sensors that rapidly consult a database of known images. The idea is that every new Audi will contain such a vast database that it will be able to drive itself anywhere, anytime. The more you drive, the more specialized a database it will construct, and the better a driver it will become.Just like a wizened cabdriver, Audi deep learning will learn the fastest routes and quickest shortcuts. Click To Tweet
The advantage of this is a) it requires much less computational power and b) it won’t require an upgrade since it will be constantly learning. It’s a smarter, leaner, more efficient computer that will make the chips in our current cars look like victrolas one day.
Our first demonstration of Audi deep learning in a production car will likely be the next-generation A8 arriving in 2017. But don’t sweat it; we likely have two decades of self-driving innovation beginning, so there will always be a refinement of Nvidia’s system in development.
If you like cars the old-fashioned way without any of these new-fangled gizmos, come to dealership today. All of our current vehicles are homegrown and organic with zero self-driving capabilities, just the way you like them.