Nvidia is best known for making graphics cards for running image-intensive games like Crysis 3. Driving a car is not a game, but Nvidia has designs on putting its graphics chips in vehicles, too.
Nvidia is shipping a development kit to several luxury car companies, including Bentley, Aston Martin, Tesla and Rolls-Royce, the Wall Street Journal reported on Thursday. The kits consist of a powerful graphics processor installed in prototype cars that reports to a server back at the carmaker's lab.
The development kit can help "train" prototype self-driving cars equipped with a battery of cameras and other sensors looking at the road around them.
Nvidia is using technology it originally developed for graphics designers and gamers in cars. Turns out, the type of artificial intelligence (AI) that self-driving cars use to identify objects on the road -- that's a curb, don't hit it -- runs much better on graphics processors than traditional processors made by companies like Intel or designed by ARM.
The key is that Nvidia's Tegra chip has only four traditional cores, but its GPU has 192, and AI can take advantage of all of them. That type of AI technique is called deep learning, and it's essential for computer vision or training a robot to spot patterns and edges in images and identifying what they are (or likely are).
"The system can, for example, detect the difference between a taxicab and a police car and will know that the driver has to pull over for the police car if it’s flashing lights at the driver," Danny Shapiro, director of automotive operations at Nvidia, told the WSJ.
For now, automakers are using Nvidia's chips to train their algorithms. That means that cars with the kit installed will collect data so that when self-driving cars eventually hit the road, the self-driving system will have seen a larger range of rare events, like when a deer crosses the road. More data means safer self-driving cars.
Nvidia is aggressively courting the automotive market because it must diversify in the face of diminishing growth in personal computers. It doesn't provide a lot of chips for smartphones and tablets, either, although Google's Nexus 9 tablet uses an Nvidia chip. But the company has also long seen applications for its GPU technology beyond graphics, and with self-driving cars on the brink, it seems like a mainstream application has finally caught up with Nvidia's ambition. Tesla CEO Elon Musk is already a fan.
"What Nvidia is doing with Tegra is really interesting and really important for self-driving in the future," Musk said at a conference in March.
So if one day you're riding in the back of a self-driving car, you have video games to thank.