As any hardened gamer will tell you, a blistering gaming rig isn’t complete without at least one high-end graphics card. Have multiple monitors or spend time in virtual reality with Oculus Rift’s VR headset, and you’ll need more than one.
Californian chip manufacturer NVIDIA, known for its range of super-fast GeForce GTX graphics cards, is unarguably one of the biggest and best graphics card manufacturers out there. But while it is best-known for its graphics card, the Santa Clara firm builds far more than just graphics cards, with a range of chips designed for use in everything from server farms and cloud computing applications to tablet computers and automotive embedded systems.
And as it detailed last night the 2015 Consumer Electronics Show in Las Vegas, it wants a piece of the autonomous vehicle market too.
Enter the NVIDIA DRIVE PX and NVIDIA DRIVE CX, two brand-new in-car computer systems that have just been launched by at CES 2015. While the DRIVE CX is designed primarily to be used as a the heart of a powerful digital infotainment and navigation system, featuring enough processing power to drive an astonishing 16.8 million pixels of dashboard screens and displays, the DRIVE PX has been built from the ground-up as a computer powerful enough to enable autonomous driving features.
What’s more, the DRIVE PX is capable of something NVIDIA calls deep learning: a process which allows it to improve and refine its driving behaviour with time.
“Mobile supercomputing will be central to tomorrow’s car,” said Jen-Hsun Huang, CEO and cofounder of NVIDIA. “With vast arrays of cameras and displays, cars of the future will see and increasingly understand their surroundings. Whether finding their way back to you from a parking spot or using situational awareness to keep them out of harm’s way, future cars will do many amazing, seemingly intelligent things. Advances in computer vision, deep learning and graphics have finally put this dream within reach.”
At the heart of the DRIVE PX is a brand-new NVIDIA Tegra X1 mobile super chip, which is built on the same Maxwell GPU architecture found in many of today’s most popular high-end NVIDIA graphics cards. Capable of more than one teraflop of processing power — 1 trillion
billion operations per second — the DRIVE PX is built with twelve discrete high-resolution video inputs, meaning it can process and display images from 360 degrees with ease.
NVIDIA says this incredible graphics processing — capable of processing 1.3 gigapixels per second — gives the DRIVE PX the ability to find a parking space and park itself without human interaction, a feature it is calling Auto-Valet.
Like so many other self-driving valet-parking features we’ve seen demonstrated, NVIDIA’s new DRIVE PX can interact with a smart phone to allow the car to meet its owner again at the drivers’ desired pickup point.
Sadly, NVIDIA’s autonomous drive and cockpit car computers aren’t the kind of things you’ll be buying off the shelf in your local Best Buy for retro-fitting to your current car. Instead, they’ll be sold directly to automakers in a tier-1 supply chain relationship, first for prototype autonomous-driving vehicles and later for production models.
Potential customers of these new chips include German automaker Audi, whose autonomous drive program has already demonstrated some impressive capabilities to date based on previous-generation technology. And based on Audi’s response to the latest hardware from NVIDIA, it’s a relationship which will continue for many years.
“Audi and NVIDIA share a common belief that machine learning is a powerful enhancement to our zFAS Piloted Driving technology,” said Ricky Hudi, executive vice president of Electrical and Electronics Development at Audi. “Thus Audi, sees DRIVE PX as a crucial tool for further research and development.”
As for other potential customers? We’re guessing that many OEMs are already looking at this new hardware with eager eyes and wouldn’t be surprised if Tesla is among them. Unlike many of its rivals, Tesla is known to make use of the latest-generation hardware in its cars, and is known to apply generational updates to cars during model years if there’s a business case for doing so.
Even though Tesla is also know to use custom-designed, custom-built hardware, we’re guessing that the appearance of this latest-generation autonomous drive hardware from NVIDIA could find its way into a future Tesla model, since it is probably more cost-effective to use existing components than design its own.
You can also support us directly as a monthly supporting member by visiting Patreon.com.