During the golden age of PC and intelligent mobile phone processors, Intel has owned keynotes at CES in Las Vegas since 2012.
But it was NVIDIA’s turn this year.
AT CES 2017 on January 4, NVIDIA founder and CEO Jen-Hsun Huang hit the prime time showing his company’s leadership in three areas: Artificial intelligence (AI), autonomous cars and video games and became a star at the show. It is a reflection of a major shift in the computational requirements for artificial intelligence and deep learning, which depend on GPUs and alternative processors to handle different style of tasks. The new technology evolution of NVIDIA unveiled by Huang would define computing in the early 21st century in the same way what Intel did in the 1990s.
In April 1993, Huang founded NVIDIA with two partners, and it soon acquired its reputation in the tech world for developing graphic processing units (GPUs) and systems on a chip (SoCs) for mobile computing. However, in recent years the company has expanded into the auto industry by its involvement of artificial intelligence and deep learning that is required by self-driving cars. It transitioned three years ago into a platform company focused on three markets: gaming, professional visualization, and automotive.
NVIDIA had a great 2016 and its stock price soared to an all-time high in December. This is largely due to the company’s vision on AI and its ability to execute on the vision all starting with very strong leadership from Huang, who has been voted as one of the world’s best performing CEOs by Harvard Business Review in 2016.
AI Auto-Pilot and AI Co-Pilot
“Now let’s talk about AI for transportation. Everything we enjoy, everything we own, everything we eat and nourish our families with is owed to transportation. Parked cars cause waste, see litter, everywhere! Humans are the cause of waste, the robots should run things. And your car will be ‘your most personal robot’.” And autonomous vehicles would contribute immensely to a reduction of waste and crowding, especially in cities. It’s nearly impossible to just tell a computer how to drive, but deep learning allows them to learn on their own. It’s about “building a mental model to perceive your surroundings. The technology to do this stuff has only recently become possible. AI is the solution,” Huang said at his keynote.
Driving is complicated and unexpected things happen. Freezing rain turns the road into a skating rink. The road to your destination is closed. A child runs out in front of the car. You can’t write software that anticipates every possible scenario a self-driving car might encounter. That’s the value of deep learning; it can learn, adapt, and improve. Deep-learning breakthroughs have sparked the AI revolution. Machines powered by AI deep neural networks solve problems too complex for human coders. They learn from data and improve with use.
NVIDIA’s self-driving car system, the AI Auto-Pilot, can understand the environment around the car. And it is based on its new Xavier AI Car Supercomputer: 8 core custom ARM64 CPU, 512-core Volta GPU, thirty trillion operations in 30 watts. NVIDIA already has a prototype vehicle named BB8 based on the Lincoln MKZ, which has been driving around in California and New Jersey to test their capability and prove them for commercial uses.
What’s more, Huang also unveiled the AI Co-Pilot.
“There are places where the AI’s confidence is not high – it knows the path, but perhaps some blocks aren’t well mapped so it will pass control over to you,” said Huang. “The AI should be a Co-pilot. And the name of the new feature is AI Co-Pilot. It’s an AI that’s watching everything inside and outside the car. The AI Co-Pilot is paying attention to you too. Maybe you’re not paying attention. Maybe you’re dozing off. Maybe it notices you’re a little agitated and should pull over. Your car will judge you!” Huang went and explained the car’s AI will be able to understand if the driver is angry and tell the driver to pull over to the side of the road if it doesn’t look like the driver is ready to drive. AI Co-Pilot performs facial recognition and gaze tracking. It will watch which direction the driver’s head is turning, where the driver’s eyes are gazing and read the driver’s lips and knows when to signal a warning.
The AI Auto-Pilot and Co-Pilot will revolutionize personal mobility services, or reduce the need for sprawling parking lots within cities, according to Huang, and self-driving cars have the potential to do amazing social good. A recent study by KPMG predicts that computerized driver assistance technologies will help reduce car accidents by 80 percent in 20 years, or nearly 1 million lives a year saved.
Audi AI car coming in 2020
Nowadays every major automaker is in a race to develop the technologies that will power the cars of tomorrow: autonomous driving. The auto industry tends to like to develop its own technology and not rely on third players, however, that has been changed lately, particularly with ever more computerized cars. No automaker can do everything alone.
“Without a great car company to partner with, it’s hard to realize this vision. So today we’re announcing Audi and NVIDIA will partner to provide the next generation of AI cars,” Huang announced at his keynote. The self-driving Audi is demonstrating skill “not end of this year, not end of next year – right now,” which is indeed quite an achievement for a vehicle with that short of a training period.
NVIDIA and Audi aim to bring autonomous vehicles to the road by 2020. This is the most solid claim of a timeline for delivering highly automated driving, with a roadmap that relies on tested automotive-grade production computing platform, huge news for self-driving car indeed. And they have developed an Audi Q7 piloted driving concept vehicle to navigate a complex course, and it has been demonstrated at CES to showcase their progress. The demonstration also shows the power of end-to-end deep learning and how it could be one of many neural networks that can run inside of an artificial intelligence car. Audi also said it will expand testing of its automated vehicles on public roads in California and select states next year.
The partnership with Audi is just one of a half a dozen announcements Huang announced during his keynote. He also announced a partnership with automotive supplier ZF, known for their involvement as a supplier for trucks and commercial vehicles, so it makes sense that NVIDIA would expand their involvement in automotive outside just self-driving cars. NVIDIA also partnered with Bosch, which is adopting NVIDIA’s computing platform to bring AI to autonomous vehicles.
“Every car company in the world, any car you build, should be able to drive anywhere in the world. Therefore map works – all the stages of mapping, from what the car sees to what the data center sees, how those datasets are reconciled, is very critical. Baidu dominates China much in the same way Google dominates the Web in the rest of the world. And China is the largest auto market, too big not to be ignored. And Baidu is an incredible partner of ours,” said Huang.
At the Baidu World Conference in Beijing last September, Huang announced NVIDIA and Baidu had agreed to collaborate on the incorporation of artificial intelligence in a cloud-to-car autonomous vehicle platform. The platform will include cloud-based high-definition maps. The partnership brings together two companies that see vast potential in AI. The companies plan to integrate Baidu’s cloud platform and mapping technology with NVIDIA’s self-driving computing platform. They will work together to create solutions for high-definition maps, autonomous vehicle control and automated parking.
“We’re going to bring together the technical capabilities and the expertise in AI and the scale of two world-class AI companies to build the self-driving car architecture from end-to-end, from top-to-bottom, from the cloud to the car,” Huang said at the Conference at the time.
Baidu will utilize the technology for its own taxi fleet as well, as it invested in Didi Chuxing. Baidu already has a huge presence in China and this partnership can essentially get up to speed very fast and deliver a product equally fast.
NVIDIA also has entered cooperation with other mapping company partners such as HERE, TomTom in Europe and Zenrin in Japan.
These partnerships are to partner on AI technology for HD mapping capabilities from cloud to car. German luxury car makers BMW, Daimler und Audi bought HERE from Nokia for €2.55 billion ($2.6 billion) last year to create a platform for developing self-driving cars and to secure their access to intelligent high-precision mapping systems.
HERE is a leading mapping technology provider instead of Google for companies in the auto business, and that’s part of the benefit to carmakers. Some carmakers are spooked that Google could ask them for certain terms in the future, like demanding service additions that all carmakers aren’t comfortable with and HERE is an important element of the ecosystem.
NavInfo, a leading provider of digital maps and location services in China, Tencent, a leading provider of internet value-added services in China, and GIC, a leading global investment firm, recently acquired a 10 percent stake in HERE. Intel is also buying a 15 percent stake in HERE in 2017, and wants a precise location system for self-driving cars
On top of the 2020 target date for Audi and NVIDIA to deliver an autonomous AI car, NVIDIA also announced a partnership with Mercedes to work with it on bringing a product to market within 12 months.
Since it began working on autonomous vehicles several years ago, NVIDIA has rolled out development platforms and formed partnerships with more than 80 automakers including Tesla and suppliers for its efforts to bring autonomous cars to public roads in production capacity. These partnerships underscore NVIDIA’s leadership in AI and their commitment to accelerate the growth of automotive innovation. They also position NVIDIA as a central player in the automotive space and not just another tech supplier that is getting into the automotive world.
“Let’s make sure none of our kids ever have to drive,” Huang said at the end of his keynote.
With help from NVIDIA, that day may come sooner than we expect.