Audi to use NVIDIA tech to create advanced self-driving cars

Now that we've covered the gaming announcements from NVIDIA, it's time to take a look at what the company announced for the car market. There was one big design win as the company announced a collaboration with Audi to put advanced AI cars on public roads starting in 2020.

NVIDIA (NASDAQ: NVDA) and Audi today shifted their decade-long partnership into high gear, announcing that they are collaborating to put advanced AI cars on the road starting in 2020.

Speaking at the opening keynote at CES, NVIDIA founder and CEO Jen-Hsun Huang, joined on stage by Audi of America President Scott Keogh, said future Audi car models will use deep learning to tackle the complexities of driving.

The first phase of this new collaboration focuses on NVIDIA DRIVE™ PX, an AI platform for self-driving cars, which uses trained AI neural networks to understand the surrounding environment, and determine a safe path forward.

"NVIDIA is pioneering the use of deep learning AI to revolutionize transportation," said Huang. "Audi's adoption of our DRIVE computing platform for AI cars will accelerate the introduction of next-generation autonomous vehicles, moving us closer to a future of higher driving safety and new mobility services."

Keogh said: "Audi drivers know the pinnacle of performance and technology. In our mutual pursuit for safer roads, the partnership between Audi and NVIDIA will expand to deep learning and artificial intelligence to bring higher automation onto the road more quickly."

One of the highlights of CES 2017 will be demos of an Audi Q7 piloted driving concept. Passengers will be able to ride in the vehicle's back seat, with no one behind the wheel.

Outfitted with a DRIVE PX 2 and running NVIDIA DriveWorks software, the Q7 uses deep neural networks -- specifically, NVIDIA PilotNet, which recognizes and understands its changing environment while driving safely. The course will be modified during the demonstration, and features a variety of road surfaces, with and without lane markings, and a simulated construction zone requiring a detour.

In his keynote, Huang addressed how AI will anticipate the driver's needs -- driving to the office in the morning and home at night, automatically opening and closing the garage door, and adjusting the climate to the individual's preferences, as well as being able to understand and respond to requests in a natural conversational language.

NVIDIA and Audi first appeared together at CES seven years ago and announced a technology partnership that has steadily expanded. The results include the award-winning Audi MMI navigation and the Audi virtual cockpit, currently available across a wide range of the automaker's sedans, SUVs and sports cars. In the coming months, Audi will roll out its new A8, the world's first Level 3 automated vehicle with its Traffic Jam Pilot system, powered by zFAS, which integrates NVIDIA hardware and software.

Audi and NVIDIA have a demo track at CES 2017 to show off the Audi Q7 piloted driving concept car. The Q7 uses NVIDIA's Drive PX 2 platform.



NVIDIA also uploaded a new video clip that showcases the capabilities of its self-driving car platform by using the firm's own self-driving car, the BB8:



On a related note, NVIDIA also announced a partnership with HERE for HD mapping from cloud to car. This is the same company Intel just purchased a 15% stake in:

With their sights set on a driverless future, HERE and NVIDIA (NASDAQ: NVDA) today extended their collaboration to develop HERE HD Live Map into the industry-leading real-time, high-definition mapping solution for autonomous vehicles.

The broad collaboration for the solution, which would span the vehicle and the cloud, includes three planned initiatives:

HERE is accelerating HERE HD Live Map using NVIDIA MapWorks AI technology.

NVIDIA is developing localization technology based on HERE HD Live Map as part of NVIDIA DriveWorks software -- enabling automakers using NVIDIA DRIVE™ PX 2 in the car to integrate localization capability.

HERE and NVIDIA intend to collaborate on a HERE HD Live Map-based in-vehicle solution to perceive changes in the environment and update the map in the cloud accordingly.

"The physical world is changing all the time and self-driving cars need to be aware of that change so they can take better driving decisions," said Edzard Overbeek, chief executive officer of HERE. "HERE HD Live Map already addresses that need and by working with NVIDIA we can ensure that automakers deploying the NVIDIA DRIVE platform can easily enable HERE HD Live Map for self-driving cars."

"HD maps are essential for self-driving cars," said Jen-Hsun Huang, founder and chief executive officer of NVIDIA. "HERE's adoption of our deep learning technology for their cloud-to-car mapping system will accelerate automakers' ability to deploy self-driving vehicles."

HERE HD Live Map, a cloud service supporting all levels of vehicle automation, is already commercially available for North America and Western Europe. Through multiple modes of sensor ingestion and aggregation, it can update itself, with rich data layers assisting the vehicle in positioning, localization and strategy planning.

At the HERE booth at CES, NVIDIA and HERE are showcasing localization using the HERE HD Live Map on the NVIDIA DRIVE PX 2 AI computer. It uses deep learning to precisely locate the vehicle's position with centimeter accuracy, as well as to detect how the environment around the car may differ from the current map. Road tests are already taking place as part of this collaboration.

Besides the partnership with HERE, NVIDIA is also working with ZENRIN, the leading Japanese mapping company, on similar cloud-to-car mapping technology.

NVIDIA (NASDAQ: NVDA) today announced that it is collaborating with ZENRIN, Japan's leading mapping company, to develop a cloud-to-car HD map solution for self-driving cars.

The collaboration encompasses processing both in the car, where the data is collected, and in the cloud:

In the mapping-survey vehicle, the NVIDIA DRIVE™ PX 2 AI car computer and NVIDIA DriveWorks software can process massive amounts of data generated by camera and lidar sensors. Deep learning will enable the real-time image recognition, feature detection and classification necessary to build detailed environment models.

In the data center, NVIDIA GPUs and NVIDIA MapWorks software can be used to process complex datasets, compile and register data from multiple vehicles, and create a 3D map.

"ZENRIN's big data includes road images and point-cloud data captured by mapping-survey vehicles," said Koji Haraguchi, head of the Research and Development Office at ZENRIN. "Combining NVIDIA's AI technologies and ZENRIN's big data will enable us to provide wider coverage of HD maps to automotive manufacturers with a dramatically shorter lead time."

"Using NVIDIA's GPU technology, DriveWorks and MapWorks, ZENRIN will be able to accelerate the map creation process, and then easily detect changes when they occur," said Rob Csongor, vice president and general manager of Automotive at NVIDIA.

NVIDIA is additionally developing localization technology based on ZENRIN HD maps, which will become part of DriveWorks. This enables automakers using DRIVE PX 2 in the car to integrate localization capabilities.

ZENRIN has been focused for more than 60 years on creating maps of Japan. To reflect rapid change in the physical world, the company is increasingly turning to artificial intelligence as a tool for keeping its maps current.