Nvidia Xavier Supercomputer Aims To Turn Cars Into AIs On Wheels

Nvidia has shown off  Xavier, its latest supercomputer designed to power its self-driving car platform in order to inject artificial intelligence (AI) into the automotive world.

Building upon Nvidia’s 512 core Volta graphics processing unit and its neural network powered AI technology, Nvidia touted ambitions to use the supercomputer to power its AI Co-Pilot system, which essentially learns how to drive from watching a driver, deep learning algorithms and cloud-delivered mapping data.

“We would like to turn your car into an AI, and that by applying this technology we can revolutionise the automobile and bring joy and delight and safety to millions of people,” said Nvidia’s CEO Jen-Hsun Huang.

Nvidia and AI

While Nvidia has been working on providing underlying hardware and deep learning neural networks for car makers to build autonomous cars upon, such as its Drive PX 2 lunchbox-sized supercomputer platform found in the latest Tesla Models S, the company best known for its graphics cards aims to make AI more central to the automotive world than just taking over driving duties.

Nvidia is taking its work further with AI Co-Pilot which will not only use the normal array of sensors and external video cameras to carryout autonomous driving functions similar to those in other self-driving cars, but also use internal cameras to watch the human driver and monitor their driving behaviour.

If it detects that a driver’s eyes are not on the road ahead while also detecting an obstacle in front of the car, it can alert the driver to the potential risk they could face.

Nvidia also noted how the system could be used to make a driver’s everyday life easier by autonomously carrying out tasks such as opening up the garage door before a commute to work or adjusting climate controls to a drivers preference which is has learnt over time; essentially living up to its co-pilot credentials.

Nvidia also championed an evolving long-term partnership with German carmaker Audi to expand the use of autonomous driving and driver-assistance systems based on Nvidia’s 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,” said Audi’s America president Scott Keogh.

Nvidia also has a partnership with commercial vehicle company ZF, which announced the creation of the ZF ProAI self-driving system, which uses the Drove PX 2 platform at its core as means to power self-driving trucks and cars for industrial uses. 

Moving into the mapping arena, Nvidia also announced it adding its MapWorks AI technology into the HD Live Map service from mapping company HERE, in order to bring smart cloud based mapping into driverless cars.

In essence, this trio of partnerships demonstrates that the road to driverless cars will involve multiple partnerships and collaboration on systems and projects in order to make semi an true autonomous driving an everyday reality.  

Smart homes and cloud

At CES 2017, Nvidia also explained how it is bringing AI technology into homes by enabling its Shield streaming game console to support the Google Assistant AI, helping turn the gadget into a smart home hub.

“We want to be able to turn your home into an AI,” Jen-Hsun said. “We believe your home will engage you and you will engage your home in natural simple ways.”

At the same time, Nvidia is continuing to mix the cloud into is technology and services, with the introduction of GeForce Now, a cloud-based gaming streaming service that works with the Shield and is backed up by servers running the latest Geforce graphics cards.

Previous services such as the now defunct OnLive attempted to deliver such a gaming streaming service, but struggled due to issues over streaming quality, adoption and the ability to generate revenue. However, as a larger company with more resources and technology expertise, Nvidia stand more of a chance on making such a streaming service work where others have failed.

Nvidia’s technology chops appear to be going from strength to strength with the company having claimed to have made the world’s most efficient supercomputer towards the end of 2016.

Think you know all about supercomputers? Try our superquiz!