NVIDIA’s Supercomputer to Accelerate the Self-Driving Car Race

In his CES 2017 keynote address, NVIDIA’s CEO Jen-Hsun Huang offered insight into the company’s ambitious strategy to “turn your car into your most personal robot”.

And that involves providing the necessary technology to give means to autonomous driving.

In this context, the American technology company showed off a new computer devised to enable self-driving cars.  And thus power the future. 

Named Xavier, the AI (artificial intelligence) car supercomputer is designed to learn how to drive by observing human drivers. 

Xavier is the latest, most powerful version of the company’s DRIVE PX 2 supercomputer.

The potential is worth discussing, so make sure to keep on reading.

How Can Deep Learning Enable Self-Driving Cars

Many people have fond memories of their parents teaching them how to ride their bikes. You felt nervous at first, but after a while, it all came pretty natural.

You just knew to ride the bike, although you weren’t exactly sure how you did it.

But how do you teach a computer to ride a bike – or drive a car? This is much, much harder to do. That’s because computers don’t have the innate cognition required to learn like we do.

Well, NVIDIA thinks deep learning is how you get past this barrier.

A branch of machine learning, it’s meant to “teach” machines to read and interpret unstructured data, such as:

  • images
  • speech
  • video.

Using algorithms, deep learning systems help computers perceive the world around them.

More about NVIDIA’s Lunch Box Sized Supercomputer

Xavier might be small, but it’s aimed to do great things.

So far, we’ve seen devices which could “see” or “hear” different obstacles or objects to assist self-driving cars.

But as Nvidia’s CEO explained in his speech, these are just senses. Perception takes things further. It involves accumulating all those senses and “building a mental model of what it is you are perceiving.”

Let’s put it this way.

The car computer can see what is going on around a vehicle in detail. But it also understands the information.  So, it can actually perceive the environment.  

Using the same technology robots use to learn how to walk like a human, a car could learn how to drive just by observing human drivers.

Backed by HD maps in the cloud, cars fitted with an AI supercomputer will know:

  • Where the vehicle is
  • Where other objects are around the car.

But through the deep learning system, it can do more than this. For instance:

  • It will know the difference between an ambulance and a delivery truck.
  • It can determine whether a parked car is empty of passengers or if the door is opening as a passenger gets out.
  • Predict where objects around the car will be in the near future.

Here’s a video of NVIDIA’s Star Wars-inspired vehicle fitted with the DRIVE PX 2 system. Take a look to get a grasp of  NVIDIA’s driverless technology.

What Are the Implications when It Comes to Safety on the Road?

Globally speaking, 3,287 people die every single day in road crashes.

An assistant accompanying you on the road – which understands what’s going on and can take the wheel – can turn out to be a game changer.

Acting like the brain of your car, an AI supercomputer can take more informed and thus better actions to prevent accidents. 

Automakers which Got Onboard

Tesla Motors has already announced IT will equip all its Tesla vehicles with an onboard “supercomputer” powered by NVIDIA AI Drive PX 2 computing platform.

Volvo will ALSO use it as an autonomous-car pilot program scheduled to start this year in Sweden.

Audi, the company’s long-time partner, is also interested in putting advanced AI cars on the road starting in 2020.

NVIDIA’s extensive list of partners includes many important names which might embark on this journey, such as:

  • MERCEDES-BENZ
  • BMW
  • Honda.

Meanwhile, Nvidia is testing different variants of the system using their own squadron of self-driving cars.

Over to You

NVIDIA is convinced the solution to self-driving is artificial intelligence.  And there’s no denying the progress they’ve made.

But I’m curious to know what are your thoughts about Xavier and AI pushing autonomous driving towards the future.

Make sure to share your opinions and ideas in the comment section down below.

Philipp Kandal