Nvidia leads the way to autonomous car
Everybody is searching on and developing the autonomous car but it’s video-cards producer Nvidia who takes the lead with his “mini-supercomputer” Drive PX Pegasus.
This small computer (in size) consumes 500 Watt and can do 320 billion calculations pro second, far more than any competitor. The Pegasus should be available next year and is powerful enough to make autonomous cars possible.
Nvidia-CEO Jen-Hsun Huang: “I think we will see the first automated taxis in 2020, because they can drive in a secured and semi closed system. When the new generation of electric cars comes in 2022, they can’t allow it to be non-autonomous, this is a so called 1 billion $ opportunity.”
Nvidia has already drawn the interest of many automotive customers. Mercedes and Volvo for example already use predecessors of Pegasus in their cars and Tesla’s Autopilot wouldn’t exist without Nvidia.
With its Metropolis platform Nvidia wants to connect observation cameras all over the city. That way, the system could warn in real-time when crime takes place or guide car drivers to a free parking spot. But it also could serve commercial goals.
Nvidia takes big steps, so it can’t be ignored. Also TomTom works together with Nvidia for his complicated mapping schemes. Even promising start-ups use Nvidia to develop other software, for example Perceptive Automata (American) who wants to score in face recognition.
Calculations are the key
Why is Nvidia becoming such a big player? “Because for autonomous cars you need a huge amount of data out of which the car can learn. Nvidia makes video programming cards since 25 years and they have to steer many different processes at the same time,” says Jaap Zuiderveld, sales manager for the EMEA region.
At the moment, Nvidia sits in the front row, but it doesn’t want to have a monopoly. CEO Huang: “Let 1000 flowers blossom. We are at the beginning, chaos is necessary then. Out of it will come a handful of interesting ‘drivers’, not a unique one.”
He gives Isaac as an example, a robot from Nvidia who learned hockey by itself. “We used a lot of Isaacs in the learning process. The one who won the first round, couldn’t win the second, etc. The same thing will happen when Artificial Intelligence (AI) starts to drive cars.”