CES 2016 : NVIDIA Announces First In-Car AI Supercomputer for Autonomous Driving
The 2016 edition of the Consumer Electronics Show is currently underway in Las Vegas, and we have some of the early news regarding autonomous vehicle tech. NVIDA has announced the first in-car artificial intelligence supercomputer, capable of 24 trillion computations a second. It is a big step towards a improved thinking power for self-driving cars. The supercomputer is called PX 2 and according to a press release from the company, is “the size of a lunchbox and with the computing capability of 150 MacBook Pros.” It features two Tegra processors, two discrete graphics processing units (GPUs), and is able to process 2,800 images a second. RELATED: Proposed California DMV Rules Spell Trouble for Autonomous Car Development
And why would you need that sort of image processing power? Because autonomous cars that are in development now rely on a series of cameras and sensors to not only stay on the proper path, but also account for other cars, obstacles, errant bicyclists, and anything else that might pop up on the road that the average human driver has to account for. It is claimed that PX 2 can process the data from 12 video cameras, LIDAR, RADAR, and ultrasonic sensors to create a real-time awareness of its surroundings on the road.
RELATED: Five Things to Know About Rinspeed's Self-Driving BMW i8
The the recent spat of Google cars getting into fender benders (even though Google places the blame with the other drivers) shows that there is work to be done when it comes to autonomous vehicles being able to account for all of its surrounding. Boosting the processing power is a big step towards creating self-driving vehicles that work effectively enough that the public will actually trust them.
Volvo has announced it will test the new NVIDIA PX 2 in 100 XC90 SUVs. It will make these high-tech, self-driving SUVs available for public use in 2017.
RELATED: See more images of the 2016 Volvo XC90