NVIDIA's new DRIVE PX 2 module is powered by two next-gen
Tegra processors. Image source: NVIDIA.
Long before this year's Consumer Electronics Show in Las
Vegas, we already knew
NVIDIA Corporation
was striving to ensure its technology would play a central role
in advancing self-driving cars.
When NVIDIA unveiled its 192-core Tegra K1 chip at CES two
years ago, for example, investors marveled as Audi showcased a
self-driving car that relied on the
supercomputer-esque power
of the tiny new processor contained in a small module in the
trunk.
Then at last year's show, NVIDIA launched the
first iteration of NVIDIA DRIVE PX
, an autopilot platform powered by two of its
even
more powerful Tegra X1 processors, and a slew of advanced
algorithms created by NVIDIA's computer vision engineering
team. The Tegra X1 for is part, packed over one full teraflops
of computing power, making it technically faster than ASCI Red,
the world's first teraflops system and the fastest
supercomputer up until the year 2000.
Deep learning on the road
But the graphics chip specialist has outdone itself this
year.
Earlier this week, NVIDIA kicked off CES 2016 by launching
NVIDIA DRIVE PX 2, describing its as "the world's first in-car
artificial intelligence supercomputer."
More specifically, NVIDIA DRIVE PX 2 is powered by a
combination of two next-generation Tegra processors and two
next-gen discrete GPUs based on NVIDIA's Pascal GPU
architecture. All that processing might is able to collectively
deliver up to 24 trillion "deep learning operations" per second
-- that is, in NVIDIA's words, "specialized instructions that
accelerate the math used in deep learning network
inference."
By that measure, NVIDIA DRIVE PX 2 packs more than 10 times
the computational power of the original DRIVE PX released one
year ago. And these specialized operations allow the vehicle to
more effectively identify and handle abnormal driving tasks
like road debris, erratic drivers, and construction zones.
Meanwhile, DRIVE PX 2's GPU architecture delivers up to 8
trillion general purpose floating point operations per second
-- or four times the GPU power of last year's DRIVE PX --
allowing NVIDIA's automotive partners to more effectively
implement the more predictable driving algorithms, including
localization, path planning, and fusion of data collected from
a multitude of sensors. On the latter, for example, DRIVE PX 2
can process inputs from as many as 12 video cameras, as well as
LiDAR, radar, and ultrasonic sensors, providing 360-degree
situational awareness around the car.
If you build it...
Of course, even the most superior technology is nothing without
partners willing to implement it. Luckily, NVIDIA has already
seen over 50 automakers adopt its AI platform for antonymous
driving development since launching DRIVE PX last year.
Volvo's XC90 self-driving SUV's will enjoy 360 degree
situational awareness thanks to NVIDIA DRIVE PX 2. Image
source: Volvo.
It would seem to follow, then, that these automakers should
have little hesitation continuing to move forward with DRIVE PX
2. To be sure,
Volvo
has already committed to be the first to implement DRIVE PX 2
using a fleet of self-driving Volvo XC90 SUV's which will hit
the road in a "public trial" next year.
Meanwhile, don't be surprised if additional automakers
announce their intent in short order; NVIDIA says the platform
will be available to early developmental partners in second
quarter of this year, and "generally available" by Q4.
As part of its end-to-end solution for partners, NVIDIA also
supplies a "deep learning" platform called DIGITS, which helps
train each system's deep neural network to better "understand"
the vast quantities of data provided by the vehicles'
sensors.
That's also not to say automakers haven't tried doing it on
their own.
"Using NVIDIA's DIGITS deep learning platform, in less than
four hours we achieved over 96 percent accuracy using Ruhr
University Bochum's traffic sign database," explained Matthias
Rudolph, Audi's director of Architecture Driver Assistance
Systems. "While others invested years of development to achieve
similar levels of perception with classical computer vision
algorithms, we have been able to do it at the speed of
light."
And according to Dragos Maciuca, technical director of
Ford
's Research and Innovation Center, "Deep learning on NVIDIA
DIGITS has allowed for a 30X enhancement in training pedestrian
detection algorithms, which are being further tested and
developed as we move them onto NVIDIA DRIVE PX."
NVIDIA's automotive growth and market potential
So where does that leave NVIDIA investors today?
As it stands, there are already more than 10 million
vehicles on the road featuring NVIDIA technology -- though
nearly all that technology is still in the form of high-tech
infotainment modules powered by NVIDIA's GPUs.
But even absent a significant contribution from self-driving
vehicle technology, automotive segment revenue still climbed
51% year over year to $79 million in NVIDIA's most recent
quarter. And though NVIDIA hasn't specifically outlined
expected automotive growth rates going forward, CEO Jen-Hsun
Huang did insist during last quarter's conference call that
they anticipate automotive will "continue to grow through next
year and a couple years after that."
To be fair, the young state of the self-driving car industry
makes it virtually impossible to forecast how growth will play
out with any reasonable degree of certainty. So the most NVIDIA
can do is continue to drive innovation to position itself
firmly at the center of the trend.
Thankfully for shareholders, it seems that's exactly what
NVIDIA is doing given the vast improvements it continues to
deliver in self-driving vehicle technology with each passing
year. In the end, that's why I remain convinced NVIDIA
stock is a great bet for investors with the foresight and
patience to buy early and profit as the self-driving car
narrative unfolds.
The next billion-dollar iSecret
The world's biggest tech company forgot to show you something
at its recent event, but a few Wall Street analysts and the
Fool didn't miss a beat: There's a small company that's
powering their brand-new gadgets and the coming revolution in
technology. And we think its stock price has nearly unlimited
room to run for early in-the-know investors! To be one of
them,
just click here
.
The article
Everything Investors Need to Know About
NVIDIA's New Self-Driving Car Tech
originally appeared on Fool.com.
Steve Symington
owns shares of NVIDIA. The Motley Fool recommends Ford and
NVIDIA
. Try any of our Foolish newsletter services
free for 30 days
. We Fools may not all hold the same opinions, but we all
believe that
considering a diverse range of insights
makes us better investors. The Motley Fool has a
disclosure policy
.
Copyright © 1995 - 2016 The Motley Fool, LLC. All rights
reserved. The Motley Fool has a
disclosure policy
.