Page 20 - EETEurope FlipBook February

20 EE|Times EUROPE — CES 2021
Mercedes-Benz: In-Cabin AI First, Software-Defined Car Next

Nvidia isn’t alone in its pursuit of enabling software-defined or “user
defined” vehicles.
NXP just announced BlueBox 3.0, a new development platform for
car OEMs looking to add a wider portfolio of user-defined applications
and AI (see “NXP BlueBox 3.0 Lets Carmakers Build Safety, Portfolio of
Apps,” below).
Nvidia has been on that path since late 2019, when it unveiled Orin.
According to the company, Orin is designed to handle myriad appli-
cations and deep neural networks while achieving systematic safety
standards such as ISO 26262 ASIL-D.
Leading up to CES 2021, Nvidia disclosed that leading Chinese EV
makers Nio, Li Auto, and Xpeng were developing their intelligent elec-
tric fleets on Nvidia Drive. Nio and Li Auto will use Orin in their future
vehicles; Xpeng is producing new EVs based on Nvidia’s Xavier SoC.
At a recent event, Nio unwrapped its ET7 sedan, slated to ship
in 2022. The car will feature Adam, a new Nvidia-powered super- The AI software learns each driver’s preferences and habits.
computer that uses Orin to deploy advanced automated driving (Source: Mercedes-Benz)
technology.
“Every carmaker that has a desire to develop a Tesla-like Junko Yoshida is global editor-in-chief of AspenCore. This article was
software-defined vehicle is turning to us,” said Nvidia’s Shapiro. ■ originally published on EE Times.

CES 2021 REVIEW
NXP BlueBox 3.0 Lets Carmakers Build Safety,

Portfolio of Apps

By Junko Yoshida

t January’s virtual CES, NXP Semiconductors launched Blue-
Box 3.0, an automotive development platform that doubles
the embedded computing power of BlueBox 2.0 and boosts
A I/O and PCIe connectivity eightfold.
The announcement underscores NXP’s platform approach. While
rivals are still engaged in the competition to offer more trillions of
operations per second (TOPS), NXP is pushing to alter the focus,
promoting a scalable development platform with which carmakers can
“experiment, develop, test out, and validate software and advanced fea-
tures that clearly enhance safety,” according to Ali Osman Örs, NXP’s
director of artificial-intelligence strategy and strategic partnerships for
automotive processing.

Carmakers have begun scaling back their
initial all-out pursuit of full autonomy.
Instead, they are investing more energy

and resources to enhance ADAS and other
features.
(Source: NXP Semiconductors)
By moving to BlueBox 3.0, “we’re also recognizing that there’s a lot
more variety in the type of compute and expansion needed,” Örs said acceleration, plugs into the motherboard via PCIe expansion cards.
in an interview with EE Times. The advantages of the BlueBox 3.0 platform center on its flexibility
Three basic building blocks of the BlueBox 3.0 platform are an and expandability. Increased PCIe slots let users plug in “elements like
LX2160A multicore processor, which is the highest-performance artificial-intelligence/machine-learning accelerator cards; additional
member of NXP’s Layerscape family; NXP’s latest gateway pro- vision-perception device cards, bringing in all the camera connectivity;
cessor (S32G274); and Kalray’s massively parallel processor array and radar and LiDAR, whose connectivity is traditionally based on a
(MPPA) processor. The first two sit on the development platform’s CAN or Ethernet,” said Örs. “Those can still come in through the regu-
motherboard. Kalray’s MPPA, which addresses heterogeneous AI lar CAN, Ethernet connectivity ports, but if there are added expansions

FEBRUARY 2021 | www.eetimes.eu