Mapping NVIDIA’s Progress in Autonomous Driving


NVIDIA Brings Home Strong Growth for the Holidays PART 12 OF 15

NVIDIA’s journey to autonomous driving

Intel (INTL) is competing with NVIDIA (NVDA) not only in the AI (artificial intelligence) data center market but also in the autonomous vehicle space. In 2015, NVIDIA launched its first architecture for self-driving cars, the Drive PX. This platform would collect and process data from cameras and sensors to understand the car’s location, assess its outside environment, and anticipate potential hazards while driving.

Mapping NVIDIA’s Progress in Autonomous Driving

In January 2016, NVIDIA launched a more powerful next-generation platform, the Drive PX 2, which was equipped with a suite of software tools and libraries that support the deployment of autonomous vehicles. This supported level-3 automation.

NVIDIA has further accelerated its efforts in this space by launching its Xavier platform in September 2016, and after 13 months, it launched its Pegasus platform, which it claims can support level-5 autonomous driving.

The Xavier platform

The Xavier SoC (system-on-a-chip) processor is the AI brain behind autonomous driving. The SoC processor can collect and process sensor data and understand the outside and inside environment of the car.

The Xavier SoC has a 360-degree surround perception with features such as eye tracking, gesture recognition, and natural language understanding. The SoC notifies the driver of any potential danger outside while ensuring the state of the driver—for example, if he is distracted or drowsy.

The Drive Xavier can also act as a co-pilot, wherein it offers adaptive cruise control, lane keeping, and automatic lane changing. All these features are enabled through deep neural networks that can be updated via Wi-Fi or cellular connections. NVIDIA aims to make its Xavier platform powerful enough to power self-driving cars from regular carmakers such as Toyota Motor (TM).

NVIDIA’s Pegasus platform 

The technology advancements are happening faster than expected and automakers are competing to launch their autonomous car first. This pushed NVIDIA to launch its new system, the Pegasus, which the company claims can power completely autonomous vehicles.

NVIDIA claims that the Pegasus can deliver over 320 trillion operations per second, which is ten times faster than that of its predecessor, Xavier. The Pegasus would be made available to automakers in 2H18.

Notably, most automakers plan to skip level-3 and level-4 automation and jump directly to level-5, which is fully autonomous cars—a move that many believe to be dangerous or even reckless.

To avoid major accidents, NVIDIA stated that it would deploy these level-5 autonomous cars in a ride-hailing capacity only at first, and only in limited settings like college campuses or around airports. After confirming its life-saving potential, these cars could then be rolled out on public roads.

Author: || World Economic Forum

Author: administrator