Follow
Subscribe

NVIDIA CEO Reveals DRIVE SIM, a VR Autonomous Driving Simulator at GTC 2018

Home > News > Content

【Summary】NVIDIA CEO Jensen Huang took the stage this morning to deliver an important keynote kicking off NVIDIA’s annual GTC Technology Conference in Silicon Valley. Among the highlights was his announcement of an update to NVIDIA DRIVE, the AI platform for autonomous driving, which now includes a autonomous driving simulator to support autonomous driving development.

Eric Walz    Mar 27, 2018 1:09 PM PT
NVIDIA CEO Reveals DRIVE SIM, a VR Autonomous Driving Simulator at GTC 2018

SAN JOSE, Calif., — Nvidia CEO Jensen Huang took the stage this morning to deliver an important keynote kicking off NVIDIA's annual GTC Technology Conference in Silicon Valley. Among the highlights was his announcement of an update to NVIDIA DRIVE, the AI platform for autonomous driving, which now includes a virtual reality autonomous driving simulator to support autonomous driving development.

GTC brings together a worldwide group of software engineers, developers and companies working on Deep-learning and AI advancements. Nvidia's 2018 GTC is the company's largest so far, with nearly 8000 registered attendees.

During this morning's opening keynote to a packed audience of developers, journalists and other industry professionals, Huang announced The new NVIDIA DRIVE Constellation, a cloud-based system simulator for autonomous driving using photorealistic simulation. Constellation offers a safer way to test self-driving cars, to prevent accidents like the one involving Uber in Arizona last week.

"Safety is really the most important thing, as we were reminded just last week,"  Huang said, referring to the Uber incident.

"We at Nvidia are dedicating ourselves to solving this problem. It's the ultimate high performance computing problem to enable safe autonomous driving."

The Constellation simulator will help Nvidia reach its goal, as million, or billions of miles can be "driven" in simulation, rather than trying out deep-learning algorithms alongside other drivers in the real world.

IMG_20180327_111131.jpg

Constellation is a new solution for the data center as well. Huang said the platform uses two different servers.

"We have to train our neural network with extreme precision for autonomous driving," Huang said. Having multiple servers helps with this.

The first server runs the NVIDIA DRIVE simulation software to simulate a self-driving car's sensor suite, including cameras, LiDAR, and radar. The second server contains the NVIDIA DRIVE Pegasus AI car computer that runs the entire software stack in an autonomous driving vehicle. This data is processed just like it is actually coming from the sensors of a car on the road—at speeds of 30 times per second.

Driving commands from DRIVE Pegasus are fed back to the simulator, completing the digital feedback loop. This unique, "hardware-in-the-loop" cycle repeats times per second and is used to validate the algorithms and software running Pegasus, to ensure it's operating the simulated self-driving vehicle correctly.

The simulation server is powered by NVIDIA GPUs, each one generating a stream of simulated sensor data, which then feeds into the DRIVE Pegasus for processing.

"Deploying production self-driving cars requires a solution for testing and validating on billions of driving miles to achieve the safety and reliability needed for customers", said Rob Csongor, vice president and general manager of Automotive at NVIDIA.

"With DRIVE Constellation, we've accomplished that by combining our expertise in visual computing and data centers. With virtual simulation, we can increase the robustness of our algorithms by testing billions of miles of custom scenarios and rare corner (edge) cases, all in a fraction of the time and cost it would take to do so on physical roads."

PhotoRealistic Images

One major advantage of DRIVE Sim software is photorealistic images. It can simulate a rainstorm, or bright sunlight at different times of the day, or complete darkness. Driving in these conditions can be scripted, to test an autonomous vehicle's ability to react, without putting other drivers or pedestrians in danger on public roads.

NVIDIA_DRIVE_Sim (1).jpg

The Image quality produced by Nvidia's Constellation simulator

"Autonomous vehicles need to be developed with a system that covers training to testing to driving." siad Luca De Ambroggi, research and analyst at IHS Markit. "NVIDIA's end-to-end platform is the right approach. DRIVE Constellation for virtually testing and validating will bring us a step closer to the production of self-driving cars."

Huang said DRIVE Constellation with be available in the third quarter of this year. Currently, over 370 partner companies and about 200 startups are using various versions of NVIDIA's DRIVE platform to develop their self-driving technology.

Prev                  Next
Writer's other posts
Comments:
    Related Content