Nvidia has revealed a self-driving car powered by its new AI supercomputer known as Xavier. The Xavier system can learn how to drive by observing a human driver, and Nvidia installed the AI in an autonomous Lincoln vehicle to demonstrate its capabilities.
Nvidia, best known for its video game graphics cards, developed its new Xavier system-on-chip (SoC), integrating Volta architecture, a custom octa-core CPU, and a new “computer vision accelerator,” specifically for use in self-driving cars.
The Xavier system employs deep learning and cloud-based map data to support its navigational capabilities. It’s connected to a set of cameras which monitor the environment, processing data like the status of traffic lights and road signs in real-time.
During Nvidia’s CES press conference, where the project was unveiled, CEO Jen-Hsun Huang said, “This is the greatest SoC endeavor I have ever known, and we have been building chips for a very long time.”
Nvidia says that Xavier SoC samples will be available in Q4 2017 for “automakers, tier 1 suppliers, startups and research institutions” involved in the development of self-driving cars.