During the CES 2016 Presentation, Nvidia surprised attendees with the revelation of the Drive PX 2, the world’s first in-car supercomputer.
One of the main holdbacks of self-driving cars is the inability to capture and learn from the millions of hours of road time from daily drivers in the world and the complexity of the infinite road. Nvidia hopes to help bridge this gap with their new Nvidia DRIVE PX 2 platform.
The DRIVE PX 2 is a complete deep-learning platform designed to capture, analyze, and learn from every day road conditions and its infinite variables to build a superhuman level of situational awareness. Utilizing two next-generation Tegra processors and two discrete Pascal GPUs, the DRIVE PX 2 can deliver up to 24 trillion deep learning operations per second. With this computational power, the DRIVE PX 2 can quickly learn the challenges of everyday driving, such as road debris, erratic drivers, and construction zones.
The DRIVE PX 2 can process 12 video cameras, LIDAR, RADAR, and ultrasonic sensors to create an accurate situational map of their driving environment and create optimal decisions for safe travel.
Nvidia DRIVE PX 2 relies on the NVIDIA DIGITS supercomputer to help train its neural networks with data from other units to build a sophisticated learning network.
Currently, in partnership with automated driving company AdasWorks, Volvo plans to deploy a fleet of XC90 SUVs in 2017 with many more manufacturers working on their own development projects soon.