Nvidia’s autonomous car brain packs power of ‘150 MacBook Pros’
Nvidia has detailed its next-generation Drive PX 2 brain for autonomous production vehicles.
The advanced hardware has been developed with a focus on artificial intelligence and ‘deep learning’ principles, allowing the system to quickly monitor sensor data and compute the best trajectory.
“Drivers deal with an infinitely complex world,” observes Nvidia co-founder and CEO Jen-Hsun Huang. “Modern artificial intelligence and GPU breakthroughs enable us to finally tackle the daunting challenges of self-driving cars.”
The executive claims GPU-based hardware will eventually be capable of achieving ‘superhuman’ levels of situational awareness. The PX 2 brings a significant jump in processing power, equivalent to that of “150 MacBook Pros” and 10 times greater than the previous-generation product.
The lunchbox-size computer requires liquid cooling to chill its two discrete Tegra GPUs, capable of running 24 trillion ‘deep learning operations’ per second or eight billion traditional floating-point operations per second.
Deep-learning capabilities are said to be particularly well suited for processing unexpected challenges, such as erratic drivers and road debris. It is also claimed to be superior for operating in poor weather or lighting conditions that limit effectiveness of one or more sensor systems.
The PX 2 hardware can simultaneously process the inputs of a dozen video cameras alongside measurements from LiDAR, radar and ultrasonic sensors. The company has also created a suite of software tools, known as DriveWorks, that allows automakers to customize and improve autonomous driving algorithms.
Audi, BMW, Daimler and Ford are among a long list of companies already working with the first-generation Nvidia PX hardware. Volvo will be among the first automakers to test the PX 2 upgrade, which will power a fleet of 100 XC90 crossovers driven by actual customers involved in the company’s Drive Me autonomous-car pilot program.