Nvidia, a visual computing company based in Santa Clara, Calif., introduced its new Tegra X1 mobile phone chip Sunday at the Consumer Electronics Show (CES) in Las Vegas, as The Independent reports. This chip, in many ways more powerful than leading desktop computers, could eventually be used to power self-driving cars. “Two of the chips built in to a car would be able to recognise 75 objects at the same time, the company said, helping the car to understand its environment,” The Independent writes.
The chip will be a key part of Nvidia’s Drive PX platform, which will help companies to build self-driving cars, The Independent writes. The Drive PX car computers platform is an autopilot system that can process data from as many as 12 cameras at a time, writes Chris Wood for Gizmag. That will allow for what Wood calls “true self parking.”
Drive PX is a next step in the evolution of such technologies as sensors, cameras, and radar, writes Electronista. The PX platform contains “deep neural network computer vision,” which makes it possible for Drive PX to learn, the same machine learning process used in artificial intelligence, Electronista writes.
Whereas previous systems recognized items straight in front and in full view, Drive PX recognizes and learns items that are partly hidden. It can even distinguish among different types of items, such as vans, heavy trucks, SUVs, and sports cars, Electronista writes. “It could also detect a person partially hidden behind a car, before reclassifying him as a cyclist once in full view,” Electronista writes. Data for any object that is not identifiable by the system is uploaded to a cloud-based system for further learning and classification, and shared with all other systems, Electronista writes.
On Nvidia’s blog, Brian Caulfield writes that Tegra X1 is scheduled to arrive in products in the first half of this year. At CES, Nvidia’s CEO Jen-Hsun Huang said that in the future, cars will be “the most advanced computers” on Earth, Caulfield writes. In fact, they will be more powerful than a supercomputer “the size of a suburban family home from 15 years ago,” Jen-Hsun said.
In another Nvidia blog post, Danny Shapiro writes that Drive PX will make it possible for a car to park itself by “giving cars the brains to take over one of driving’s biggest headaches.” The system does that by using complex algorithms to convert images from cameras on the front, rear, and sides of a car into a “real-time obstacle map” as the vehicle drives on its own through a parking garage. Drive PX controls the car’s throttle, brake and steering as it moves to find a spot to park in.