Nvidia’s New TX2 Board Does Dual 4K-Camera Object-Detection in Real Time

Drones Drones & Vehicles Other Boards Robotics
Nvidia’s New TX2 Board Does Dual 4K-Camera Object-Detection in Real Time

Machine learning is complex, but nonetheless has pushed its way to professional and maker communities alike. Nvidia has lead much of this with their TK1 and TX1 modules; now, with the new release of the Jetson TX2, the AI capabilities we have access to have just doubled.

The new hardware, announced last night at a press event in San Francisco, retains the same form factor as the TX1 — roughly the size of a credit card, it’s meant as a drop-in replacement. It replaces the TX1’s Mawell GPU with a Pascal unit, doubles the TX1’s storage and memory, and increases its video encoding and decoding specs. With it, the company states that it can get either twice the performance of the TX1 (handling object detection and tracking from two 4K cameras simultaneously), or get double the efficiency running the same configuration as a TX1.

Putting unprecedented power into the GPU, we’ve seen the adoption of these Nvidia boards by the DIY autonomous car enthusiasts, from 1/10 scale DIY Robocar entrants to full-scale automobiles. But really, this is a board for professional, serious use. Nvidia is focusing the high-performance, low-power board  on “AI at the edge” — building artificial intelligence processing capabilities into products directly, rather than sending data to cloud-based super computers. During the TX1 announcement event, they described the benefits of this approach, including the processing of a projected 1 billion surveillance cameras by 2020, offering increased privacy to consumers, and giving AI capabilities to geographic regions where the necessary broadband speeds for cloud processing aren’t available.

With the TX2, Nvidia is also releasing an update of its software, Jetpack 3.0, with a stated 2x system performance.

For the release event, Nvidia brought 18 demos from various groups to show how they’re using TX architecture, including an onstage demo from Cisco of their new Spark Board collaboration tool. A large flatscreen display, it connected the Cisco rep with his companion Cisco team in Norway, who then showed a new camera device for the Spark Board that automatically recognizes and labels the names of the people in a conference, and automatically crops in to smaller groups to eliminate that “large empty conference room” aesthetic that we’ve become accustomed to with teleconferencing. Despite some bandwidth choppiness, the display and camera worked smartly and look like useful workplace tools.

Other demos included drones from MIT, EnRoute, and Teal, an Artec 3D scanner, assistance robots from Toyota and Fellow Robots, and a 4K stereoscopic 360º camera for live VR streaming from Live Planet.

Preorders for the TX2 are available now; it will begin shipping next week. Meant for professional applications, its power doesn’t come cheaply; a developer kit will be sold for $599. The TX1 will continue to be sold alongside it at a $499 price point.


Discuss this article with the rest of the community on our Discord server!
Mike Senese

Mike Senese is a content producer with a focus on technology, science, and engineering. He served as Executive Editor of Make: magazine for nearly a decade, and previously was a senior editor at Wired. Mike has also starred in engineering and science shows for Discovery Channel, including Punkin Chunkin, How Stuff Works, and Catch It Keep It.

An avid maker, Mike spends his spare time tinkering with electronics, fixing cars, and attempting to cook the perfect pizza. You might spot him at his local skatepark in the SF Bay Area.

View more articles by Mike Senese


Ready to dive into the realm of hands-on innovation? This collection serves as your passport to an exhilarating journey of cutting-edge tinkering and technological marvels, encompassing 15 indispensable books tailored for budding creators.