April 12, 2021

NVIDIAs $60 Jetson Nano 2GB Is an Entry Point to Machine Learning and Robotics

A photo of the NVIDIA Jetson Nano 2GB board.
It’s like a Raspberry Pi, but with an emphasis on GPU power. NVIDIA

As NVIDIA continues to bulk up its RTX 3000 GPUs, the company is trimming the fat from its popular Jetson Nano developer kit. The new Jetson Nano sports just 2GB of RAM rather than the usual 4GB, but has the same CPU and a 4K video processor. At $60, it’s an unusually accessible platform for machine learning and robotics.

The Jetson Nano 2GB is a single-board computer. It’s quite similar to the Raspberry Pi, albeit with a heavier emphasis on GPU performance than CPU performance. And when you’re dealing with AI, you need all the GPU performance you can get.

Like the first Nano, the new 2GB model support 4K 60Hz video decoding, which is perfect for AI applications like object-identification or motion-tracking. It can also decode up to eight video feeds simultaneously, although the feeds will bottleneck at 1080p and 30Hz.

Hobbyists shouldn’t have any trouble getting started with the Jetson Nano 2GB, as it has the standard GPIO pins, USB ports, and HDMI-out that you expect from a modern single board computer. It also sports a Gigabit Ethernet connector and a built-in Wi-Fi module—a slight quality-of-life upgrade from the Jetson Nano 4GB.

NVIDIA’s 2GB Jetson Nano developer kit is available for preorder today and ships October 26th. Use it with NVIDIA’s Jetson AI Specialist certification program for a quick and robust introduction to AI and robotics.

Source: NVIDIA via ZDNet