In recent years, advances in AI have produced algorithms for everything from image recognition to instant translation. But when it comes to applying these advances in the real world, we're just getting started. A new Nvidia product announced today at GTC, an $ 99 AI computer called Jetson Nano, should help accelerate that process.
The Nano is the latest in Nvidia's line of integrated Jetson computer cards, which is used to provide the brains of robots and other AI-powered devices. Connect one of these to your latest creation and you can handle tasks such as object recognition and autonomous browsing without relying on the processing power in the cloud.
This type of configuration is known as perimeter computing and, as it means that the data that is processed from cameras and microphones never leave the device, the final result is usually faster, more reliable and more secure hardware. So everyone wins. Previous Jetson boards have been used to power a variety of devices, from Lowe & # 39; s rack scanning robots to Skydio autonomous drones. But the Nano points even smaller.
Nvidia is launching a Nano development kit aimed at "integrated designers, researchers and DIY creators" for $ 99 and production modules ready for commercial companies for $ 129 (with a minimum purchase of 1,000 modules).
The company also presented a fun DIY project for any advanced manufacturer: a $ 250 open source, stand-alone robotics kit called JetBot (above). It includes a Jetson Nano along with a robot chassis, a battery pack and motors, which allows users to build their own automatic driving robot.
With the $ 99 devkit, you get 472 gigaflops of computing with a four-core ARM A57 processor. , Nvidia Maxwell GPU of 128 cores and 4 GB of LPDDR RAM. The Nano also supports a variety of popular AI frameworks, including TensorFlow, PyTorch, Caffe, Keras, and MXNet, so most algorithms will be virtually plug-and-play. And there is the usual support of ports and interfaces, including USB-A and B, Gigabit Ethernet and support for microSD storage.
Nvidia says it expects the Nano price point to open AI hardware development to new users. "We hope that a large part of the manufacturing community wants to join the IA, but in the past it could not, the Jetson Nano will allow them to do so," Nvidia vice president and general manager of autonomous machines, Deepu Talla, told reporters. in an informative meeting
It is true that the Nano has a competitive price, although it is not the only one. Intel, for example, sells its Neural Compute Stick for $ 79, while Google recently introduced two similar devices under its Coral brand: a $ 150 devkit and a $ 75 USB accelerator. But if this proves anything, it's that Nvidia is entering a fertile market. Let's see that creations driven by AI begin to grow.