The new NVIDIA Jetson Nano board is intended for Artificial Intelligence applications such as IoT or robotics. Its (small) size (70 x 45 mm) and its (small) price ($99) are impressive. This new system includes a SOM (System On Module) type module and a development board. It joins the Jetson range with its big brothers, Jetson TX2 and Jetson AGX Xavier.
The development kit offers the well-known Artificial Intelligence frameworks: TensorFlow, PyTorch, Caffe, Keras, and MXNet.
Nvidia has packed a lot of features in this module: 128 CUDA cores, a 4-core ARM Cortex processor, 4 GB 64-bit LPDDR4, 16 GB eMMC 5.1 Flash, a H.264/H.265 video encoder/decoder, an GigaEthernet interface, a HDMI 2.0, a microSD interface, 4 USB ports including 1 USB 3.0 and 1x SDIO / 2x SPI / 6x I2C / 2x I2S / GPIOs! Nano but not by the functionalities nor the power (472 gigaflops of computing power!).
The board is intended primarily for Edge computing, where data does not leave the system to go to the Cloud. This makes it possible to guarantee better autonomy and better controlled safety. Applications are mainly object recognition, simple autonomous navigation, smart input terminals or small drones.
Intended for both Makers and companies, it will be interesting to follow the adoption curve of this system, in direct competition with the latest Google modules from the Coral range (a bit low in memory in our opinion).