Maker Board Spotlight: NVIDIA Jetson Nano Developer Kit
If you have been working on complex high-end data-intensive AI applications, it is very likely that you have come across NVIDIA Jetson Nano Developer Kit. As a part of AI accelerators designed to operate at optimal conditions on heavy AI workloads, NVIDIA Jetson Nano Developer Kit has captured the market as the cost is very reasonable for the amount of data-processing it can perform. In the last decade, AI at the edge (which is nothing but embedded machine learning) has been growing at an exponential rate, and there are already several pieces of hardware available fulfilling the requirements. Even with its high demand for an easy-to-use and powerful AI platform for makers and embedded developers, NVIDIA Jetson Nano Developer Kit seems to become a benchmark for many common TinyML applications.
NVIDIA believes that the hardware brings AI computing to everyone through its powerful Quad-core ARM Cortex-A57 processor running at a clock frequency of 1.43GHz integrated with 128-core Maxwell GPU making it one of the most compact and powerful computers for running multiple neural networks in parallel. Announced back in 2019, NVIDIA Jetson Nano Developer Kit delivers 472 GFLOPS of computing performance through its CPU and NVIDIA GPU. The robust software support makes it one of the best choices for critical AI applications even for accelerated graphics. The ecosystem and developed software environment make it easy to bring real-time computer vision and inference of complex DNN models to the edge. The onboard functionalities have helped the hardware to top the leaderboard for IoT edge analytics and advanced AI systems.
What is NVIDIA Jetson Nano Developer Kit?
NVIDIA Jetson Nano Developer Kit is a compact, easy-to-use, power AI computer that lets you run common AI tasks like image classification, object detection, and speech processing at low-power consumption of as less as 5 watts. The embedded electronic device, Jetson Nano Developer Kit is built around the native NVIDIA CUDA-X which is a collection of libraries, tools, and technologies that aim to deliver high performance for artificial intelligence to high-performance computing workloads.
NVIDIA JetBot open-source deep learning autonomous robotics kit based on Jetson Nano [Image Credit: NVIDIA Website]
Former Communications Manager, Nefi Alarcon published the news on the release of Jetson Nano Developer Kit and its module says, “The kit comes with out-of-the-box support for full desktop Linux, compatibility with many popular peripherals and accessories, and ready-to-use projects and tutorials that help makers get started with AI fast.” “Jetson Nano supports high-resolution sensors, can process many sensors in parallel and can run multiple modern neural networks on each sensor stream.”
NVIDIA Jetson
Nano Developer Kit Specifications
Let us take a look at the specifications of the NVIDIA Jetson Nano Developer Kit
- CPU: ARM Quad-core Cortex-A57 processor core clocked at up to 1.43GHz
- GPU: 128-core NVIDIA Maxwell architecture-based GPU
- Memory: 4GB LPDDR4 at 25.6 gigabytes/second transfer speed
- Storage: microSD card storage slot
- Connectivity: Gigabit Ethernet, M.2 Key E
- Camera connector: 2x MIPI CSI-2 DPHY lanes
- Display interface: HDMI and display port
- USB: 4x USB 3.0 and USB 2.0 Micro-B
- Video encode: 4K at 30 fps (H.264/H.265)
- Video decode: 4K at 60 fps (H.264/H.265)
- Software: Linux for Tegra
- Connectors: GPIO, I2C, I2S, SPI, UART
- Dimension: 70x45 mm (Approximately)
Note: NVIDIA has already introduced a smaller version of the developer kit that comes with 2GB of LPDDR4 memory. Moreover, there are several versions with minor changes to the hardware with interfaces and connectivity.
Add-on Boards for NVIDIA Jetson Nano Developer Kit
There are many add-on boards for this popular piece of hardware, however, today we will be discussing external cameras for computer vision and grove modules for other external sensors.
1.IMX219-130 8MP Camera
This is a high-quality camera with an 8-megapixel Sony IMX219 image sensor capable of viewing images at a high resolution of 3280x2464 pixels. The field of view of 130 degrees makes it suitable for machine vision projects that can capture better quality videos from the camera. There are several other options with different diagonal FOV and minor tweaks in the specifications. The camera is compatible with Jetson Nano and Xavier NX Developer Kit.
Specifications of IMX219-130 8MP Camera
- Image sensor: Sony IMX219
- Megapixels: 8 megapixels
- Resolution: 3280x2464 pixels
- Pixel size: 1.12x1.12µm
- Aperture: 1.8
- Focal length: 1.88mm
- FOV: 130 degrees
- Maximum image transfer rate: 30 fps for QSXGA
- Lens size: 6.5x6.5 mm
- Operating temperature: 20°C to 70°C
- Dimension: 25x24x14 mm
2.Grove Sensors for Jetson Nano
Interestingly, along with being the official distributor for the hardware, Seeed Studio has also supported the hardware through their standard Grove sensors with grove.py Python library. Using this library, the developer can get the sensors up in running within minutes and there are more than 20 grove modules ready to be interfaced with Jetson Nano. To connect grove modules to the hardware, you will require the Base HAT for Raspberry Pi.
How to get started with NVIDIA Jetson Nano Developer Kit?
The basic requirement is to get all the hardware ready. Most importantly, you will require a MicroSD card for storage. Along with that, you will need to power the developer kit with a power supply that can deliver 5V@2A through the Micro-USB port. Subsequently, you will now have to prepare the microSD card and connect that to your computer with an internet connection. There is a detailed guide for various operating systems, Windows, macOS, and Linux.
NVIDIA Jetson Nano Developer Kit Set-up [Image Credit: NVIDIA Website]
Head to the official getting started guide for more details and graphics. Once you are ready with the initial setup, it’s time for setting up Jetson Nano with JetPack. JetPack is a comprehensive software development kit that is designed for AI and computer vision applications and contains drivers for CUDA toolkit, cuDNN, TensorRT, OpenCV, VisionWorks and multimedia APIs. More details on this can be found on the GitHub repository.
Final thoughts on NVIDIA Jetson Nano Developer Kit
For beginners, the manufacturer has also developed a Jetson AI course and certification that gives you the credibility and skills to design advanced projects. The hardware requirements to enroll for the course are the Jetson Nano Developer Kit, a MicroSD card for storage, Micro-B USB cable and power supply. With this setup, you are ready to get started with the hardware.
All these powerful AI features in a compact form factor offered at just $118.75 (at time of writing) makes Jetson Nano Developer Kit an interesting piece of hardware to consider for your complex AI applications.
Abhishek Jadhav is an engineering student, freelance tech writer, RISC-V Ambassador, and leader of the Open Hardware Developer Community.
Leave your feedback...