Antmicro's Open Hardware Baseboard Aims to Broaden Outreach of New NVIDIA Jetson Nano Edge AI Platform
Topics: Open HW, Edge AI
Earlier today at the opening keynote of GTC Silicon Valley 2019, Jensen Huang, founder and CEO of NVIDIA, announced a new Jetson product, the Jetson Nano. As a Jetson ecosystem partner widely known for its early adoption of all members of the Jetson family, Antmicro is participating in the event and releasing a unique, fully open source Jetson Nano baseboard the day NVIDIA debuted its latest brainchild on the keynote stage.
Jetson Nano is a compact AI computer that aspires to revolutionize a much broader market by targeting entirely new applications including low-cost NVRs, home robots, and intelligent gateways with full analytics capabilities. Addressing the size, power and AI compute density constraints faced by many companies, the smart design of the small but mighty Jetson Nano module promises to reduce overall development time and bring end products to market faster. Today you can buy the developer kit for $99, and the production module will be available for $129 in a few months.
Jetson Nano is supported by NVIDIA JetPack, which includes a board support package (BSP), Linux OS, NVIDIA CUDA, cuDNN, and TensorRT software libraries for deep learning, computer vision, GPU computing, multimedia processing, and more. It comes with a 128-core NVIDIA Maxwell GPU, a quad-core ARM Cortex-A57 processing system.
Antmicro has been helping customers take advantage of the capabilities of the heterogeneous Jetson platforms for years across dozens of projects, offering design services for AI applications, BSP development (including camera drivers), and customised hardware. Through Jetson Nano, NVIDIA is reaching out to new, broad industrial markets previously targeted by other solutions, where customers had to settle on platforms that could not provide the compute needed to do modernAI. Based on many years of experience with both the Jetson platform and a deep understanding of the needs of customers in those markets, Antmicro is in the perfect position to help accelerate the development of early adopter products looking to do edge AI in a compact form with the Jetson Nano.
In readiness for this new approach, Antmicro has prepared a dedicated baseboard to allow NVIDIA customers looking to develop right away to leverage the possibility of introducing state-of-the-art edge AI in completely new, mass-market applications. And by making this baseboard completely open source, we’re adding another valuable asset to Antmicro’s university outreach programme, enabling academic researchers and students globally to use - and freely modify - an affordable, yet bleeding-edge-powerful computing platform for AI at the edge.
Antmicro’s Jetson Nano Baseboard is compact and combines a set of typical IO interfaces such as Gigabit Ethernet, HDMI, DisplayPort or USB host (exposed with a USB-C socket) with unified MIPI CSI-2 camera interfaces compatible with other Antmicro video board accessories. The board exposes all of the CSI data lanes that may be configured in order to support up to 4 x 2-lanes or up to 3 x 4-lanes. A built-in power sequencer combined with high-performance DC/DC power supply circuitry makes it operational over a broad range of power supply voltages.
The code is now available on Antmicro’s github.
If you’re interested to learn more and see how Antmicro can help you design your next mass-market device based on Jetson Nano, write to email@example.com.