Updated NVIDIA® Jetson Nano / Xavier NX open source baseboard with M.2 PCIe connector
Topics: Open hardware, Open machine vision, Edge AI
Antmicro’s highly successful Jetson Nano™ / Xavier™ NX open source baseboard has reached two new important milestones: increased extensibility thanks to the newly-added M.2/PCIe slot, and compatibility with the new and exciting Jetson Xavier NX SoM from NVIDIA that reached the market just this month.
Combine Jetson with multi-port Ethernet, FPGA, external accelerators
Owing to the M.2 connector with the M key, the updated version now offers more extension possibilities, enabling multi-Gigabit Ethernet connectivity, plugging in external AI accelerators, FPGA processing cards, LTE modems, etc.
The M.2 slot greatly increases the number of potential applications and use cases for the board. We can help you rapidly create a prototype of your target application using external M.2 cards (or PCIe, with the relevant extender) and then roll it into one design using our custom hardware development services.
Keep an eye out for upcoming blog posts about the integrations and advanced products we created based on this platform.
Compatibility with Jetson Xavier NX and cameras
The new version of our baseboard supports the NVIDIA Jetson Nano SoM as well as the brand-new Xavier NX SoM, which can easily process data from the board’s 4 camera interfaces thanks to its unprecedented deep-learning capabilities enclosed in a small form factor. We have successfully run the Xavier NX modules we received on the latest revision of our open hardware board, and tested video input with OV5640 and Allied Vision’s 1500 C-500 cameras. This means we can add preliminary support for Xavier NX to the already impressive hardware lineup possible to use with the plug-and-play Alvium cameras we are working on with Allied Vision.
If you missed the news, the Xavier NX is a smaller version of the powerful AGX Xavier. Almost as capable for deep learning applications as its bigger brother, the Xavier NX SoM has a lot to offer with its 6-core CPU, NVIDIA Volta GPU and a maximum of 21 TOPS of deep learning performance. On top of that, its small footprint, reduced power envelope and a much more affordable cost - 399 USD - make it a really interesting counterpart of the heavy-duty AGX Xavier. Compatible with the very inexpensive Jetson Nano, the Xavier NX offers an excellent migration path for product variants that might require different levels of compute power depending on the circumstances.
In fact, some custom hardware and software projects based on the Xavier NX are already underway at Antmicro, as several of our customers were interested in leveraging the new capabilities of the Nano / NX platform.
Great starting point for space-constrained, edge AI capable devices
Our aim behind developing this open source Jetson Nano / Xavier NX platform was to provide an ideal starting point for the demanding engineering projects that we carry out for our customers. With a footprint as little as 55 x 100 mm, it perfectly suits high-end but space-constrained devices for applications such as robotics, on-board computers and compact smart cameras.
So far we have used the Jetson Nano / Xavier NX baseboard for projects involving stereovision, data fusion with synchronous camera triggering, integration with external MCUs, IMU, motor control, FPGA, radar and other sensors, or to enable camera interfaces such as SDI. Using the platform, we have been implementing OTA update systems, reproducible and traceable BSPs using OpenEmbedded, hardware-in-the-loop Continuous Integration, ROS environments, containerization and - of course - deep learning for our customers.
Features and summary
Summarising, the key features of the Jetson Nano / Xavier NX baseboard are:
- Jetson Nano / Xavier NX SO-DIMM connector
- Gigabit Ethernet RJ45 connector
- Micro USB debug connector
- USB-C DFP (Host) port
- Mini HDMI video output interface
- Mini DisplayPort video output interface
- 2 x 50-pin FFC connectors exposing MIPI CSI-2 camera interfaces
- M.2 (key M) connector exposing PCIe x4 interface
- RTC battery backup
- EEPROM for storing board ID/SN
We are continuing to deliver engineering projects based on the updated version of the platform, helping our customers get the most of the Jetson platform and put their products out to market faster. If you are building your next product on the Jetson Nano or Xavier NX, please write to us at email@example.com to find out how we can assist you.