Product of the Week: Aetina’s AIE-KX and AIB-MX Series of AI Inference & DeviceEdge Jetson Systems

September 26, 2022

Sponsored Story

Product of the Week: Aetina’s AIE-KX and AIB-MX Series of AI Inference & DeviceEdge Jetson Systems

Embedded systems are often deployed in challenging environments, such as a smart factory, smart city, or smart transportation. And now, in addition to being able to withstand those harsh environments, many of the applications they support now require high-accuracy, server-class AI inferencing on data collected in the field.

The AIE-KX series and the AIB-MX series AI Inference and DeviceEdge Jetson Systems from Aetina Corporation provide the aforementioned capabilities on the power of the NVIDIA® Jetson AGX Orin™ module that features 2048 NVIDIA® CUDA® cores and 64 Tensor Cores to deliver up to 275 TOPS of performance. There are also an additional 12 NVIDIA Arm® Cortex A78AE v8.2 64-bit CPU cores with 3MB L2 + 6MB L3 of cache on-chip.

 

An integrated fan helps keep all that horsepower cool enough to operate in extreme temperatures ranging from -25°C to 80°C and up to 95 percent humidity, where it can shuttle data from sensors over a either of two CAN 2.0b interfaces or from external cameras connected over the platform’s 16-lane MIPI expansion connector or four RJ-45 GbE PSE ports that support POE cameras. Those data captures would make their way to up to 64 GB of 256-bit LPDDR5 memory onboard the Jetson AGX Orin module at 204.8 GB/s, and from there could be analyzed by sophisticated inferencing algorithms executed on the awesome compute subsystem mentioned above.

The Aetina AIE-KX and AIB-MX also include 64GB of eMMC 5.1 flash storage and a range of other expansion options, such as a microSD slot, 1x M.2 M-Key 2280 slot for additional NVMe, and additional M.2 B- and E-Key support for LTE/5G and Wi-Fi/Bluetooth connectivity, respectively – backed by the platform’s CE/FCC Class A certification.

 

The native I/O and connectivity consists of five USB interfaces ranging from version 2.0 to 3.2 Gen 1, as well as a 10 GbE port for high-speed networking and HDMI Type A interface for connecting a multimedia display. A TPM 2.0 is also present.

Aetina’s AIE-KX and AIB-MX Series in Action

To support deployment in varied environments, Aetina’s AIB-KX and AIE-KX have a flexible input power range from 9 to 36VDC, deliverable over either a 4-pin DC jack or 6-pin ATX power supply. And by varied environments we mean almost anything an industrial engineer can throw at it, drop on it, or roll over it thanks to adherence to IEC 60068-2-27 and IEC 60068-2-64 shock and vibration standards.

For the safety-conscious, it’s also IEC 62368 certified.

 

Those ready to develop on either the AIB-KX and AIE-KX series will be happy to find a few UARTS for debugging and familiar buses such as two I2Cs, an I2S, and a SPI, as well as an RS-232 and RS-422/485 interface. There’s also access to five GPIOs.

And if you need to get your hands on the Aetina AIE-KX and AIB-MX from far away, support for Innodisk’s Out-of-Band Remote Management module can help you realize your IoT deployment goals.

Getting Started with the AIE-KX and AIB-MX Series: Software, OS Support, and Security

Those interfaces are your entry point to getting started with the Aetina AIE-KX and AIB-MX Series, which is backed by NVIDIA’s JetPack SDK development environment. The JetPack SDK provides an end-to-end software stack complete with bootloader, Linux or Ubuntu 20.04 OS, libraries, tools, and frameworks for building GPU-accelerated AI, computer vision, and other multimedia-centric systems at the edge.

Through the JetPack SDK, which is free to download, you can also access demos, sample code, documentation, and platforms like NVIDIA DeepStream, the Isaac for Robotics SDK, or RIVA Conversational AI to help jumpstart your application development even further.

To start building your own rugged AI edge system, go to https://aetina.com or check out the resources below.

Additional Resources: