Product of the Week: Syslogic AG AI Embedded PC RS A3N
March 15, 2021
Blog
Computer vision, AI inferencing, and sensor fusion are being deployed into industrial use cases to increase the productivity and efficiency of applications ranging from smart cities and agriculture to transportation and construction vehicles. These functions often demand the supercomputing-class performance available on modules like the NVIDIA Jetson Xavier NX, but must be executed on rugged, reliable systems that can withstand the rigors of harsh environments.
According to Syslogic, their AI Embedded PC RS A3N is the first embedded system based on the Jetson Xavier NX that brings 21 TOPS of INT8 horsepower via a 384-core Volta™ GPU, 48 Tensor cores, and a 6-Core ARM v8.2 64-bit NVIDIA Carmel CPU. Designed for industrial-grade AI inference, CV, and sensor fusion tasks, the fanless industrial box PC is encased in a protective aluminum sealed housing rated to IP67 or IP69 ingress protections.
In addition to resisting dust and liquid contaminants, the enclosure helps protect the RS A3N supercomputer from high levels of shock and vibration. Temperature supervision helps manage the platform’s thermal load in extreme deployments, while an integrated hardware watchdog monitors system components to ensure reliable operation in use cases that require continuous 24/7 operation.
The Syslogic AI Embedded PC RS A3N in Action:
The NVIDIA Jetson Xavier NX module in the AI Embedded PC RS A3N sits on a carrier board developed and manufactured by Syslogic. The carrier transmits signals out through interfaces including:
- 2x 10/100/1000 Mbps Ethernet with rugged screw-on connectors
- 1x CAN ESD (Active/Passive) with rugged screw-on connectors
- Serial (RS-232, I2C)
- 6x USB (2x 2.0, 1x 2.0 OTG, 1x 3.1)
- 2x PCIe signals
- 1x DisplayPort 1.4
- 1x HDMI 2.0
- MIPI CSI-2, GMSL2, or FPDLinkIII Camera Interface (Optional)
GNSS, LTE, and Wi-Fi connectivity are provided on the B102S variant and optional on the A102S, while an additional USB 2.0 port and GNSS module can also be added.
It is through these interfaces that sensor input data makes its way into the RS A3N’s 8 GB of soldered-on 128-bit LPDDR4x RAM (M.2 expansion can extend flash memory up to 2 TB). From there, inputs are run against perception, recognition, and detection algorithms by the massively parallel Jetson Xavier NX SoC, enabling real-time autonomous decision making at roughly 10 W of total platform power consumption.
These perception, recognition, and detection workloads can be optimized for on the Xavier NX-based RS A3N system using accelerated libraries and APIs for deep learning that are available as part of NVIDIA’s CUDA parallel programming toolkit.
Getting Started with the Syslogic AI Embedded PC RS A3N:
In addition to the CUDA toolkit, engineers working with the Syslogic AI Embedded PC RS A3N can leverage other components of the NVIDIA JetPack software development kit (SDK) such as the Linux4Tegra (L4T) operating system, the TensorRT runtime, cuDNN libraries, the DeepStream SDK, real-time operating system support, and other tools.
When combined with the pre-installed board support package (BSP), the JetPack SDK provides developers working with the Syslogic AI Embedded PC RS A3N a quick path to market by eliminating the need to start developing drivers and application software from scratch.
The Syslogic AI Embedded PC RS A3N is available now, and customization options are available. To get started, visit www.syslogic-group.com or check out the resources below.
Additional Resources:
- Syslogic AI Embedded PC RS A3N Product Page – www.syslogic-group.com/eng/ai-embedded-pc-rs-a3n-nvidia-jetson-xavier-nx-111092.shtml?parentPageId=111605
- Syslogic AI Embedded PC RS A3N Data Sheet – www.syslogic.com/ProductDocuments/deu/DS_IPC_COMPACTA3N_RS_AI_v0.3.pdf
- NVIDIA JetPack SDK – developer.nvidia.com/embedded/jetpack
- NVIDIA CUDA Toolkit – developer.nvidia.com/cuda-toolkit
