Harsh-environment embedded computing board for artificial intelligence (AI) applications introduced by WOLF

Single-board computer offers NVIDIA Jetson AGX Xavier module and the WOLF FGX for applications that need NVIDIA GPGPU CUDA processing like machine vision.

Wolf 10 June 2019

WHITCHURCH-STOUFFVILLE, Ontario – WOLF Advanced Technology in Whitchurch-Stouffville, Ontario, is introducing the VPX3U-XAVIER-SBC 3U VPX single-board computer for use in military, aerospace, and other harsh-environment applications.

The VPX3U-XAVIER-SBC includes an NVIDIA Jetson AGX Xavier module and the WOLF FGX, and is suited for applications that require artificial intelligence (AI) inferencing, machine vision, or NVIDIA general-purpose graphics processing unit (GPGPU) CUDA processing.

The embedded computing board offers autonomous operation and a power draw of less than 50 Watts. "The combination of 4K video, high-speed LAN, excellent SWaP, PCI Express with non-transparent bridging, and single-slot autonomous computing for AI and computer vision at the edge opens up numerous systems integration options for defense and aerospace," says Greg Maynard, chief technology officer at WOLF. “Including the WOLF FGX in the module also allows for expanded video format support, for example, SDI, CVBS, STANAG 3350, ARINC 818, and other video formats.”

The NVIDIA AGX GPGPU provides advancements in edge and autonomous computing, without the need for continued cloud connectivity. It enables local visual odometry, sensor fusion, localization and mapping, obstacle detection, and path planning algorithms for next-generation autonomous platforms.

Related: WOLF video graphics embedded computing modules help design windowless supersonic aircraft cockpit

Developers using this module can benefit from NVIDIA’s artificial intelligence (AI) tools and workflows as well as the NVIDIA JetPack software-development kit.

Features include NVIDIA Volta GPU with 512 CUDA cores and 64 Tensor Cores, ; NVIDIA Carmel 8-core ARMv8.2 64-bit CPU; two NVIDIA Deep Learning Accelerator (DLA) engines; as much as 32 TOPS of deep learning inference computing using the GPU and Deep Learning Accelerator (DLA) engines; two NVIDIA Vision Accelerator engines; 16 gigabytes of 256-bit LPDDR4 memory; and wo NVENC / NVDEC hardware-based video codecs for H.264/H.265.

The 3U VPX WOLF-12TP module includes factory configuration options to solve system integration challenges. This module has been designed to support VPX REDI (VITA 48) and OpenVPX (VITA 65).

In addition, the WOLF-12TP includes a WOLF FXG to support additional video formats that are not native to the Xavier GPU. This also allows WOLF to react quickly to customer requirements for additional video formats. Early access units of the computer board are available for order, and production will begin in September 2019.

For more information contact WOLF Advanced Technology online at https://wolfadvancedtechnology.com.

More in Computers