This article appeared in Microwaves & RF and has been published here with permission.
Check out our AUVSI Xponential 2023 coverage. This article is also part of the TechXchange: AI on the Edge.
Looking for a rugged, wireless, IP67-qualified platform to deploy artificial-intelligence (AI) applications on the edge? Then check out Elma’s JetSys-5330 (see figure), which is built around NVIDIA’s Jetson AGX Orin system-on-module (SOM). The platform is suitable for a range of applications like intelligent-virtual-reality (IVR) and augmented-reality (AR) applications that can take advantage of the Jetson’s machine-learning (ML) support.
Equipped with MIMO antennas, wireless functionality includes Wi-Fi and 4G/5G LTE. On the wired side, there’s support for 10G BASE-T Ethernet. The system can handle half-a-dozen GMSL2/GMSL1 cameras in x2 or x4 CSI mode. In addition, the hardware is able to manage up to three 1000BASE-T Power-over-Ethernet (PoE) cameras.
The carrier board features three Mini-PCIe sockets with one that handles mSATA. Also in the mix are eight additional GPIOs. The system has door-accessible drive bays with support for a pair of removable 2.5-in. solid-state drives.
The 1.3-GHz Jetson AGX Orin 64-GB module gets its AI acceleration from 2,048 NVIDIA CUDA cores and 64 Tensor Cores. The combo delivers up to 275 TOPS and 5.3 FP32 TFLOPS of compute performance. Two NVIDIA Deep Learning Accelerator (NVDLA) engines plus a 7-way VLIW vision processor are incorporated as well. Furthermore, there are a dozen Arm Cortex-A78AE cores that can run Linux.
The platform meets other rugged requirements such as MIL-STD-810G, MIL-HDBK-704F, MIL-STD-1275D, and more.
Check out our AUVSI Xponential 2023 coverage. This article is also part of the TechXchange: AI on the Edge.
