AI

Annapolis Micro Systems COTS Boards and Systems with integrated FPGAs and GPUs are ideally suited for high-performance Artificial Intelligence applications due to their balance of performance and SWaP efficiency.

FPGAs offer low-latency, reconfigurable hardware acceleration, making them ideal for real-time AI inference tasks with strict timing constraints. GPUs provide massive parallelism, efficiently handling deep learning workloads and complex AI models. By leveraging both, Annapolis COTS products enable high-performing and scalable AI solutions at the edge.

Annapolis supports deep learning acceleration frameworks for a variety of AI applications:

  • ELINT/SIGINT applications focused on deep learning powered spectral sensing and signal identification
  • AI-powered Pulse Descriptor Word (PDW) generation over MORA/VITA 49.2
  • EW tools for Cognitive Radar and Smart Jammers/ECCM
  • Communication applications spectral agility and Cognitive Radio
  • Metacognition and game theory optimization over contested spectral resources
  • AI-powered filtering and channel optimization

These toolsets for AI accelerated inference are supported:

  • Agilex OpenVINO toolset
  • AMD Xilinx Vitus AI toolset and the Deep-Learning Processing Unit (DPU) and acceleration on dedicated Versal AI Engines
  • Vitus Model Composer for AI Engine Kernel development in MATLAB/Simulink
  • Nvidia Jetson AGX Orin module for CUDA supported AI workflows
  • Open CoreFire blocks for transferring data from compute kernels into traditional DSP flowgraphs and programmable logic

 

What Can AI-Optimized Embedded Computing Do for You?

If you require high-performance data digitization, processing or storage in a compact, rugged environment, Annapolis Baseboards, Switches, Storage and Chassis/Backplanes are for you. Contact us to have an FAE recommend the best approach for your AI application.