Federal And Aerospace AI Suite

The suite is a comprehensive set of resources designed to accelerate the development and deployment of edge AI solutions across federal and aerospace domains, as well as demonstrate Intel’s integrated AI acceleration and real-time performance on Intel® Core™ Ultra Series 3 and Intel® Core™ Series 2 processors in defense-focused use cases.

It aims at providing a foundation for evaluating, benchmarking, and validating from handheld systems to avionics, drones, urban air mobility, space, and ground station infrastructure. It currently offers:

• Handheld Multi-Modal application

• UAV Vision Analytics

The applications are optimized for AI inference on portable devices, focusing on SWaP-C compliance (Size, Weight, Power, and Cost). They also help meet certification, ruggedness, and interoperability requirements that are important for field hardware use.

How to use the Federal And Aerospace Suite

1. Install the Edge Node Infrastructure software

The most efficient way to deploy the Handheld Multi-Modal app is to use the preconfigured Edge Node Infrastructure software system. To learn how to do it, refer to:

• Build the standard handheld OS image

Optionally, to learn more about customizing and optimizing Operating System images, refer to:

• Image Composer Tool documentation - Infra developer composes a slim, Intel-tuned Ubuntu based image for the target use case.

2. Deploy the application

Once your system is installed, you can deploy selected apps:

Handheld Multi-Modal app:

• Download the package

• Deploy the Handheld Multi-Modal application - User runs the multi-modal handheld app (speech + chat) on Intel XPUs.

UAV Vision Analytics:

• Download the package

• Deploy the UAV Vision Analytics - standalone

• Deploy the UAV Vision Analytics - SDK mode

3. Run benchmarks

Use the benchmark guide for your application, or run platform-level benchmarks:

• Benchmark the Handheld Multi-Modal application

  • Use ViPPET to compare pipeline performance across CPU, GPU, and NPU.

• Benchmark the UAV Vision Analytics application

  • Measure UAV video stream density and hardware utilization. ViPPET is also available as an optional tool.

• Edge Workloads and Benchmarks: Handheld Multi-Modal and UAV Vision Analytics

  • Run platform-level vision, media, edge-AI-pipeline, and GenAI benchmarks.

4. Use AI agent to develop your solution

Use the AI agent guide for your application:

• Handheld Multi-Modal AI agent guide

  • Build and provision the platform, tune power and thermal settings, and generate DL Streamer pipelines.

• UAV Vision Analytics AI agent guide

  • Includes the shared infrastructure and DL Streamer workflows plus UAV SDK commands and MCP tools.