← OSINT Library

AI/Deep Learning Wavefront Sensing for HEL Using Target Image to Simplify Adaptive Optics Systems

NPS Calhoun · Agrawal, Brij ·

Executive Summary

This project explores using deep learning to simplify adaptive optics (AO) systems for high-energy lasers (HEL) by predicting wavefront errors directly from target images. Researchers at the Naval Postgraduate School developed a CNN model, trained with UAV images, to characterize atmospheric turbulence, aiming to reduce the complexity and cost of traditional AO systems.

Why It Matters

This research could significantly enhance the performance and reduce the complexity of high-energy laser systems by integrating AI for atmospheric turbulence correction, making them more viable for defense applications.

Key Takeaways

  • Deep learning can simplify adaptive optics for high-energy laser (HEL) systems by eliminating the need for traditional wavefront sensors and reference beacons.
    Source evidence · PDF page 3
    The objective of this project is to determine wavefront error from a target image using deep learning algorithms. Traditional correction methods use wavefront sensors with adaptive optics (AO), which can be expensive, complex, and resource intensive.
  • A convolutional neural network (ResNet-18) was successfully adapted and trained using turbulent images of UAVs to predict atmospheric distortions.
    Source evidence · PDF page 3
    Researchers conducted a transfer learning experiment with a convolutional neural network (CNN), ResNet-18, specialized in image classification. It was modified to fit a regression problem of characterizing the Zernike polynomials corresponding to the turbulence. Next, large datasets of turbulent images of the Reaper Unmanned Aerial Vehicle (UAV) were created.
  • The developed model demonstrated strong generalization, accurately predicting turbulence on unseen data and different UAVs, suggesting its potential for real-world deployment in HEL systems.
    Source evidence · PDF page 4
    Deep learning-based wavefront sensing from object scenes in adaptive optics for high-energy laser (HEL) systems is a promising approach, as it eliminates the need for both a reference beam and a wavefront sensor.

Strategic Relevance

Improving the effectiveness and deployability of high-energy laser systems through AI-driven adaptive optics has direct implications for military defense capabilities, particularly in areas like missile defense and counter-UAV operations.

Source Website View PDF

Key Entities

Naval Postgraduate SchoolOffice of Naval ResearchBrij AgrawalLeonardo HerreraNicholas A. MessinaReaper UAVHigh-Energy Laser Systems

Best For

Defense ResearchersAI/ML EngineersLaser Systems DevelopersMilitary StrategistsAcquisition Professionals

Related Themes

Directed Energy WeaponsArtificial Intelligence in DefenseAdaptive OpticsUnmanned Aerial VehiclesAtmospheric Turbulence Compensation

Get the week's most relevant defense documents

Subscribe to our free newsletter to receive curated OSINT reports, analysis, and strategic updates directly in your inbox.

Join the briefing Subscribe