Atmospheric Turbulence Aberration Correction Based on Deep Learning Wavefront Sensing.

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Atmospheric Science

Background:

• Wavefront sensing is crucial for adaptive optics systems to correct optical aberrations.
• Traditional methods like Shack-Hartmann Wavefront Sensing (SHWS) have limitations in speed and complexity.
• Deep learning offers a promising alternative for real-time wavefront measurement and correction.

Purpose of the Study:

• To develop and validate a novel Deep Learning Wavefront Sensing (DLWS) system.
• To evaluate the performance of the DLWS model against established methods like SHWS.
• To demonstrate the effectiveness of DLWS for long-range laser aberration correction.

Main Methods:

• Development of a DLWS neural network integrating attention mechanisms and Convolutional Neural Networks (CNNs).
• Training and validation using simulated atmospheric turbulence datasets and an in-house experimental platform.
• Deployment and testing of the trained DLWS model in kilometer-range laser transmission experiments.

Main Results:

• The DLWS model achieved accuracy comparable to the Shack-Hartmann Wavefront Sensing (SHWS) method in indoor experiments.
• The DLWS model, trained indoors, was directly applied to kilometer-scale experiments without retraining.
• Closed-loop correction using DLWS resulted in a 5.35-fold increase in the average peak intensity of the laser spot at the target.

Conclusions:

• The proposed DLWS model demonstrates high accuracy and robustness for atmospheric turbulence compensation.
• DLWS offers a practical and efficient solution for real-time wavefront sensing and aberration correction in long-range optical systems.
• The study highlights the potential of deep learning for advancing adaptive optics and laser communication technologies.