Beaconless adaptive optics for atmospheric laser propagation with multi-plane convolutional neural network

Area of Science:

• Optics
• Machine Learning
• Laser Physics

Background:

• Adaptive optics (AO) systems correct wavefront distortions caused by atmospheric turbulence.
• Traditional AO requires a beacon laser for wavefront sensing, which is not always available.
• Laser propagation through scattering media presents challenges for AO correction.

Purpose of the Study:

• To develop a machine learning (ML) based method for emulating adaptive optics (AO) in laser propagation.
• To enable AO correction in scenarios lacking a dedicated beacon laser.
• To improve the figure of merit for laser beams propagating through scattering media.

Main Methods:

• Utilized a convolutional neural network (CNN) to correlate scattered-light intensity profiles with phase profiles.
• Developed a ML method to emulate AO action without a beacon laser.
• Recorded scattered-light intensity profiles in image planes near conjugate to the object plane.

Main Results:

• The ML-based AO emulation achieved an improved figure of merit compared to tip-tilt correction alone.
• Demonstrated the feasibility of beacon-less AO for laser propagation through scattering media.
• Quantified the performance improvement achieved by the developed technique.

Conclusions:

• The developed ML method effectively emulates AO for laser propagation without a beacon.
• This technique provides a viable solution for AO correction in scattering environments lacking beacons.
• The study highlights the potential of ML in advancing AO applications for laser propagation.