Smart Machine Vision for Universal Spatial-Mode Reconstruction

Area of Science:

• Optoelectronics
• Optical Communications
• Artificial Intelligence

Background:

• Structured light beams, particularly those with orbital angular momentum (OAM), offer enhanced transmission capabilities for communication systems.
• OAM-based communication faces challenges from mode distortion in disordered media and high-order mode divergence.
• Current AI solutions for OAM distortion are computationally intensive, requiring high processing time and power.

Purpose of the Study:

• To develop a low-power, low-cost solution for detecting and reconstructing distorted OAM-carrying beams.
• To demonstrate an image sensor acting as an artificial neural network for OAM beam reconstruction.
• To overcome the limitations of existing AI algorithms in terms of processing time and power consumption.

Main Methods:

• Utilizing a low-power, low-cost image sensor engineered to function as an artificial neural network.
• Implementing the sensor for simultaneous detection and reconstruction of distorted OAM beams.
• Testing the system's efficacy with individual Vortex, Laguerre-Gaussian (LG), and Bessel modes, as well as hybrid superpositions.

Main Results:

• The image sensor successfully reconstructed distorted OAM beams with 95% efficiency.
• Demonstrated reconstruction of individual and hybrid (nonorthogonal) OAM modes.
• Validated the sensor's capability as a low-power artificial neural network for optical signal processing.

Conclusions:

• A novel, low-power image sensor can emulate an artificial neural network for OAM beam reconstruction.
• This approach significantly reduces processing time and power consumption compared to conventional AI methods.
• The developed device offers a promising foundation for future low-power, high-efficiency optical communication technologies.