Efficient prediction for arbitrary trajectories of butterfly vortex beams using a deep neural network

Area of Science:

• Optics and Photonics
• Biomedical Engineering
• Artificial Intelligence

Background:

• Longitudinal particle manipulation requires precise control over particle motion via beam propagation.
• Existing methods struggle with multi-directional manipulation and accurate prediction of light spot deviation over distance.
• Single trajectories are insufficient for complex particle guidance needs.

Purpose of the Study:

• To introduce butterfly vortex beams (BVBs) capable of arbitrary trajectory design for longitudinal particle manipulation.
• To develop a deep learning model for accurate prediction of light spot positions in BVBs.
• To demonstrate the efficacy of BVBs in guiding particles and advancing deep learning applications in manipulation and biomedicine.

Main Methods:

• Proposal and experimental demonstration of butterfly vortex beams (BVBs) with adjustable propagation trajectories.
• Development of a deep learning-based model to predict the spatial position of the light spot.
• Validation of particle guidance capabilities using the proposed BVBs.

Main Results:

• BVBs successfully demonstrated adjustable propagation trajectories for particle guidance.
• The deep learning model achieved high accuracy in predicting light spot positions, with a mean square error near 0.01.
• Experimental validation confirmed the effectiveness of BVBs in controlling particle motion.

Conclusions:

• Butterfly vortex beams offer a flexible solution for longitudinal particle manipulation with arbitrary trajectories.
• Deep learning significantly enhances the predictive accuracy for light spot positioning, crucial for precise control.
• This integrated approach holds promise for advancements in particle manipulation techniques and biomedical applications.