AI-enhanced precision alignment of panda polarization-maintaining fibers for next-generation photonic applications

Abstract

This study introduces an artificial intelligence (AI)-based approach for high-precision alignment of Panda polarization-maintaining optical fibers. Using the YOLOv8 model for object detection, our method effectively aligns the slow axis of the Panda fiber with the edge of a pre-designed groove, which is essential for preserving polarization properties in optical communication and sensing applications. A 1000× microscope camera captures images of the fiber and groove, allowing the AI model to accurately detect the angle between the fiber’s slow axis and the groove edge. This angle information is then used to control a motor that rotates the fiber until alignment is achieved. Extensive experiments reveal that our system achieves an angular alignment error of < 2°, limited mainly by image quality and groove irregularities. This automated alignment system, driven by a deep learning model, offers significant improvements over traditional methods, optimizing alignment accuracy and operational efficiency and presenting new possibilities for the integration of AI in photonic device fabrication.