Ambient Light Robust Eye-Tracking for Smart Glasses Using Laser Feedback Interferometry Sensors with Elongated Laser Beams

Smart glasses, including audio, augmented reality, and auto focal glasses, are emerging as a new class of wearables, designed to enhance user interaction with surroundings. Despite their potential, smart glasses suffer from limited natural interaction methods. Gaze tracking offers a promising solution, allowing users to interact with the device and their surroundings using their eyes. However, current video oculography (VOG) systems face challenges such as high-power consumption, bulky sensors, and reduced robustness in varying lighting conditions. To address these issues, we propose Laser Feedback Interferometry (LFI) sensors as an alternative to VOG. LFI sensors offer advantages in size, power efficiency, and ambient light robustness. The proposed LFI eye-tracking sensor integrates seamlessly into lightweight eye wear, requiring no user-specific calibration and enabling continuous, all-day use. To ensure reliable tracking across a wide range of users with varying interpupillary distances, we employ elongated laser beams that cover a broader eye region. This design enhances tracking precision by accurately capturing the user's gaze vector while maintaining robustness under different lighting conditions. We developed a comprehensive simulation model that integrates the sensor and eye model, validated against real-world experiments, and designed a convolutional neural network (CNN) based model to accurately estimate gaze from LFI sensor readings. Our results demonstrate that the LFI-based system provides a robust, efficient, and user-friendly gaze interaction solution, achieving high gaze prediction accuracy with an error margin of 2.23° on a custom simulation dataset. This represents a significant step toward next-generation smart glasses.