A wearable non-contact optical system based on muscle tracking for ultra-long-term and indirect eyetracking

Abstract

Eye signals have become popular worldwide due to their potential in multiple fields. Such data can be regarded as indications of diseases such as epilepsy, a sign of fatigue, expression of potential interests during shopping and so on. In this paper, we demonstrate a novel optical eye-tracking system, NIRSense, which is able to track eye activities based on muscle movements, without any touching with the cornea. The NIRSense sensors, six pairs of near-infrared (NIR) light-emitting diodes (LEDs) and photodiodes are emitting 940 nm lights on skins below eyes in a contactless distance, where lights are able to penetrate and be reflected back to photodiodes. These signals are then transmitted into electric currents and processed to build relationships between signals and pupil locations. It is proved that the NIRSense system can track eye activities not only accurately (1.05 degrees) and precisely (0.6 degrees), but also in a safe, long-term used way. Moreover, the hardware design of the NIRSense system allows the assembling easily on diverse near-eye frames, involving Virtual reality (VR) goggles, Augmented reality (AR) headsets and normal glasses frames, via attaching sensors near the bottoms of lenses. In this way, this system has enriched the possibilities of exploring human reactions to virtual or real environments. Functions of accurate and precise eye trace prediction, pupil tracking with eyelids closure, comparison of performances with a commercial eye tracker Tobii Pro Glasses 3 and clinical application on the concussion assessment are presented in this paper.