Precise VR Eye Tracking using Single-Shot Deflectometry
Department of Electrical and Computer Engineering, Northwestern University; 2Department of Computer Science, Northwestern University; 3Wyant College of Optical Sciences, University of Arizona
Abstract
Although the task has been studied for several decades, a robust solution to precise and fast eye tracking remains an unsolved problem. With the advent of Virtual, Augmented, or Mixed Reality (VR/AR/MR), precise eye tracking recently attracted considerable research interest - mainly because it enables many procedures that significantly improve the performance and experience in VR/AR/MR headsets, such as foveated rendering, or compensating for the accommodation-convergence reflex. In this contribution, we present a novel method for precise and fast eye tracking in VR/AR/MR headsets, which utilizes recent advances in single-shot Phase-Measuring-Deflectometry. In contrast to current state-of-art methods that only sparsely sample the eye surface, our approach allows for a dense capture of 3D surface information. This information is used, eventually, to extract 3D surface features for a precise estimation of the gazing direction. We will explain the working principle of our method, introduce our prototype setup, and show real experiments with an evaluated gazing vector precision below 0.5°.