Eyeball Kinematics Informed Slippage Robust Gaze Tracking

Gaze movement is a crucial index of human attention and thus shows great potential in human-computer interaction. Head-mounted devices (HMDs) are developing rapidly and show a great demand for head-mounted gaze-tracking techniques. However, the lack of slippage robustness and excessive calibration time still bother current gaze-tracking systems. This article proposes STARE, a head-mounted real-time gaze tracking system with slippage-robust gaze estimation and minimal calibration. STARE leverages the eyeball kinematics, specifically Listing’s law and Donder’s law, to propose a mapping function for slippage robust gaze estimation that holds physical significance. Our succinct mapping function minimizes personal calibration time to its lowest. The experimental results of 40 subjects demonstrate that our system achieves a mean angular error of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${0}.{71}^{\circ } $ </tex-math></inline-formula> under varying levels of device slippage and decreases the personal calibration time to less than 1 s. STARE outperforms state-of-the-art methods in gaze tracking accuracy and precision. Our system is convenient for practical usage and shows excellent potential for gaze tracking.