A resonant eye-tracking microsystem for velocity estimation of saccades and foveated rendering

We demonstrate the first MEMS-based eye tracking system that operates in resonance to obtain 3300 eye position measurements per second, a 25-fold improvement over previously reported microsystems. The present system achieves at least a 10x improvement in bandwidth, volume (0.5cm
³
), power consumption (15mW), and cost when compared to state-of-the-art camera-based systems, all without compromising resolution (0.4° RMS). A system with these specifications produces real-time velocity measurements within saccades, enabling on-the-fly prediction of fixations for the first time. This predictive capability will significantly reduce the power consumption of graphical processing units (GPUs) in augmented and virtual reality (AR/VR) headsets because the region of a display within the fovea may be rendered exclusively (“foveated rendering”) [1].