In the past decade, the development of augmented and virtual reality systems has accelerated. Nonetheless, there are still challenges that need to be addressed, particularly when implementing such technologies on more portable devices such as contact lenses. A key challenge with such implementations is eye-tracking and angle detection. In this work, we demonstrate a passive, battery-free (lens-side) approach that estimates gaze angle by measuring inductance changes from four detector coils embedded in a polydimethylsiloxane (PDMS) eyeglass lens in response to a resonator embedded in a PDMS contact lens. We tuned the detector tanks and the contact lens resonator to nearly the same frequency, which increases reflected impedance and frequency sensitivity, extending the usable range at millisecond-scale acquisition. Three runs were measured with a four-channel inductance-to-digital converter (LDC), scanned sequentially, from which one run was utilized for training a Random Forest model and two independent runs for testing. The LDC was configured for 1.64 ms sensor conversion time, but due to sequential scanning and host logging, the effective sampling interval was 18 ms (∼55.6 Hz) in the data sets. After preprocessing the mean absolute error ± standard deviation (MAE ± SD) was 4.27° ± 4.67° (horizontal) and 6.44° ± 5.13° (vertical) corresponding to 8.61° ± 5.80° Euclidean. At higher sensor conversion time (4.782 ms), the Euclidean MAE decreased to 7.53° ± 5.60° due to noise reduction, illustrating the latency–accuracy trade-off. The results support passive inductive sensing as a viable route to camera-free, battery-free coarse gaze-region estimation in low-power wearables, while higher-precision gaze interaction remains for future work.
Azhari et al. (Thu,) studied this question.