Radar-Based Eyeblink Detection Under Various Conditions

Abstract

This paper aims to assess the ability of Frequency Modulated Continuous Wave (FMCW) radar to detect blinks not only in direct facing and limited-range situations but also by exploring different factors that may affect the accuracy of the detection process such as distance, angle, and movement. To achieve this, we propose a three-layered processing chain that relies on the use of Adaptive Variational Mode Decomposition (AVMD) algorithm to extract the blink signal as it can correctly separate it from complex radar reflections. We use an off-the-shelf FMCW radar operating at 77 GHz from Texas Instruments (TI) to perform several experiments and evaluate the feasibility of blink detection in each scenario, including distances up to 1.2 meters, different angles, while chewing gum, and across different subjects. The evaluation results show that FMCW radar combined with our processing chain can detect eyeblinks correctly under different conditions and farther distances than previous works.