Detection of Eye Blinking in Automotive Environments using a FMCW Radar

Accurate eye-blink detection serves as a valuable biometric input for various applications, including driver-drowsiness monitoring, assistive computer interfaces for individuals with disabilities, and deception detection. Traditional methods, such as camera vision and electrooculography, have been widely explored but come with noteworthy limitations. This study investigates a novel, non-contact approach using FMCW radar to detect eye blinking in realistic environments. Experiments were conducted in an indoor setting and a stationary vehicle. Six subjects were tested at elevation angles of observations ranging from 0 to 90 degrees. Radar measurements successfully detected eye blinks at all angles and were validated against camera vision. Under optimal conditions, the method achieved 95% blink detection accuracy, with a 50% variance in velocity measurements compared to validation techniques.