Radar-based In-Vehicle Heart Rate Estimation with an AI-based Validity Check

This paper presents a framework for radar-based heart rate estimation of the driver of a moving vehicle. To reduce the influence of driver and vehicle movements on the radar signal, various sensor mounting positions were evaluated. Based on signal quality and estimated heart rate accuracy, mounting the radar in or behind the driver’s seat was identified as optimal. Additionally, an AI-based validity check is proposed to prevent false heart rate estimations under poor signal conditions or large driver movements. The framework was validated against electrocardiogram data, yielding an average absolute error of less than 8.4 bpm respect to the true heart rate.