A Radar-Based In-Cabin Health Monitoring System

The topic of in-cabin health care monitoring within vehicles has recently garnered significant attention. This technology serves two primary applications in a vehicle. First is the monitoring of the vital signs of drivers and passengers. Given the significant amount of time individuals spend driving daily, it is essential to monitor their vital signs to identify potential health issues at an early stage. If there is a health condition, autopilot mode of vehicle can be used. Second, it facilitates occupancy detection, which is crucial in detecting instances of a child being left behind in a vehicle. As a result, the need for in-cabin health care monitoring is rapidly increasing.
\n Radar technology is particularly popular for use in health monitoring systems due to a number of reasons, one of which is the privacy concern. While vision and thermal cameras can also be used for health monitoring, they may be perceived as invasive to an individual's privacy. Additionally, radar-based health monitoring systems are contactless, making them more suitable for in-vehicle applications where maintaining a certain distance from the subject is necessary. In addition, radar technology is often more cost-effective than other types of sensors.
\n In this thesis, frequency-modulated continuous wave (FMCW) radar systems are employed for in-cabin health monitoring. A dual radar system has been developed to monitor breathing patterns during driving, with a specific focus on detecting potential breathing issues. Because abdominal breathing may result in reduced chest displacement, it's essential to monitor both the chest and abdomen for early detection of any breathing abnormalities. In this system, separate radars are employed to monitor the movements of the chest and abdomen simultaneously. Various breathing abnormalities, including Tachypnea, Bradypnea, Biot, Cheyne–stokes, and Apnea, are explored. The proposed algorithm can detect the mentioned breathing abnormalities through breathing rate (BR) estimation and breath-hold period detection. In addition, the proposed method in this thesis estimates BR based on the multiple range bins. The experimental results demonstrate a maximum BR error of 1.9 breaths per minute using the proposed multi-bin technique. In addition, the dual radar fusion system can detect breath-hold periods with minimal false detections.
\nSecondly, multi-input-multi-output (MIMO) FMCW radars have been developed to monitor multiple people inside the vehicle in two different applications, including vital sign monitoring and occupancy detection. For vital sign monitoring, digital beamforming algorithms are explored to monitor various angles inside the vehicle. Different scenarios involving either a single subject or multiple subjects were deployed. The results indicate that the proposed system can monitor the breathing patterns of multiple subjects simultaneously when they are seated in the same row. However, when they are seated in different rows, the reflected signals from subjects in the second row are combined with the subjects in the first row due to the multipath inside the vehicle. For occupancy detection, a novel approach that involves detecting the occupied space in each seat is presented in this thesis. The variance of detected points is suggested as an indicator of volume occupancy. In the conducted experimental study, which covers 70 different scenarios involving both single-subject and multi-subject situations, each seat is categorized into one of three labels: adult, baby, or an empty seat. The proposed approach achieves an overall accuracy of 96.7% using an AdaBoost classifier. Additionally, a miss-detection rate of 1.3% is achieved when detecting babies. The proposed approach demonstrates better robustness to multipath compared to the more commonly used energy-based approaches.
\nThirdly, a radar system operating at 60 GHz and using FMCW technology is positioned behind a seat to monitor an individual's heart waveforms. The suggested algorithm accurately recognizes specific patterns in healthy subjects' heart waveforms, depicting two peaks followed by a valley in each cycle. High-frequency components related to breathing, often present in the heart band, are eliminated through variational mode decomposition (VMD) to refine the reconstructed heart waveform. The proposed method effectively detects and compensates body movements in seated individuals in the time domain, utilizing multiple range bins to identify and remove signals affected by strong body movements. A comprehensive investigation into heart rate variability (HRV) and heart rate (HR) estimation yields a median interbeat interval (IBI) estimation error of 30 ms and an average relative error of 4.8% for HR estimation using the VMD and multi-bin approach. Furthermore, the study focuses on analyzing a group of older adults to detect heart conditions, with those exhibiting a prolonged corrected QT interval (QTc) showing distinct heart waveforms compared to those without this condition. This specific heart waveform can serve as an indicator for detecting the mentioned heart condition. Additionally, the research delves into human body vibrations within vehicles, particularly in the presence of car body vibrations induced by road defects like cracks and potholes. A threshold based on z-axis acceleration is set to detect these road defects; exceeding 12 m/s² leads to the omission of the corresponding signal, followed by employing an autoregressive integrated moving average (ARIMA) model with forward forecasting to reconstruct the omitted sections. The experiments reveal a median IBI estimation error of 37 ms and an average relative error of 5.9% for HR estimation.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.