The goal of this Safety IDEA project was to develop a non-contact sensing platform to monitor the physiological signals of drivers such as the electrocardiography (ECG) and/or electroencephalography (EEG), from which the on-set and extent of drowsiness can be detected. Clinical research has found physiological signals are good indicators of drowsiness. Conventional bioelectrical signal measurement system requires the electrodes to be in contact with the human body. That not only interferes with the normal driver operation, but also is not feasible for long term monitoring purposes. Therefore, a non-contact physiological signal sensing platform as developed in this project will be very helpful to detect driver drowsiness and reduce crashes. Such sensors can be integrated readily into a wireless health monitoring system for drivers. In this project, we designed a non-contact ECG sensor based on high input impedance circuitry. With delicate sensor electronics design, the bioelectrical signals associated with electrocardiography (ECG), breathing, and eye blinking can be measured. This sensor package can detect the ECG signals with an effective distance of up to 30 cm (11.81 inch) away from the body. It also provides sensitive measurement of physiological signals such as heart rate, breathing, eye blinking etc. The sensor performance was validated on a high fidelity driving simulator. Digital signal processing algorithms were developed to remove the signal noise and simultaneously automate signal analyses. The characteristics of physiological signals indicative of driver fatigue, i.e., the heart rate (HR), heart rate variability (HRV), breath frequency and eye blinking frequency, can be determined. A drowsiness indicator was developed by coupling the multiple physiological parameters to achieve high reliability in drowsiness detection. Evaluation of sensor performance was conducted under various conditions in this project. These include evaluation under ordinary laboratory and office environmental conditions. Sensor performance was also evaluated in a high fidelity driving simulator as well as an operational truck. The sensor would have applications for railroad train operators and truck drivers. Results of the evaluation indicate that the sensor is accurate, robust, and easily deployed. All of these evaluations point to great promise for this technology. This project showed that the proposed sensing concept is feasible. Recommendations are made for further development of the sensor prototype. (Author/publisher)
Abstract