Research on driver status monitoring system using ECG sensor based on dry electrode

Abstract

The traffic accident is one of the most serious problems we have at present. Especially, traffic accident caused by incapacitation of drivers (i.e., falling asleep or heart attack) brings fatal consequence, and the cause accounts for 30 to 50% of the total traffic accidents. Therefore, the Driver Status Monitoring (DSM) systems and the real-time signal processing techniques can be an innovative solution to significantly reduce the number of traffic accidents around the world. This dissertation, firstly, introduces a wide range of research trends in the development of DSM systems, and it introduces the latest DSM systems based on physiological signal measurements, which overcomes the disadvantages of the conventional DSM systems and achieves highly accurate and reliable detection of the driver’ status. Secondly, in this dissertation, The dry electrodes that the automotive manufacturers use to measure electrocardiogram (ECG) signals for driver status monitoring have three technical problems: a lack of attachment fexibility, low ECG signal detection stability, and the potential to harm or irritate the driver. In addition, the complicated signalconditioning circuits employed by automotive manufacturers to improve the SNR, increase the ECG signal detection stability, and to ensure a wider dynamic range of the electrodes increase the cost and complexity of driver ECG measuring systems. In this paper, we propose a driver ECG measuring system that resolves these three technical issues using a steering wheel covered with a conductive fabric-based dry electrode material, which is manufactured by an electroplating method. In addition, we employ a conductive fabric-shaping procedure in the development of the fabric-based dry electrode to improve the attachment fexibility and to reduce the cost and complexity of the required signal-conditioning circuit. We verify the ECG signalmeasuring performance of the proposed system by comparing it with ECG signal measurement results from a clinical ECG monitoring system. In addition, despite applying a simpler signal conditional circuit than those used in conventional ECG measuring systems, we verify that the proposed system achieves higher SNR and ECG signal detection stability than conventional ECG measuring systems through various field tests in actual driving environments. We also analysis a signal-conditioning circuit that is utilized to acquire the driver’s ECG signals reliably. In particular, we analyze the envrionmental durability assessment that are necessary for the commercialization of the system. Lastly, we suggest we propose DSM algorihm based on MLP considering various noise that should be generated while driving. The proposed algorithm shows average accuracy of 69.19% and maximum accuracy of 80.71%.