Hybrid Driver Fatigue Detection System Based on Data Fusion with Wearable Sensor Devices

The increasing popularity of wearable devices has raised the importance of human-centric services, such as healthcare, physical training, smart assistant, and its applications. In this paper, we focus on a hybrid drowsiness detection system to use wearable sensor devices for measuring drowsiness. We make following contributions: (1) an auto-configurable and adaptive middleware framework to manage flexible devices, (2) a hybrid drowsiness detection module using the wearable sensor devices and ordinary camera, (3) a real-time driver drowsiness detection system to evaluate classifying result to provide more accurate drowsiness status classification, (4) we experimentally compare the results of the above method with video-based method with and without wearable sensor device, then show that our method is crucial to achieve better prediction accuracy and robustness. Overall, the experimental result shows that the proposed system allows at the most 13 % of accuracy improvement. Therefore, the proposed system can be used to improve the accuracy of driver drowsiness detection based on fusion wearable device measures in real-time. Keywords— PERCLOS; Wearable MEMS sensors; Sensor Fusion; Driver fatigue detection; Drowsiness detection;