Online Detection of Driver Fatigue Using Steering Wheel Angles for Real Driving Conditions

This paper presents a drowsiness on-line detection system for monitoring driver fatigue level under real driving conditions, based on the data of steering wheel angles (SWA) collected from sensors mounted on the steering lever. The proposed system firstly extracts approximate entropy (ApEn) features from fixed sliding windows on real-time steering wheel angles time series. After that, this system linearizes the ApEn features series through an adaptive piecewise linear fitting using a given deviation. Then, the detection system calculates the warping distance between the linear features series of the sample data. Finally, this system uses the warping distance to determine the drowsiness state of the driver according to a designed binary decision classifier. The experimental data were collected from 14.68 h driving under real road conditions, including two fatigue levels: wake and drowsy. The results show that the proposed system is capable of working online with an average 78.01% accuracy, 29.35% false detections of the awake state, and 15.15% false detections of the drowsy state. The results also confirm that the proposed method based on SWA signal is valuable for applications in preventing traffic accidents caused by driver fatigue.