Estimation of Driver's Fatigue Based on Steering Wheel Angle

Driver's fatigue has been verified as a major factor in many traffic accidents. The estimation of driver's vigilance by steering wheel angle is good way because it is a non-invasive method compared with EEG. An adaptive vigilance estimation methodology based on steering wheel angle information is proposed. The sample data classification index is built from EEG and PVT information of ten driver's virtual driving experiment on driving simulator. According to the geometry information of road centerline and the location of the automobile center, a new algorithm is proposed to compute the lane deviation. The correlation coefficient between steering wheel angle and lane deviation are computed, and the results show that their correlation level is 0.05. Based on the steering wheel angle, the driver fatigue evaluation model is established by the Bayesian Network (BN). The structure and parameters for BN model are determined after adaptive training. The experiment results verified that this model is effective to identify driver's fatigue level.