Physiological Assessment of Driving Mental Fatigue Using Wavelet Packet Energy and Random Forests

This study measured multiple physiological parameters of mental fatigue induced by the simulated driving task. The subjects’ mannerisms and driving performance reveal that the driving mental fatigue is induced after long-term simulated driving. The wavelet packet energy (WPE) of EEG were used to extract the features of brain activity in the normal and the mental fatigue states, then random forests (RF) was employed to select key features for differentiating the two states. The experimental results show that the WPE of EEG and the amplitude of P300 are strongly related with driving mental fatigue. The power spectral indices of heart rate variability (HRV) indicate that the sympathetic nervous system may play a leading role and the parasympathetic nervous system is retrained at the end of driving task. Furthermore, RF method can select features effectively, reduce the dimensionality of the feature vectors significantly, and then achieve a recognition accuracy of 97%. The WPE-RF method could be a potential way of assessment of driving mental fatigue.