Distinguish method of fatigue state based on driving behavior wavelet analysis

Being a direct reflection of the drivers´ states, the driving behavior research is getting widely attention recently. This paper presents a new identification method of fatigue driving state, which is obtained from driving behavior data analysis both about normal driving state and the fatigues ones. PERCLOS80 is utilized as reference to distinguish two different states. During identification process, the driving behavior data is dealt with wavelet transform. Then modulus maxima values and Lipschitz exponents which reflected smooth level of data signal are performed as index to identify driving states: normal or fatigue. Among various experimental driving behavior data, the error to driving center line is chosen as information source here, and the result shows remarkable identified effect.