Design & analysis of performance of K-Means algorithm for cognitive fatigue detection in vehicular drivers using Skin Conductance signal

The fatigue level of the driver changes due to many factors such as monotonous job of continuous attentive driving, road traffic, un-healthy road conditions, insufficient sleep, stress level, changing work habits and adverse environmental conditions. This paper provides an alternative approach for design of fatigue classifier for vehicular drivers using Skin Conductance (SC) signal to save the loss of lives by reducing the number of occurrences of vehicular accidents specifically those caused by cognitive fatigue of the drivers. We have implemented basic K-means algorithm and proposed two modifications as classifier for detection of two state fatigue data of Skin Conductance signal recorded from vehicular drivers. The classifiers were trained and tested for wavelet and statistical features of SC signal. The features so extracted were treated as individual decision making parameters. From test results it could be found that some of the statistical features and wavelet features could fetch 100 % classification accuracy.