Biosignal based on-road stress monitoring for automotive drivers

Design of a bio-signal based wearable driver assist system requires a functional algorithm capable of mapping features extracted from time series data of physiological signals to a known mental / affective state. The paper presents a comprehensive analysis of extraction and signal processing techniques of features from physiological signals. Self Organizing Map based approach was adopted to cluster data into topographically distinct clusters of low, medium and high stress states. Further, Profile Analysis of the observed drivers led to categorization on the basis of the stress susceptibility index. A cumulative sum-based stress metric, capable of detecting over-stress conditions, was developed using Page?s Technique. These techniques will help in identification of stressful situations where the driver is susceptible to temporal loss of concentration and vehicle control. Thereby, appropriate actuation of relaxation procedures in such situations can mitigate mishaps like road accidents.