Quantifying driver stress: developing a system for collecting and processing bio-metric signals in natural situations

A system for quantifying the physiological features of emotional stress is being developed for use during a driving task. Two prototypes, using sensors that measure the driver's skin conductance, respiration, muscle activity, and heart activity are presented. The first system allows sampling rates of 200 Hz on two fast channels and 20 Hz on six additional channels. It uses a wearable computer to do real-time processing on the signals and has an attached digital camera which was used to capture images of the driver's facial expression once every minute. The second system uses a car-based computer that allows a sampling rate of 1984 samples per second on eight channels. This system uses multiple video cameras to continuously capture the driver's facial expression and road conditions. The data is then synchronized with the physiological signals using a video quad-splitter. The methods for extracting physiological features in the driving environment are discussed, including measurement of the skin conductance orienting response, muscle activity, pulse, and respiration patterns. Preliminary studies show how using multiple modalities of sensors can help discriminate reactions to driving events and how individual's response to similar driving conditions can vary from day to day.