Design and Implementation of Driver State Detection System using Physiological and Physical Sensors

Abstract

Time-frequency technique of synchrosqueezing based continuous wavelet transform (CWT SST) is proposed to estimate the Heart Rate Variability(HRV) of given ECG data. From the CWT-SST, 10-20 Hz energy band of wavelet coefficients information is extracted to detect R-peaks. Furthermore, the reconstructed signal from the 10-20 Hz energy band and local maxima are useful to determine the R-peak locations. HRV is a dominant pa rameter in detecting stress state of a driver or any individual person with abnormalities. In addition to this, physical sensor can also be helpful in detection of some of driver state conditions such as drowsy and distraction. Accelerometer sensor based head movement analysis for jerk detection to determine state of the driver. CWT-SST technique is ap plied on the collected accelerometer data along with x-axis and 0-1.5 Hz band energy of wavelet coefficients information is used to determine a jerk locations. Flexible inter dig itated capacitive sensor is designed for hand touch and position sensing on the steering wheel rim, which helps in understanding the pattern of driver behavior while driving on the road. The physiological and physical sensor data is collected from different subjects and tested with the proposed algorithms, it shows good capability to estimate the driver states and behavior.