Localizing Changes in Driver Behavior via Frequency-pattern-analysis

Explorative analysis of driver behavior, a key variable in the context of automotive research and development, can be tedious. The authors present a quick and easy method to identify changes in recorded driver behavior data. The method consists of a data processing algorithm that uses Fourier-series and statistical t-tests to identify points in time where changes in the frequency of the recorded signal occur. An exemplary usecase for the method is presented for driver steering torque data obtained in an experiment with an automatic obstacle-avoidance maneuver. The results allow for the assumption that changes in frequency of driver steering torque may mark meaningful, implicit changes in driver behavior even when driver behavior does not explicitly change, thereby making obvious the potential of the proposed analysis method.