抄録
In this research an online driver steering control modeling and driver state assessment system is investigated. Using ARMAX time series model with online system identification, useful dynamic characteristics of the driver can be derived. Using the dynamic characteristics of the driver, two different driver state assessment algorithms are investigated. Compared to the fuzzy logic classifier, which is developed by manually tuned parametric functions, the probabilistic neural network classifier utilizes more systematic methods in choosing the input variables and performing the classifications. The classification results indicate the advantage of using the online driver model in assessing the driver behavior.
