Online driver behavior classification using probabilistic ARX models

Paper in proceeding, 2013

This paper shows how a probabilistic ARX model can be used to predict the driver's steering behavior and classify the current driving style, based on measurements from the vehicle sensors. An algorithm, online classifying the driving style and predicting the steering behavior, is designed and validated on data recorded on a test track. The algorithm is designed to distinguish between two driving styles corresponding to normal and aggressive driving.