Citation

Raksincharoensak P, Khaisongkram W, Nagai M, Shimosaka M, Mori T, Sato T. Veh. Syst. Dyn. 2010; 48(Suppl 1): 55-71.

Copyright

(Copyright © 2010, Informa - Taylor and Francis Group)

DOI

10.1080/00423111003668229

PMID

unavailable

Abstract

This paper describes the modelling of naturalistic driving behaviour in real-world traffic scenarios, based on driving data collected via an experimental automobile equipped with a continuous sensing drive recorder. This paper focuses on the longitudinal driving situations which are classified into five categories - car following, braking, free following, decelerating and stopping - and are referred to as driving states. Here, the model is assumed to be represented by a state flow diagram. Statistical machine learning of driver-vehicle-environment system model based on driving database is conducted by a discriminative modelling approach called boosting sequential labelling method.