TY - JOUR PY - 2010// TI - Development of a new driver behavior model through a novel object detection technique JO - Proceedings of the Road Safety on Four Continents Conference A1 - Sarvi, M. A1 - Zavabeti, A. SP - 494 EP - 506 VL - 15 IS - N2 - In this paper a comprehensive framework is proposed to model driver behavior considering the surrounding vehicles. With the proposed structure, the driver consideration and action spaces can be measured and related. Furthermore, a platform is developed and presented which could facilitate a dynamic real traffic data collection to study driver behavior. The application of the developed platform to collect associated traffic data in a follow the leader experiment is presented in which the real microscopic traffic data (e.g. vehicles speed, acceleration, relative distance and speeds to surrounding vehicles) is measured through a unique moving object detection technique. Different schemes of filtering and smoothing are applied to the collected data. A comparison among various car-following models is used to analyze the predicting ability of the models. Two linear models as well as the GM model are used to observe the car-following behavior. The developed platform provides promising capabilities in capturing driving behavior in complex driving conditions using its advanced features and unique moving object detection software. The potential applications of the proposed driving framework and developed platform in the context of collision prevention and driving assistance systems are also discussed.

LA - SN - UR - http://dx.doi.org/ ID - ref1 ER -