The research of driver's safety and driving assistance systems was focused on the study of the interaction between driver, vehicle, and environment in the last years. Based on a general algorithmic model of driving in combination with drivers interaction developed in previous contributions, the personalization of the model becomes the focal point of the actual work. In this paper, an approach to personalize the human driver model is developed. The analysis of Multivariate Normal Distribution (MND), which depends on the selected driving signals to calculate the probability of individual driving behaviors, is used to build individualized models. By implementing a task specific human driver model, a closed-loop algorithm has been developed according to the tasks of driving into/off the highway, overtaking, lane changing, etc, in which the individual elements are playing important roles in deciding the following actions of the driver. The results of the proposed personalized driver model approach allow the cognitive supervision and also autonomous driving.
展开▼
