On Task Labeling Method of Driving Behavioral Data Based on Formal Grammar for Driver Assistance System

Abstract

Analysis of driving behavioral data is recently developing and recognized as a necessary technique for driver assistance system. A driver model based on the analysis is used for understanding and predicting the behavior of drivers. In the present study, we propose dividing the driver model into two parts: a task model and a user model. The task model objectively expresses how driving tasks occur according to the geographic infromation and the traffic environment while the user model shows how the driver perceives the environment and makes decisions. We also show formulation and labeling of driving tasks based on formal grammars in the task model.