Extraction and Recognition Methods of Vehicle Driving Tendency Feature Based on Driver-vehicle-environment Dynamic Data under Car Following

Abstract

Driving tendency is the comprehensive embodiment of driver's physiological and psychological characteristics. It reflects driver's psychological condition during the process of driving. Identification of the driving tendency has great significance to researching of the driver assistant system, especially the active security warning system. In this paper, data from simulative driving and real driving experiment of different drivers (risk-taking, cautious and conservative type, etc) under the state of car-following flow are obtained through questionnaires on psychological test. The feature eigenvectors with good ability to classify the type of driving tendency is extracted based on rough set theory. Dynamic recognition method of driving tendency is put forward with model calibration and verification making use of real data. The results of verification are compared with the results of the questionnaires. It is shown that the pattern recognition model is feasible, and can realize real-time recognition of the driver tendency under the state of car-following.

Keywords:

• driver assistance
• driving tendency
• pattern recognition
• rough set theory
• feature extraction