Abstract
This paper addresses a method of unusual driving behavior detection based on naturalistic driving database in urban area. This contributes to the development of an active safety system such as advisory and warning devices to prevent accidents in early stages. Naturalistic driving behavior data in real-world traffic situation is the key point of good cooperative characteristics between driver and driver assistance system. On-road experiments are conducted to capture the data by using continuous sensing drive recorder. The acceleration during pedal operation and time headway during following situation are analyzed to understand the differences of these feature values in running speed region, road environment and driver. In addition, the feature values in hurry driving classified by the distinctive factor are analyzed compared to usual driving data.
