This study attempts to clarify driving behavior on a large-scale crest-shaped intersection, where traffic accidents have often been observed, and to examine the effects of an in-vehicle warning information (IVWI) system on driving behavior. To examine the effects of providing IVWI, a well-equipped probe vehicle, called the Advanced Cruise-Assist Highway Systems (AHS) test vehicle, is employed to measure driving behavior across space and over time in a more accurate manner, in that driving behavior can be recorded every 0.1 second. In addition, in the same field experiment, the effects of providing static and dynamic information with different human-machine interfaces (HMI) are tested by comparing them with those in the case of no information provision. It is confirmed that the locations of decelerations over 0.3 G are observed near road sections on which more traffic accidents occur, while some of these large decelerations are also found on road sections where no traffic accidents have occurred. Information provision usually results in the clear spatial dispersion of peaks of dangerous driving behavior, measured using autocorrelations, which are observed in the no information case. Dynamic information seems more effective for drivers to avoid dangerous driving than does static information. In the case of static information, providing information with both voice and image is better than providing information with only voice; however, such a trend does not emerge clearly for dynamic information. It is also found that the simplified HMI could have the same effects as those of the complicated HMI in the case of dynamic information provision.
Abstract