Although significant advances have been done with respect to vehicle technology and roadway
construction, driver behaviour remains the number one contributing factor of traffic crashes worldwide.
Studies show that one of the major causes of crashes is driver inattention. Driver inattention may occur
when drivers are involved with secondary activities (e.g., texting, talking on the phone, or eating), and
when they fail to follow the cues of the surrounding environment while driving. The latter is particularly
important when drivers are negotiating maneuvres and are required to interact with other vehicles as in
the case of changing lanes or merging onto the freeway. The main objective of this research is to
investigate the relationship between driver behavior and safety, by looking at the actual body movements
and posture, as well as the eye fixation of the drivers when they are performing lane changing and
merging maneuvers under different traffic conditions. To accomplish this objective, a total of 35 drivers
were recruited to participate in an instrumented vehicle field study, where each participant drove for
approximately two hours along a pre-selected route. Participants’ 3D body posture was recorded with the
use of a low-cost infrared depth sensor (Microsoft Kinect). In addition, participants’ eye gaze throughout
the entire data collection effort was recorded with the help of eye-tracking equipment. Lastly, the vehicle
was equipped with two cameras that faced the front and the rear which allowed for information about the
traffic conditions during the data collection period to be obtained. A rich dataset of driver behavior was
developed and analyzed as part of this research. The analysis findings relate the 3D sequence of driver
motion and posture with the actual eye and head movement of drivers. Based on the analysis, head
movements were the predominant type of movement when driving. The average duration of head
movements was 4 seconds and 3.75 seconds for freeway merging and lane changing maneuvers
respectively, and 2.3 seconds for arterial lane changing. Analysis of the magnitude of movements while
driving showed that the right arm was more active than the left arm for the majority of the drivers, and
differences between different driver groups were observed. However, given the small sample available in
this study, the findings should be treated with caution. This report presents the research approach,
summarizes findings, and provides recommendations accordingly. The research approach is useful for
establishing guidelines for monitoring driver behavior as part of advanced driver assistance systems. The
proposed framework has also potential in developing appropriate alert mechanisms for increasing driver
alertness by monitoring driver body posture.
Authors
Angelos Barmpoutis, Alexandra Kondyli, Virginia P. Sisiopiku
Related Organizations
- Published in
- United States of America