An Approach to Determining Insufficient Safety Confirmation at Stop-Sign Intersections

Many traffic accidents occur at intersections and in their vicinity. The main cause of such accidents is the driver's failure to make safety confirmations, such as driving without keeping their eyes on the road ahead or driving while absorbed in thought. Therefore, the development of driver support systems, such as a system that determines whether safety confirmation is sufficient and alerts the driver if it is not, are useful to prevent intersection accidents. In this paper, a determination method is proposed for the development of such a system. At first, a driving simulator is set up with a course that includes intersections with good visibility and blind intersections surrounded by buildings. The test subjects run the course repeatedly and the vehicle's behavior, the driver's behavior, their gaze angles, etc., are measured. The driving behavior in these situations is analyzed and the characteristics extracted and accumulated. Secondly, a determination method is constructed based on the analyzed results and the accumulated characteristics. This method calculates symbols for gaze angles like right, left, right to left, etc., using SOM (Self-Organized Map) and determines safety confimation insufficiency based on the time series of the symbols. Finally, the usefulness of the proposed method is demonstrated using the experimental results derived from the driving simulator.