Effect of hazard perception training with artificial near-miss accident scenes made by adding computer graphics to real traffic recordings

Although hazard perception training with accident videos is effective, accident videos can only be recorded by chance. It is impossible to carry out training which covers all desired points. To solve this problem, we created near-miss accident scenes by adding computer graphics to traffic recordings taken by experimenters during normal traffic situations on commuting route of participants. Eleven participants were recruited to evaluate the effectiveness of training using these scenes. Each of their vehicles was equipped with a drive recorder to document their driving behavior while commuting. After five round-trips, the participants took part in training utilizing these near-miss scenes. After training, five additional round-trips were recorded, and their driving behaviors before and after training were then compared. No significant differences were observed in driving speed; however, both duration and frequency of eye fixation at intersections along their commuting route were significantly increased.