| Traffic accidents have become an important obstacle hindering the improvement of traffic safety and the public has paid more and more attention to traffic safety.The human factor is the main cause of traffic accidents.Among them,the driver’s ability of hazard perception is of great significance to ensure traffic safety and reduce traffic accidents.This dissertation studies the driver’s hazard perception ability,explores the factors that affect the ability of hazard perception and the indicators reflecting that,so as to scientifically intervene the drivers to improve the level of traffic safety.Firstly,the dissertation builded virtual driving simulation platforms for various types of dangerous driving scenes and recruited the participants to carry out the test.During the test,the driving behavior and eye movement data of the subjects were obtained.Then,the two types of data were applied to analyze the performance differences of different groups of drivers in dangerous scenes.With the help of eye movement indicators,the differences between dualtarget scenes and single-target scenes,and between visible hazard scenes and invisible hazard scenes were emphatically analyzed.The results show that gender,driving experience,and types of dangerous scenes are all factors that affect the driver’s ability to perceive hazards.The average speed and average depth of brake pedal of female drivers are significantly lower than that of male drivers,but their number of fixations to hazards is lower than that of men.Compared with novice drivers,experienced drivers have more attention to hazards and responds to the hazard earlier.It is found that the horizontal spread of search of the experienced driver is significantly larger than that of the novice driver.This visual search mode makes the experienced driver more safe in the driving process.In addition,the driver’s overall response to the hazard in the single-target scenario is slower than the dual-target scenario and the perception of the visible hazards is also worse than the invisible hazards.However,in a further comparative study,it was found that compared with a single-target scenario,the driver’s perception to visible hazards in a dual-target scenario is better,but the perception to invisible hazards is worse,that is,in the dual-target scenario,the driver’s perception to the visible hazard hindered their perception to the subsequent invisible hazard in the same scene,that is,a “subsequent search misses”(SSMs)occurred.A study of the area of interest found that although there were no significant differences in the number of fixations on traffic signs among driver groups,experienced drivers had more fixations on construction signs than novices,but less attention on attention pedestrian signs.Finally,the TTC index was used to judge the driver’s performance and hazard perception ability in the scene,and the hazard perception ability is graded.Based on the pattern recognition method,multiple discriminant models of hazard perception ability were established,that is,identify and predict the driver’s level of hazard perception through driving behavior indicators and eye movement indicators.And by comparing the performance characteristics of the discriminant model,a better recognition method for the driver’s hazard perception ability was selected.The results show that the quadratic discriminant model containing driving behavior indicators and eye movement indicators has higher accuracy and lower false alarm rate. |
