Abstract:

Hazard perception refers to drivers’ cognitive process of identifying, predicting and reacting to the obvious and latent hazard. Previous studies have shown that both experienced and novice drivers were good at detecting pedestrian-related hazard. We doubted whether the experienced drivers can detect pedestrians well if the hazard were combined with a potential hazard(eg pedestrian and car braking together). We therefore, conducted two experiments to explore different drivers’reaction time and their eye movements by setting two traffic scenarios:single hazardous condition( only pedestrians) and multiple hazardous conditions(pedestrian combined with the vehicle in front of the driver). In the first study, we used Tobii T120 type eye tracker to examine the behavior and eye movements by combined the reaction method and eye movements technique. Experiments is a 2 (traffic scene: pedestrian scene, pedestrians - the front car scene) ×3 (group: no driver's license, novice, experienced driver) mixed design, with traffic scene being the the within subject variable and the groups being the between subject variable. Participants were asked to watch a video of traffic scenes, when they detected a hazard, they should press the key immediately. The results show that: In the case of pedestrian-car condition, the experienced group’s reaction time is faster than novice group (p = .039) and no driver's license group (p = .013), F (2,30) = 3.98, p = .029. No driver's license group’s average fixation duration on pedestrian was longer than the experience group (p = .043) and the novice group (p = .024), F (2,30) = 3.39, p = .047, η2 = .18. No driver's license group’s level of searching breadth was less than (p = .006) the experienced group and the novice group (p = .016), F (2,30) = 5.17, p = .012, η2 = .26. In the second study, we further examined the impact of double hazard scene on drivers’ hazard perception and information processing efficiency. Experiments is a 2 (AOI type: pedestrian, the vehicle in front) × 3 (group: no driver's license group, the novice group, experienced group) two-factor mixed design, with AOI type being the the within subject variable and the groups being the between subject variable. The results showed that: the experienced group’s total fixation time on the front car was significantly shorter than the novice group (p = .014) and no driver's license group (p = .003), F (2,30) = 5.99, p = .006 . The experienced group’s average search time of the front car was significantly shorter than the novice group (p = .014) and no driver's license group (p = .004), F (2,30) = 5.50, p = .009. In the case of the front car condition, the average search time is longer, the longer the time their reaction time to pedestrians by controlling of the sex and age (β = .393, p <0.05, R2 = .308). These results indicate that: the inexperienced and novice drivers’ bad behavior were not just due to the fact that operations of the car snatch cognitive resources or caused by emotional factors. Lower efficiency of information processing, the inflexible search mode will also leads to a higher accident rate, which is one of the reasons for the novice driver traffic collisions. In the future, driver training should be strengthened for the novice driver on their visual search skills, and continue to provide more educational opportunities for the drivers.

摘要：

危险知觉是指驾驶员在交通情境中对明显危险和潜在危险的识别、预测和反应的认知加工过程，是驾驶员对当前交通情境形成的连续而富于变化的复合认知表征。以往研究表明，对于行人过马路这类明显危险，新手和经验驾驶员危险知觉表现都非常出色。但是当明显危险（行人）和潜在危险（前车制动）并存时，经验驾驶员对明显危险的知觉反应是否会突显出优势呢？本研究通过设置两种交通场景：行人单一危险条件和行人-前车双重危险条件，考察不同经验人群对行人危险的反应及眼动特点。在第一个研究中，使用Tobii T120型眼动仪，通过反应时和眼动结合的方法，考察不同危险场景中，不同驾驶经验对驾驶员知觉行人危险的行为反应数据（反应时和正确率）及眼动指标（平均注视时间和水平搜索广度）的影响。采用2（交通场景：行人场景、行人-前车场景）×3（组别：无驾照组、新手组、经验组）两因素混合实验设计，其中交通场景是组内变量，组别为组间变量。因变量为驾驶员对行人危险进行反应的行为数据（反应时和正确率）及眼动数据（平均注视时间和水平搜索广度）。实验要求被试观看交通场景视频，当发现危险时，立刻按键反应。实验结果表明：在有前车条件下，经验组被试的反应时快于新手组（p=.039）和无驾照组（p=.013），F(2，30)=3.98，p=.029。无驾照组被试对行人的平均注视时间长于经验组（p=.043）和新手组（p=.024），F(2，30)=3.39，p=.047，ηp2= .18。无驾照组被试对场景的水平搜索广度均小于经验组（p=.006）和新手组（p=.016），F(2，30)=5.17，p=.012，ηp2= .26。在第二个研究中，对研究一的数据进行了深入分析，考察双重危险场景中驾驶经验对驾驶员知觉行人危险及信息加工效率的影响。采用2（兴趣区类型：行人、前车）×3（组别：无驾照组、新手组、经验组）两因素混合实验设计，其中兴趣区类型是组内变量，组别为组间变量。因变量为搜索行人和前车时的眼动数据（总注视时间、搜索次数和平均搜索时间）。研究结果表明：经验组驾驶员对前车的总注视时间显著短于新手组（p=.014）和无驾照组（p=.003），F(2，30)=5.99，p=.006。经验组驾驶员对前车的平均搜索视时间显著短于新手组（p=.014）和无驾照组（p=.004），F(2，30)=5.50，p=.009。在有前车条件下，控制了性别、年龄后，驾驶员对前车的平均搜索时间越长，对行人的危险反应时越长（β=.393，p＜0.05，R2=.308）。这些结果说明：无经验和新手驾驶员糟糕的搜索行为不仅仅是由于操作抢夺认知资源或者情绪紧张导致的。危险的信息加工效率更低，搜索模式不够灵活，也是导致新手驾驶员行人碰撞事故率更高的原因之一。驾驶员培训有助于提高驾驶员对明显危险的信息加工效率和搜索模式灵活性。积累更多的驾驶经验可以提升驾驶员对潜在危险的的视觉搜索和注意资源分配效率，从而加快危险反应时间。未来的驾驶员培训应加强对新手驾驶员搜索技能的培训，并持续提供更多的驾驶员继续教育机会。