Developing Advanced Road Safety Analysis Techniques based on Understanding of Human Psychology and Behavior

过去，道路安全一直是根据历史事故记录进行评估的。然而，由于碰撞的罕见发生和收集碰撞数据所需的时间延长，使用替代安全措施（SSM）进行安全评估最近很普遍。SSM通常使用车辆轨迹计算，并确定车辆之间碰撞的风险。考虑到车辆轨迹还不容易获得，复制单个车辆运动的计算机交通模拟模型被用来生成车辆轨迹并计算SSM。然而，由于缺乏对人为因素的考虑，现有的仿真模型产生的车辆轨迹并不真实，对实际碰撞风险的描述存在偏差。尽管驾驶中的人为因素在心理学研究中得到了广泛的研究，但在碰撞风险的模拟模型和预测中，人为因素还没有得到充分的适应。因此，本研究旨在利用SSM弥合人为因素、交通模拟模型和安全评估之间的差距。本研究的目的是将人为因素纳入交通模拟模型和SSM中，以发展先进的道路安全分析技术。该研究将修改现有的车辆跟随和变道模型，这些模型包括交通模拟模型，以反映驾驶员在各种周围环境中的实际行为。为此，我们将使用驾驶模拟器观察驾驶员在受控交通和天气条件下的行为。该研究还将开发SSM，以捕捉不同驾驶员和驾驶条件下驾驶员感知、反应时间和减速行为的变化。该研究将验证所提出的仿真模型和SSM反映实际驾驶员行为和碰撞风险使用真实世界的数据。本研究亦会利用建议的SSM评估驾驶员警告信息对安全的影响，以展示SSM的潜在应用。虽然自动驾驶汽车技术的发展正在进行中，但半自动驾驶汽车仍然需要部分人为干预，并且会在同一条道路上与手动车辆发生冲突。因此，在交通仿真模型中可以考虑人工驾驶和半自动驾驶车辆的人为因素差异。然后，利用该模型和SSM预测不同混合方式下的交通性能和碰撞风险的变化。因此，这项研究将为确定未来加拿大道路安全面临的重要挑战的方法的发展提供见解。

Road safety has been evaluated using historical crash records in the past. However, due to rare occurrence of crashes and the extended time required for collecting crash data, the use of surrogate safety measures (SSM) for safety evaluation has been prevalent recently. The SSM are typically computed using vehicle trajectories and they determine the risk of collision among vehicles. Given that vehicle trajectories are not readily available yet, computerized traffic simulation models which replicate individual vehicle movements have been used to generate vehicle trajectories and compute the SSM. However, due to a lack of consideration of human factors, the existing simulation models produce unrealistic vehicle trajectories and misrepresent actual crash risk. Although human factors of driving have been extensively investigated in psychology studies, human factors have not been sufficiently adapted in the simulation models and prediction of crash risk. Thus, this research aims at bridging the gap among human factors, traffic simulation models and safety assessment using the SSM. The objective of this research is to incorporate human factors in the traffic simulation model and SSM for developing advanced road safety analysis techniques. The study will modify the existing car-following and lane-changing models which comprise the traffic simulation model to reflect actual driver behaviour in various surrounding environments. For this purpose, driver behaviour will be observed in controlled traffic and weather conditions using a driving simulator. The study will also develop the SSM which capture variations in drivers' perception and reaction time and deceleration behaviour among different drivers and driving conditions. The study will validate that the proposed simulation model and SSM reflect actual driver behaviour and crash risk using the real-world data. The study will also evaluate safety impacts of the driver warning information using the proposed SSM to demonstrate potential application of the SSM. While the development of automated vehicle technology is in progress, semi-automated vehicles still require partial human intervention and will have conflicts with manual vehicles on the same roadway. In this regard, differences in human factors between manual and semi-automated vehicles can be incorporated in the traffic simulation model. Then, the model and SSM can be used to predict changes in traffic performance and collision risk with different mix of manual and automated vehicles. Thus, this research will provide insights into the development of the method for identifying important challenges with Canada's road safety in the future.