交通冲突
观测误差
可靠性(半导体)
计算机科学
弹道
度量(数据仓库)
噪音(视频)
测量不确定度
毒物控制
随机误差
数据挖掘
统计
模拟
计量经济学
共同事业与特殊事业
系统误差
准确度和精密度
误差分析
噪声测量
计算机安全
错误检测和纠正
实时计算
流量分析
冲突理论
不确定性传播
作者
Yang-Jun Joo,Zubayer Islam,Mohamed A. Abdel-Aty,Dong‐Kyu Kim
标识
DOI:10.1177/03611981251398748
摘要
Traffic conflict analysis emerges as a proactive approach in traffic safety studies, complementing traditional crash-based methods. However, positional measurement errors in trajectory data compromise the accuracy of traffic conflict measures and reduce the replicability of traffic conflict-based safety analysis. This study examines the impact of these errors, analyzing conditions that amplify their effects. We injected controlled random noise into precise vehicle trajectories to observe their propagation into two traffic conflict measures: time-to-collision (TTC) and modified time-to-collision (MTTC). This study found that unbiased positional measurement errors in trajectory data tend to bias traffic conflict measures toward an overestimation of risk. This is contrary to the symmetrical effect expected from the initial error distribution. Among the conflict types, sideswipe conflict is most vulnerable to errors, with the error decreasing as the conflict angle approaches a right angle. Conflicts involving low relative distances, speeds, and accelerations are particularly vulnerable to inaccuracies from positional measurement errors. In addition, different conflict measures have varying sensitivities to positional measurement errors. MTTC is about four times more sensitive to error than the TTC owing to the underlying errors in the kinetic quantity of each measure used. The results emphasize the importance of assessing positional measurement errors in trajectories before traffic conflict analysis. This approach guides researchers to avoid using error-sensitive measures when positional inaccuracies are high, thereby improving the reliability and replicability of conflict-based safety analyses across various traffic conditions.
科研通智能强力驱动
Strongly Powered by AbleSci AI