考虑自动与人工驾驶车辆运动异质性的冲突风险影响因素研究

Influencing Factors Analysis of Conflict Risk Accounting for Motion Heterogeneity between Automated and Human-Driven Vehicles

摘要: 本研究旨在探究无信号交叉口中自动驾驶车辆(AV)与人工驾驶车辆(HV)混合交通环境下的冲突风险影响因素。研究基于公开数据集Argoverse-2，提取了直行与左转车辆的冲突事件。研究采用后侵入时间(PET)作为风险度量指标，并通过平均残余寿命图法来确定高/低风险冲突的阈值。核心方法上，研究构建了考虑均值异质性的随机参数Logit模型，以分析两种交互场景下(AV左转-HV直行；AV直行-HV左转)车辆运动特征、交互特征以及通行顺序对冲突风险的影响。研究发现：在AV左转场景中，HV直行时的紧急制动是关键风险因素；在AV直行场景中，HV左转时较高的速度和加减速度，以及对AV路权优先性的不遵守，会显著增加冲突风险。研究同时指出，AV能通过主动的速度调控有效降低风险。该研究为混合交通环境下的交叉口安全管理提供了理论依据。

Abstract: This study aims to investigate the factors influencing conflict risk in mixed traffic environments involving an Autonomous Vehicle (AV) and a Human-driven Vehicle (HV) at unsignalized intersections. Based on the publicly available Argoverse-2 dataset, conflict events between straight-moving and left-turning vehicles were extracted. Post-Encroachment Time (PET) was adopted as the risk metric, and the mean residual life plot method was applied to determine the threshold between high-risk and low-risk conflicts. A random parameters Logit model with heterogeneity in means was developed to analyze the effects of vehicle motion features, interaction features, and passing order on conflict risk under two interaction scenarios (AV left turn and HV straight; AV straight and HV turning left). The results indicated that in the AV left-turn scenario, emergency braking by a straight-moving HV was a key risk factor. In the AV straight-moving scenario, higher speed and greater acceleration/deceleration of a left-turning HV, as well as non-compliance with the right-of-way priority of the AV, significantly exacerbated conflict risk. It was also found that the AV could effectively reduce risk through active speed regulation. Theoretical support for intersection safety management in mixed traffic environments can be provided by these findings.