面向换道冲突互信息的复杂交织区车辆风险识别

Vehicle risk identification in complex weaving areas for conflicting mutual information on lane change

为提升复杂交织区行车安全性,提出了一种考虑换道冲突互信息的车辆换道风险识别方法。首先,基于无人机视频提取复杂交织区的微观换道轨迹,测算换道风险关键参数;然后,考虑换道紧邻车辆信息,聚焦于车辆换道行为的矢量性,改进拓展TTC理论,构建复杂交织区换道风险识别模型,并对换道风险的等级进行细分;最后,基于实测数据进行模型的优化与验证。结果表明,模型可有效反映汇入或汇出主线的迫切需求,描述车辆在不同位置、不同方向的换道风险差异。研究成果有助于智能网联环境下复杂交织区的车辆换道决策与轨迹优化,为交管部门制定动态预警方案提供理论支持。

Based on high-precision micro-vehicle trajectory data, this paper proposes a method for the lane-changing risk identification of vehicles in complex weaving areas that can be applied to ICV. Firstly, based on the UAV video, the lane-changing trajectory with a time precision of 0.1 s and the spatial precision of 0.1 m/pixel in the complex interleaving area is extracted from a wide-range perspective. The lane-changing risk perception information such as vehicle distance, velocity vector, acceleration, approaching rate, and velocity angle is measured. A lane-changing risk identification model for complex interweaving areas is constructed. This model can effectively reflect the urgent needs of importing or exporting the mainline and describe the differences in the lane-changing behavior of vehicles at different positions. Moreover, this model evaluates the dynamic risk of lane changing in real-time based on indicators such as the ratio of lane changing in different directions and the density of lane changing;then, the risk level of lane changing is divided according to the TTC threshold. In addition, the results of risk level classification are verified. The results show that the lane-changing risk in the complex weaving area can be accurately recorded. The process of changing lanes in all directions is also accurately captured. The risk distribution shows a trend of radiating outward along the main road and gradually decreasing with the weaving area as the center. Among them, the first-level high lane-changing risk accounts for 4.55%, the second-level medium-level lane-changing risk accounts for 20.58%, the third-level low lane-changing risk accounts for 4.60%, and the situation with basically no potential lane-changing risk accounts for 70.27%. Being helpful for the lane-changing decision and trajectory optimization of the connected vehicle, the results of this research help guide the traffic control department to formulate the ICV dynamic early warning scheme.