关键词: 交通工程； 碰撞风险评价； 交通冲突分析； 环行交叉口； 安全势场

Abstract: To address the issues of traffic conflict analysis methods using surrogate safety indicators in the evaluation of vehicle collision risk at unsignalized roundabouts such as difficult detection of conflict vehicle, difficult modeling of multi-vehicle conflicts and road boundary constraints, a vehicle collision risk assessment method based on safety potential field theory was proposed. Firstly, a classified detection method of conflict vehicles was designed to identify rear-end conflicts on circular lanes and lateral conflicts on entrances of the lanes by improving the time-to-collision (TTC) indicator. Then, based on the safety potential field theory, potential and kinetic fields were respectively constructed, aiming to impacts of road boundaries, lane markings and surrounding vehicles on the subject vehicle. Furthermore, based on the safety potential field resulting from the superposition of these factors, a collision risk assessment indicator, namely CPF (collision potential field), was proposed. Finally, the particle swarm optimization algorithm was applied to carry out parameter calibration of CPF indicator, based on real vehicle trajectories collected from an unsignalized roundabout. The applicability of the conflict vehicle detection method and the CPF indicator was verified from the perspectives such as conflict severity, conflict vehicle type and conflict event type. The results show that the proposed conflict vehicle detection method identifies 6568 pairs of conflict vehicles, which is more than 4097 pairs identified by the virtual lane method. The proposed method also shows higher sensitivity in detecting conflict events with different severities and improves the defect that the virtual lane method cannot identify conflict events at the entrance of circular lanes. The potential field exerted by lane markings, road boundaries, and surrounding vehicles, which is measured by the CPF indicator, can reflect the conditions in real driving scenarios. Compared to the traditional TTC indicator, the CPF indicator is more sensitive and accurate in reflecting the dynamic variation features in collision risk in specific scenarios, such as changes in the driving direction of conflicting vehicles, proximity of conflicting vehicles, and deceleration of the rear vehicle.

Key words: traffic engineering; collision risk assessment; traffic conflicts analysis; roundabout; safety potential field