路段环境自动驾驶汽车通行权决策方法

为了解决路段自动驾驶汽车的通行权决策问题，提高交通流的运行效率和稳定性，基于可接受间隙模型和谈判理论构建路段自动驾驶汽车通行权决策模型. 综合考虑多种因素，基于可接受间隙模型对行人风险进行建模，将行人风险划分为低风险、中风险和高风险. 综合考虑风险、性格（激进和保守）和等待时间等对行人行为的影响，分析不同因素组合下行人和自动驾驶汽车可能采取的行为策略，基于该行为策略，利用谈判理论对自动驾驶汽车的通行权决策过程进行建模. 利用Python联合SUMO开源交通仿真软件对模型进行验证，仿真持续10 h. 3个模型的（保守模型、Gupta模型和本文模型）仿真结果表明，当行人产生频率为15 s时，自动驾驶汽车的平均行驶时间分别为661.5、399.5和327.6 s，平均延误时间分别为618 s、336 s和260.7 s，总流量分别为6 699辆、10 583辆和11 568辆. 当行人产生频率为30 s时，自动驾驶汽车的平均行驶时间分别为643.5、311.7和81.9 s，平均延误时间分别为599.9、244.4和6.5 s，总流量分别为6 879辆、11 741辆和11 971辆. 通行权决策方法的加入有助于降低自动驾驶汽车的行驶时间和延误，提升流量.

Decision-making method of autonomous vehicles for right of way on road segments

A decision-making method for the right of way based on acceptable gap model and negotiation theory was developed in order to solve the decision-making problem of autonomous vehicles for the right of way on road segments and improve the efficiency and stability of traffic flow. Various factors were comprehensively considered to model the perceived risk based on the acceptable gap model. Perceived risk was divided into low risk, medium risk and high risk. The potential behavior strategies of pedestrians and autonomous vehicles under different combinations of above factors were analyzed by comprehensively considering the impact of risk, personality (radical and conservative) and waiting time on pedestrians’ behaviors. The negotiation theory was used to model the process of decision-making for the right of way based on these behavior strategies. The model was simulated and verified by using Python and SUMO for ten hours. The simulation of three models (conservative model, Gupta model and our model) was conducted. When pedestrian generation frequency was 15 s, the average travel time of autonomous vehicles was 661.5, 399.5 and 327.6 s respectively; the average delay was 618, 336 and 260.7 s respectively; the total traffic volume was 6 699, 10 583 and 11 568 vehicles respectively. When the pedestrian generation frequency was 30 s, the average travel time of autonomous vehicles was 643.5, 311.7 and 81.9 s respectively; the average delay was 599.9, 244.4 and 6.5 s; the total traffic volume was 6 879, 11 741 and 11 971 vehicles respectively. The introduction of decision-making model helps to reduce the travel time and delay of the autonomous vehicles, and increase the traffic volume.