Parameters for each approach of the intersection.

In this paper, we propose a right-of-way optimization model considering multi-objective DEA evaluation for intersections in mixed driving environments with automated and human driving. Considering average speed, number of cars, penetration of automated vehicles, queuing pattern, left-turn rate, and number of buses as factors influencing intersection rights-of-way. Comprehensively consider the per capita delay, travel time and traffic volume as the optimization objectives, and then determine the weights of the three optimization objectives for each strand of traffic flow, and calculate the cross-benefit by interchanging the weight evaluation through the Crossing Efficiency Evaluation Method (CREE) to determine the optimal order of traffic flow in each direction at the intersection. In this paper, the optimization strategy is compared with existing benchmarks (e.g., actuated control) using SUMO simulation software, and the simulation results show that the proposed optimization strategy is able to shorten the per capita delay and travel time at intersections in order to improve the efficiency of the traffic flow compared to actuated control and the First-Come, First-Served strategy.

本文针对自动驾驶与人工驾驶混合的交通环境下的交叉口，提出一种考虑多目标数据包络分析（Data Envelopment Analysis，DEA）评价的通行权优化模型。研究选取平均车速、小型汽车数量、自动驾驶车辆渗透率、排队特征、左转率及公交车数量作为交叉口通行权的影响因素，以人均延误、出行时间与交通流量为综合优化目标，进而为各交通流单元确定三项优化目标的权重；随后通过通行效率评价法（Crossing Efficiency Evaluation Method，CREE）开展权重互换评价以计算交叉效益，最终确定交叉口各方向交通流的最优通行次序。本文借助SUMO仿真软件，将所提优化策略与现有基准策略（如感应控制）进行对比。仿真结果显示，相较于感应控制与先来先服务策略，本文提出的优化策略可有效缩短交叉口的人均延误与出行时间，进而提升交通流运行效率。