Although Autonomous Vehicles (AVs) will enhance mobility and safety, their impact on congestion is not clear yet. AVs may increase roadway capacity due to their connectivity features. The capacity enhancement highly depends on the AV proportion in traffic. This study models user equilibrium traffic assignment when the link capacity is a function of AV proportion of traffic. The mixed traffic flow of AVs and human-driven vehicles is considered as a multiclass traffic assignment problem. This problem is formulated as a non-linear complementarity problem which is solved to find optimal traffic management policies. We show that simple policies such as AV exclusive links can improve network performance in mixed traffic of AVs and human-driven vehicles. We also show that if these policies are implemented the network performance would be very close to system optimal condition even when users choose their routes selfishly following a user equilibrium. Results of numerical examples for a real size network show that management policies can decrease the gap between user equilibrium and system optimal to less than 1%.

虽然自动驾驶汽车（AVs）会增强机动性和安全性，目前尚不清楚它们对拥堵的影响。视音频设备的连接功能，它们可能会增加道路通行能力。容量的提高很大程度上取决于AV在流量中的比例。当链路容量是流量AV比例的函数时，本研究对用户均衡流量分配进行建模。影音和人类驾驶车辆的混合交通流被认为是多类交通分配问题。此问题被公式化为非线性互补问题，可以解决该问题以找到最佳的流量管理策略。我们表明，诸如AV专有链接之类的简单策略可以提高AV和人工驾驶车辆的混合流量中的网络性能。我们还表明，如果实施了这些策略，即使用户遵循用户均衡自私地选择其路由，网络性能也将非常接近系统最佳条件。实际规模网络的数值示例结果表明，管理策略可以将用户均衡与系统最优之间的差距减小到小于1％。