伴随自动驾驶技术逐渐普及，网联自动驾驶车辆的引入将改变传统交通组成特征，为交通管理带来新的挑战。实施专用路权管理可以充分发挥自动驾驶交通流特性，优化交通整体表现。本文聚焦混合驾驶单交叉口和城市道路网络场景，面向不同自动驾驶渗透率条件，基于交通优化等技术研究专用道布设、车流引导和信号控制联合优化模型。面向未来全自动驾驶场景探索车辆级路权分配方法，本文应用深度强化学习技术研究自动驾驶高效时空路径规划策略，辅助交通管理者进行科学决策。具体地，本文的主要工作内容和创新点包括：1.研究混合驾驶场景单交叉口自动驾驶专用道布设、车流引导和信号控制联合优化问题。本文首先提出基于给定车道配置和车流分配的最佳信号控制问题。接着推导出最优流量分配和最优车道配置问题的理论特征。之后将专用车道配置、车流引导和信号控制的联合优化建构成混合整数非线性规划问题，基于理论特征提出一种启发式算法有效解决该问题。本文通过数值算例验证算法的有效性，同时在不同车道数场景下分析使专用道发挥优势的自动驾驶渗透率条件。2.研究混合驾驶场景城市路网自动驾驶专用道布设、车流引导和信号控制联合优化问题。该部分工作推导混合驾驶场景下车辆物理排队延误模型，将联合优化问题融入路网交通分配框架，同时系统评估由于设置专用道导致额外换道造成的延误。本文将路网联合优化问题拆分成交通分配和交叉口联合优化两个子问题，进行交替迭代直至收敛。本文通过数值算例验证在路网中布设专用道的优势。3.研究全自动驾驶场景城市路网车辆级路权分配及优化方法。面向未来自动驾驶渗透率达到100\%场景，本文建立冲突点网络并提出两种时空路径规划算法。本文提出“车队策略”，组织自动驾驶车辆形成车队通过冲突点，并采用深度强化学习动态优化车队规模。数值测试表明，本文所提出的算法可以在求解质量和计算负荷之间取得平衡，保障实时求解。算法在低需求场景下车辆延误接近于0。在高需求场景下通过车队策略大幅减少拥堵导致的的平均车辆延误。以上三部分内容对智能网联环境专用路权管理和交通控制优化开展了系统性研究。本文深入讨论多种交通场景和条件下设置自动驾驶专用道的优势，提出一套完整分析方法，为城市道路交通管理决策提供支持。

As autonomous driving technology gradually becomes popular, the introduction of autonomous vehicles (AV) will change the traditional traffic, and mixed traffic flow will bring new challenges to traffic management. Implementing dedicated right-of-way management can fully leverage the characteristics of AVs and optimize the overall traffic performance. This study focuses on the isolated intersection and urban network scenarios, targeting different AV penetration rate conditions. Based on traffic optimization and related technologies, a joint optimization model for dedicated AV lane deployment, flow distribution, and signal control is developed. At the same time, vehicle-level right-of-way allocation methods are explored for future scenarios with fully autonomous driving, and efficient space-time routing strategies for AVs are researched using technologies such as deep reinforcement learning to assist traffic managers in making scientific decisions. The primary contributions and novel ideas of this study are as follows.1. Study the joint optimization of dedicated AV lanes deployment, flow distribution and signal control at isolated intersections under mixed autonomy. The study first proposes an optimal signal control problem based on a given lane configuration and flow assignment. Then the theoretical characteristics of the optimal lane-level flow assignment and optimal lane configuration problems are derived. After that, the joint optimization of dedicated AV lane configuration, flow distribution and signal control is constructed as a mixed integer nonlinear programming problem, and a heuristic algorithm is proposed to solve the problem effectively based on the theoretical features. Numerical examples verify the effectiveness of the algorithm and analyze the conditions of AV penetration that enable the advantage of dedicated AV lanes under different numbers of lanes in one approach.2. Study the joint optimization of dedicated AV lanes deployment, flow distribution and signal control under mixed traffic network. The approach integrates the joint optimization problem into the network traffic assignment framework and systematically evaluates the impact of dedicated AV lanes on traffic delays resulting from additional mandatory lane changes. The study decomposes the network-level joint optimization problem into two sub-problems: network-level traffic assignment and intersection-level joint optimization. Alternating iterations are performed until convergence is achieved. The numerical results confirms that the distinct path selection strategies of autonomous and manually driven vehicles lead to different traffic allocations in the network. Consequently, the benefits of assigning dedicated lanes for autonomous driving are more pronounced in the road network.3. Study the vehicle-level right-of-way allocation and optimization methods for urban networks in fully automated driving scenarios. This part of the research establishes a conflict point network and proposes two space-time routing algorithms for a future scenario with 100\% autonomous driving penetration. The research develops a "platoon strategy" to organize autonomous vehicles to form a platoon through conflict points, and uses deep reinforcement learning to dynamically optimize the platoon size. Numerical tests show that the algorithm can achieve a balance between solution quality and computational load, guaranteeing real-time solutions with delays close to 0 in low-demand scenarios, and that the platoon strategy can significantly reduce the average vehicle delay under congestion in high-demand scenarios.The aforementioned three parts provide a systematic study of the management and control methods for the introduction of AVs into traditional traffic. This paper discusses in depth the advantages of setting up dedicated lanes for autonomous driving under a variety of traffic scenarios and conditions, and proposes a complete set of analysis methods to support urban road traffic management decisions.