Joint Optimization of Longitudinal and Lateral Locations of Autonomous-Vehicle-Dedicated Lanes on Expressways

Appropriate deployment of autonomous-vehicle-dedicated lanes (AVDLs) is crucial for optimizing transportation system with mixed human-driven and autonomous vehicles. Existing studies mainly focus on matching the capacity and traffic flow volume to maximize throughput under different AV penetration rate, while lane changing scenarios and network geometry impacts were ignored. Therefore, the lateral locations in optimal deployment on expressway cannot be determined. This study developed an analysis framework to jointly optimize the longitudinal and lateral locations of AVDLs on expressways. The problem is formulated as a bi-level optimization model. In the lower level, a lane-level multiclass equilibrium assignment model is built to predict traffic flow distribution among lanes on both longitudinal and lateral locations regarding AVDL deployment plan. In the upper level, a 0–1 integer programming model is established to obtain the optimal deployment plan of the AVDL with minimal total travel time (TTT). Additionally, a problem-specific heuristic-based adaptive large neighborhood search algorithm is developed to solve the problem. The advantages of the proposed model are validated in North–South Elevated Expressway, Shanghai, China. Results indicate that the proposed framework outperforms existing methods in terms of TTT reduction. Finally, sensitivity analysis demonstrates the robustness of the developed method in reducing the TTT under different headways, demand levels, and OD distributions. The proposed framework enables the optimal deployment of AVDLs regarding both longitudinal and lateral locations in urban expressways containing multiple on- and off-ramps, which supports decision-making for managing expressway with mixed HV and AV traffic flow.