Lane-Change Leading Strategy for Cav-Dedicated Lanes in Mixed Traffic Environments Considering Human Driver Randomness and Compliance

Trajectory planning for connected and automated vehicles (CAVs) can substantially improve both the operational efficiency and energy efficiency of mixed traffic flow. Despite abundant research in this field, current studies primarily focus on single-lane scenarios and often overlooks the randomness and compliance of human-driven vehicles. This study proposes a Lane-Change Leading (LCL) strategy for CAV-dedicated lanes in mixed traffic environments, which accounts for the randomness and compliance of connected human-driven vehicles (CHVs), as well as the lane-changing behavior of CAVs. The LCL control framework contains four modules: arrival time prediction, leading demand generation, cooperative lane change, and longitudinal trajectory optimization, thereby decoupling the complexity control tasks. First, a Time-Space-State network is developed for CAVs, and a dynamic programming algorithm based on state space reduction is designed to further reduce the search space. Second, a stochastic car-following model, a compliance model for suggested speed, and a leading-demand generation model are proposed for CHVs to comprehensively capture their uncertainties.

Finally, a parallel receding horizon framework is applied to address uncertainty in mixed environments. Numerical simulations demonstrate that the LCL strategy reduces fuel consumption and decreases the frequency of vehicle stops. The sensitivity analysis indicates that a CHV compliance rate exceeding 50% is sufficient to ensure the satisfactory performance of the proposed strategy. The results can provide valuable insights on CAV-dedicated lane management and cooperative eco-driving of mixed platoons.