自动驾驶环境下交叉口车辆路径规划与最优控制模型

自动驾驶环境下的交叉口基于车车/车路之间的双向信息交互, 能保障自动驾驶车辆相互穿插与协作地通过交叉口, 而无需信号灯控制. 因此, 如何设计高效的面向自动驾驶车辆通行的交叉口管控模型, 已成为研究的热点. 已有研究在建模时, 均基于自动驾驶车辆在交叉口内部的行驶路径已知并作为模型输入, 且大多对交叉口内部的冲突点进行简化. 本文首先将交叉口空间离散化处理, 考虑车辆的实际尺寸并面向非常规交叉口, 使用椭圆曲线建立转弯车辆行驶路径的精确轨迹方程, 再通过外边界投影降维法建立轨迹方程和交叉口空间的映射关系. 建立了基于混合整数线性规划(Mixed integer linear programming, MILP)的自动驾驶交叉口管控模型, 以交叉口总延误最小为控制目标, 同时优化车辆在交叉口的最佳行驶路径和驶入时刻, 使用AMPL (A mathematical programming language)对模型进行编译并使用CPLEX求解器求解. 与经典感应控制和先到先服务模型进行对比, 结果表明, 本文所提出模型能对车辆进入交叉口的时刻和行驶路径进行双重优化, 显著降低自动驾驶车辆通过交叉口的车均延误, 提高交叉口空间的利用效率.

A Novel Autonomous Vehicle Trajectory Planning and Control Model for Connected-and-Autonomous Intersections

By utilizing the two-way information communication between vehicle and vehicle/infrastructure, autonomous vehicles can traverse the intersection without traditional traffic signals. Therefore, how to design an efficient and automated intersection control model for autonomous vehicles becomes a hot topic in recent years. Most previous studies often assumed that the travel routes/trajectories of autonomous vehicles inside the intersection area are exogenous inputs, and the vehicles' conflicts inside the intersection area have been simplified into a small set of points. In this paper, we firstly discretize the intersection space, and then consider the actual size of the vehicle and the unconventional intersection conflicting points by utilizing the elliptic curve to describe turning vehicles' travel trajectory, and establish the mapping relationship between the trajectory equation and the intersection space by the method of outer boundary projection and dimensionality reduction. Finally, a mixed integer linear programming (MILP) model has been developed to optimally plan and control autonomous vehicle trajectories. By minimizing the total delay of the intersection, the proposed model provides the travel route/trajectory and the time to enter the intersection for each autonomous vehicle, simultaneously. The model is compiled using AMPL (A mathematical programming language) and can be efficiently solved by CPLEX. We then compare the proposed model with traditional actuated signal control and the first-come-first-served model. Our test results clearly indicate that the proposed model can optimize the travel route and entry time for autonomous vehicles efficiently, which decreases the average delay for autonomous vehicles significantly and improves the utilization of the intersection spaces.