智能网联汽车可利用网联信息进行智能协同决控，已被广泛应用于多车道队列及编队控制。现有多车道编队控制主要采用分层式控制架构，存在车群数目固定、编队构型单一的局限，其编队机动性受限。此外，多车道编队控制多采用运动学模型进行状态递推及约束构建，其控制精度较低、安全保障不足。针对上述难题，本文面向多车道场景的智能网联汽车协同驾驶任务，探索更加灵活机动的多车道编队控制架构，考虑网联局部信息进行约束型模型预测控制算法设计，依托多车道仿真场景进行基本编队功能验证，依托校园场景利用网联物流实车平台进行拓展应用功能验证。首先，针对多车道场景下的智能网联汽车，提出了一种实时确立车间队序的集成式协同编队控制架构，将决控流程分为静态轨迹规划、车间队序确立及动态优选跟踪三层。其额外引入的车间队序确立层可利用网联平台及感知信息，在各静态轨迹上按前车距离远近筛选获取跟驰前车集合，宏观确立网联车群队序，为后续模块提供决控基础。在所述控制架构下，智能网联汽车可实现自主编队及离队、构型切换等基本编队功能，满足网联车群数目不定及编队构型灵活切换等需求。其次，针对部分编队控制存在的车辆模型简单、安全保障不足局限，设计考虑非线性动力学模型及多元避撞约束的约束型模型预测控制算法。研究利用后向欧拉法搭建了低速可行的车辆动力学模型，采用双圆轮廓描述构建了多元车辆避撞约束，设计了考虑网联局部信息的约束型模型预测控制算法，在异质交通流仿真平台下开展了多组网联车群编队实验。仿真结果表明：稳态行驶状态下最大横向位置误差在0.2cm内，最大纵向跟驰误差在6.0cm内，满足了协同编队控制的多维性能需求与行驶安全保障。最后，在考虑通信时延的真实路况下验证编队控制性能，在校园场景下依托网联物流实车平台开展编队配送实验。研究对物流实车平台进行了硬件平台搭建及通信方案设计工作，以实现物流实车平台智能化与网联化；开展了控制特性辨识与系统模型重构工作，以提升协同编队控制精度；依托车云长距通信机制对协同编队控制架构进行应用功能拓展，在校园场景下依托 4 台实车开展编队配送实验，实现了网联车群停车取货、分队配送等拓展应用功能。
Connected and automated vehicles, capable of intelligent decision and control based on network information, have been widely applied to platoon and formation control in multi-lane scenarios. Most multi-lane formation control methods utilize a hierarchical decision and control framework with limitations such as a fixed number of vehicles and inflexible formation configuration, leading to inflexible formation control. In addition, most formation control methods utilize kinematic models to roll out states and construct constraints, leading to a lack of control accuracy and safety assurance. To address these difficulties, this paper searches for a more flexible formation control framework for connected and automated vehicles in multi-lane scenarios, which can establish the sequence of vehicles in real-time. A constrained model predictive control algorithm that considers the local network information is designed. The basic formation functions are verified on a multi-lane simulation platform, and the extended logistics functions are verified on a real vehicle platform in a campus scenario.Firstly, a cooperative integrated decision and control framework is proposed for connected and automated vehicles in multi-lane scenarios, which can establish the sequence of vehicles in real-time. It includes three modules: static trajectory planning, formation determining, and dynamic optimal tracking. The additional formation determining module can utilize the network and perception information to obtain the following vehicles set by relative distance on each static trajectory. Such a module can establish the real-time sequence of the connected vehicles in a macroscopic manner and provide the decision-control basis for the subsequent modules. Under the proposed control framework, the vehicle group can realize the basic formation functions of autonomous formation or departure and formation configuration switching to meet the needs of a variable number of connected vehicles and a flexible formation configuration switching process.Secondly, a constrained model predictive control algorithm considering the nonlinear dynamics model and multiple collision avoidance constraints is designed to address the limitations of simple vehicle models and insufficient safety guarantees in some formation control methods. By using the backward Euler method, a low-speed feasible vehicle dynamics model is constructed. The vehicle collision avoidance constraints are described using the bicircular contour. A constrained model predictive control algorithm considering network information is designed. Multiple formation control experiments are conducted under a heterogeneous traffic flow simulation platform. The simulation results show that under steady-state conditions, the maximum lateral tracking error is within 0.20 cm, and the maximum longitudinal car-following error is within 6.0 cm. Hence, the multidimensional performance and safety guarantee of the proposed formation control method are satisfied.Finally, a real-road test based on the connected logistics vehicle platform is conducted under a campus scenario to verify the formation control performance. To realize the intelligence and connectivity of the formation group, this paper reconstructs the hardware platform and designs communication schemes for the connected logistics vehicle platform. Control characteristics identification and dynamics model reconstruction are carried out to improve the accuracy of the proposed formation control method. The cooperative integrated decision and control framework is reconstructed by introducing the vehicle-cloud communication scheme. The real-road test is carried out using four connected logistics vehicle platforms under the campus scenario. The field experiment shows that extended logistics functions, such as parking and picking up, and separate dispatching, are realized.
