Strategic and tactical decision-making for cooperative vehicle platooning with organized behavior on multi-lane highways
Strategic and tactical decision-making for cooperative vehicle platooning with organized behavior on multi-lane highways
复制标题
多车道高速公路上有组织行为的合作车辆编队的战略和战术决策
DOI：
10.1016/j.trc.2022.103952
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Ma, Jiaqi
中科院分区：
文献类型：
--
作者：
Han, Xu;Xu, Runsheng;Xia, Xin;Sathyan, Anoop;Guo, Yi;Bujanović, Pavle;Leslie, Ed;Goli, Mohammad;Ma, Jiaqi
驾驶自动化和车对车(V2V)通信提供了部署协作自动驾驶系统(C-ADS)的机会，以实现可持续发展、安全和效率等交通系统目标。在C-ADS的各种应用中，车辆编队通过建立C-ADS车辆之间的轨迹感知V2V合作策略，具有实现上述系统管理目标的巨大潜力。此前，合作自适应巡航控制(CACC)的概念-即多辆车辆紧密跟随的单车道分散自组织操作-已被研究人员广泛研究。该研究在已有研究的基础上，提出了一种综合的多车道排队算法，该算法通过层次化的框架来组织行为。该算法采用现代先进的SOTA C-ADS软件平台框架，由感知层、规划层和控制层组成。多车道排队算法结合了战略层(即任务层)和战术层(即运动层)的决策，以应对复杂的多车道高速公路挑战，包括相同车道的排列、多车道的连接和入口匝道的合并。根据算法的策略，排长在排员和外部车辆之间进行协调，以指导排长通过复杂和现实的驾驶场景。在战略任务层，提出了一种基于确定性有限状态机(FSM)的排队行为协议，用于指导成员的操作。此外，针对启发式协议在明确表达复杂协作场景方面的不足，以有限状态机为基线训练遗传模糊系统，以扩展算法在协作入口匝道合并场景下的能力。在战术运动层面，提出了一般ADS机动(即车道跟随和换道)的轨迹生成和排队行为规则，使得其他相关车辆的规划轨迹能够被充分考虑(即具有预测性的意图共享)。在交通模拟器和自动驾驶模拟器上进行了性能评估，结果表明，所提出的综合多车道排队算法能够有效、安全地调节装有C-ADS的车辆的行为，满足系统目标。
Driving automation and vehicle-to-vehicle (V2V) communication provide opportunities to deploy cooperative automated driving systems (C-ADS) for transportation system goals such as sustainability, safety, and efficiency. Among various C-ADS applications, vehicle platooning has great potential to achieve the above system management goals by establishing trajectory-aware V2V cooperative strategies among C-ADS vehicles. Previously, the concept of cooperative adaptive cruise control (CACC)—that is, single-lane decentralized ad-hoc operations of multiple vehicles that closely follow each other—has been studied by researchers extensively. This study builds upon the existing research and proposes a comprehensive multi-lane platooning algorithm with organized behavior via a hierarchical framework. The proposed algorithm adopts the modern state of the art (SOTA) C-ADS software platform framework, which consist of perception, plan and control levels. The multi-lane platooning algorithm incorporates both the strategic level (i.e., mission level) and the tactical level (i.e., motion level) decision-making to cope with complex multi-lane highway challenges, including same-lane platooning, multi-lane joining, and on-ramp merging. Based on the algorithm’s strategies, the platoon leaders coordinate between platoon members and external vehicles to guide the platoon through complicated and realistic driving scenarios. On the strategic mission level, a platooning behavior protocol based on a deterministic finite state machine (FSM) is developed to guide the member operations. Additionally, as heuristic protocols fall short in explicitly expressing complex cooperative scenarios, a genetic fuzzy system was trained with FSM as a baseline to extend the algorithm’s capability under the cooperative on-ramp merge scenarios. On the tactical motion level, trajectory generation for general ADS maneuvers (i.e., lane following and lane changing) and platooning behavior regulation is proposed such that planned trajectories of other relevant vehicles can be fully considered (i.e., intent sharing of predictive nature). The performance is evaluated in both traffic and automated driving simulators, and the results indicate that the proposed comprehensive multi-lane platooning algorithm can efficiently and safely regulate C-ADS-equipped vehicle behavior and meet system goals.