This paper proposes a framework for path tracking under additive disturbance when a vehicle travels at high speed or on low-friction road. A decoupling control strategy is adopted, which is made up of robust model predictive control and the stability control combining preview G-vectoring control and direct yaw moment control. A vehicle-road model is adopted for robust model predictive control, and a robust positively invariant set calculated online ensures state constraints in the presence of disturbances. Preview G-vectoring control in stability control generates deceleration and acceleration based on lateral jerk, later acceleration, and curvature at preview point when a vehicle travels through a cornering. Direct yaw moment control with additional activating conditions provides an external yaw moment to stabilize lateral motion and enhances tracking performance. A comparative analysis of stability performance of stability control is presented in simulations, and furthermore, many disturbances are considered, such as varying wind, road friction, and bounded state disturbances from motion planning and decision making. Simulation results show that the stability control combining preview G-vectoring control and direct yaw moment control with additional activating conditions not only guarantees lateral stability but also improves tracking performance, and robust model predictive control endows the overall control system with robustness.

中文翻译：

---

综合鲁棒模型预测控制和稳定性控制的自主车辆路径跟踪框架

本文提出了一种车辆在高速或低摩擦道路上行驶时在加性扰动下进行路径跟踪的框架。采用解耦控制策略，由鲁棒模型预测控制和结合预览G矢量控制和直接偏航力矩控制的稳定性控制组成。采用车路模型进行鲁棒模型预测控制，在线计算的鲁棒正不变集确保存在干扰时的状态约束。当车辆通过转弯时，稳定性控制中的预览 G 向量控制基于横向加加速度、后加速度和预览点的曲率生成减速度和加速度。带有附加激活条件的直接偏航力矩控制提供外部偏航力矩以稳定横向运动并增强跟踪性能。在仿真中对稳定性控制的稳定性性能进行了比较分析，此外，还考虑了许多干扰，例如变化的风、道路摩擦以及运动规划和决策中的有界状态干扰。仿真结果表明，预G矢量控制和直接横摆力矩控制相结合的稳定性控制，不仅保证了横向稳定性，而且提高了跟踪性能，鲁棒模型预测控制赋予了整个控制系统的鲁棒性。此外，还考虑了许多干扰，例如变化的风、道路摩擦以及来自运动规划和决策的有界状态干扰。仿真结果表明，预G矢量控制和直接横摆力矩控制相结合的稳定性控制，不仅保证了横向稳定性，而且提高了跟踪性能，鲁棒模型预测控制赋予了整个控制系统的鲁棒性。此外，还考虑了许多干扰，例如变化的风、道路摩擦以及来自运动规划和决策的有界状态干扰。仿真结果表明，预G矢量控制和直接横摆力矩控制相结合的稳定性控制，不仅保证了横向稳定性，而且提高了跟踪性能，鲁棒模型预测控制赋予了整个控制系统的鲁棒性。