【作者】 李博文；

【导师】 杨洪玖； 李洪波；

【作者基本信息】 天津大学 ， 电子信息， 2024， 硕士

【摘要】 无人驾驶汽车是未来交通领域的重要发展方向之一。其中,弯道行驶是无人驾驶技术的基础场景之一,多种复杂场景都需求无人驾驶汽车具备稳定精准的弯道行驶功能。在无人驾驶技术中,运动规划和运动控制是关键环节。运动规划通过算法和模型确定车辆的行驶路径和动作,为运动控制提供指导。安全可行的运动规划是确保车辆行驶安全的基础,而可靠的运动控制则是车辆安全行驶的前提。在运动控制与运动规划中,确保无人驾驶汽车在弯道上平稳安全的行驶是非常重要的。本文的主要研究工作如下:首先,针对无人驾驶汽车的弯道轨迹跟踪提出了一种基于区间预测器的鲁棒模型预测控制(Model Predictive Control,MPC)算法。无人驾驶车辆轨迹跟踪系统被建模成具有外部干扰和不确定参数的线性变参数(Linear Parameter Varying,LPV)系统。通过设计区间预测器来构造可变鲁棒集,可以使系统满足针对外部干扰和模型不确定参数的鲁棒状态约束与控制输入约束。通过设计鲁棒控制器,提高了无人驾驶汽车轨迹跟踪系统在外部干扰和模型不确定参数影响下的抗干扰能力和动态性能。证明了闭环系统的输入到状态稳定性以及鲁棒MPC算法的迭代可行性。并通过仿真和实验结果证明了基于区间预测器的鲁棒MPC算法在无人驾驶车辆弯道轨迹跟踪中的有效性。然后,针对在具有道路超高的大曲率弯道上行驶的无人驾驶汽车,提出了一种速度规划和无终端约束MPC速度跟踪控制算法。基于车辆与乘客在大曲率弯道上的受力平衡原理,设计了一种针对乘客舒适度提高的纵向速度规划方法。提出了带有跟踪微分器的无终端约束MPC算法,用于准确且稳定地跟踪规划出的纵向参考速度。证明了无终端约束MPC算法的递归可行性和闭环系统的渐近稳定性。通过实验结果验证了所提出的无人驾驶汽车速度规划和无终端约束MPC速度跟踪算法的有效性。最后,针对公路掉头场景,设计了一种基于非线性MPC的快速公路掉头规划与控制方法。针对方向盘转向滞后问题,通过系统辨识方法建立无人驾驶汽车的二阶转向模型。利用五次多项式规划和基于非线性MPC的数值优化算法规划出快速掉头轨迹。设计了非线性MPC算法来跟踪所得到的掉头轨迹,并补偿了方向盘的转向滞后特性。分析了非线性MPC算法的递归可行性和闭环系统的渐近稳定性。实验结果表明,所设计的无人驾驶汽车快速公路掉头方法具有可行性和有效性。更多还原

【Abstract】 Autonomous vehicles are one of the important directions for future development in the field of transportation.Among them,curve driving is one of the fundamental scenarios for autonomous driving technology,and various complex scenarios require autonomous vehicles to have stable and precise curve driving capabilities.The main research work of this thesis includes:Firstly,a robust Model Predictive Control(MPC)algorithm based on an interval predictor is proposed for the bend trajectory tracking of autonomous vehicles.The driverless vehicle trajectory tracking system is modeled as a linear variable parame-ter(LPV)system with external interference and uncertain parameters.By designing an interval predictor to construct a variable robust set,the system can satisfy robust state constraints and control input constraints for external interference and model un-certainty.By designing a robust controller,the anti-interference ability and dynamic performance of the driverless car trajectory tracking system under the influence of ex-ternal interference and model uncertainty are improved.The stability of input-to-state and the iterative feasibility of the robust MPC algorithm are proven.The effectiveness of the robust MPC algorithm based on an interval predictor in driverless vehicle bend trajectory tracking is demonstrated through simulation and experimental results.Secondly,a speed planning and terminal-free MPC speed tracking control algorith-m is proposed for autonomous vehicles driving on highways with high superelevation and large curvature bends.Based on the force balance principle of vehicles and pas-sengers on high-curvature bends,a longitudinal speed planning method aimed at im-proving passenger comfort is designed.A terminal-free MPC algorithm with a tracking differentiator is proposed for accurately and stably tracking the planned longitudinal reference speed.The recursive feasibility of the terminal-free MPC algorithm and the asymptotic stability of the closed-loop system are proven.The effectiveness of the pro-posed driverless car speed planning and terminal-free MPC speed tracking algorithm is verified through experimental results.Finally,a fast highway U-turn planning and control method based on nonlinear MPC is designed for U-turn scenarios.To address the problem of steering lag in au-tonomous vehicles,a second-order steering model is established through system iden-tification.A quintic polynomial planning and a nonlinear MPC-based numerical opti-mization algorithm are used to plan a fast U-turn trajectory.A nonlinear MPC algorithm is designed to track the obtained U-turn trajectory and compensate for the steering lag characteristics of the steering wheel.The recursive feasibility of the nonlinear MPC algorithm and the asymptotic stability of the closed-loop system are analyzed.Exper-imental results show that the designed driverless car fast highway U-turn method is feasible and effective.更多还原