摘要
为了保证智能车辆在弯道动态场景下实现高效、稳定的自主换道并提高跟踪精度,提出了一种双层智能车辆换道策略,该策略包括混合轨迹规划和自适应预测时域轨迹跟踪控制。在上层,综合考虑车辆运动学约束、轨迹安全性与舒适性等要求,将路径与速度规划问题视为对偶合作博弈问题进行并行规划,结合综合博弈收益函数和二次规划生成最优换道轨迹。在下层,基于车辆动力学模型建立线性时变模型预测控制器(Linear Time-varying Model Predictive Control,LTV-MPC),增加了包括前轮转角约束、前轮转角增量约束、侧偏角约束和侧向加速度等动力学约束,并引入松弛因子以避免出现非可行解;分析车速与预测时域的关系,通过曲线拟合不同车速与对应最优预测时域的控制率,并基于车速与预测时域函数拟合关系设计了一种根据车速实时更新预测时域的自适应轨迹跟踪控制器,通过收敛性分析证明了所设计的轨迹跟踪控制器能够有效地跟踪规划轨迹。将轨迹规划求解的最优轨迹输入到自适应预测时域轨迹跟踪控制器,验证所提出的双层智能车辆换道策略在弯道动态场景下的可行性。仿真结果表明:所提出的双层换道策略能在弯道动态场景下合理快速地规划出换道轨迹,且能保证车辆稳定有效地跟踪最优轨迹,与传统的轨迹跟踪控制器相比,自适应预测时域轨迹跟踪控制器的轨迹跟踪控制精度和稳定性都有所提高,为处理智能车辆在弯道动态场景下自主换道轨迹规划与跟踪控制的问题提供了参考价值。
To ensure efficient and stable autonomous lane changing and improve tracking accuracy for intelligent vehicles in dynamic curve scenarios,a two-layer intelligent vehicle lane-change strategy was proposed.This strategy includes hybrid trajectory planning and adaptive predictive horizon trajectory tracking control.At the upper layer,considering vehicle kinematic constraints,trajectory safety,and comfort requirements comprehensively,the path and speed planning problem was planned as a dual cooperative game.An optimal lane-change trajectory was generated by combining a comprehensive game utility function with quadratic programming.At the lower level,a linear time-varying model prediction controller(LTV-MPC)was established based on the vehicle dynamics model,incorporating dynamic constraints such as front-wheel angle,front-wheel angle increment,slip angle,and lateral acceleration.A relaxation factor was introduced to ensure the feasibility of the solution.By analyzing the correlation between vehicle speed and prediction horizon,a curve-fitting method was used to derive optimal prediction horizons at varying speeds.An adaptive trajectory tracking controller was subsequently designed to dynamically adjust the prediction horizon according to real-time velocity.The optimal trajectory obtained from the trajectory planning layer was provided as input to the adaptive model predictive control trajectory tracking controller to verify the feasibility of the proposed two-layer lane-change strategy in dynamic curve scenarios.Simulation results indicate that the two-layer lane-change strategy proposed in this study can reasonably and rapidly plan lane-change trajectories and ensure stable and effective vehicle tracking of optimal trajectories.Compared to traditional trajectory tracking controllers,the adaptive predictive horizon trajectory tracking controller improves both accuracy and stability,providing a valuable reference for trajectory planning and tracking control of intelligent vehicle autonomous lane changes in dynamic curve scenarios.
作者
陈峥
姜圣男
沈世全
刘永刚
郭凤香
CHEN Zheng;JIANG Sheng-nan;SHEN Shi-quan;LIU Yong-gang;GUO Feng-xiang(Faculty of Transportation Engineering,Kunming University of Science and Technology,Kunming 650500,Yunnan,China;College of Mechanical Engineering,Chongqing University,Chongqing 400030,China)
出处
《中国公路学报》
北大核心
2025年第5期248-260,共13页
China Journal of Highway and Transport
基金
国家自然科学基金项目(52272395)。
关键词
汽车工程
智能车辆
轨迹跟踪控制
自主换道
二次规划
合作博弈
automotive engineering
intelligent vehicle
trajectory tracking control
active lane change
quadratic planning
cooperative game
