摘 要：针对自动驾驶客车在主动避障过程中的路径规划与轨迹跟踪问题，本文提出一种基于纵横向解耦控制的优化算法。首先，建立车辆二自由度动力学模型，通过坐标变换将大地坐标系映射至Frenet坐标系，进而构建车辆路径跟踪误差模型。在此基础上，横向控制器基于线性二次型调节器(LQR)理论并结合前馈补偿进行设计；纵向控制器则采用改进的无模型自适应控制(MFAC)方法实现。此外，利用五次多项式进行平滑路径规划，生成满足运动边界条件的可行轨迹。实车试验结果表明，本文算法在路径跟踪精度、横向稳定性及速度控制平滑性方面均优于传统PID与模型预测控制(MPC)方法，且具有良好的实时性与鲁棒性。
关键词：路径规划；误差模型；LQR；前馈；MFAC；五次多项式
中图分类号：U471.15 文献标志码：A DOI：10.15917/j.cnki.1006-3331.2026.01.002
Research on Active Obstacle Avoidance Control Strategies for Autonomous Buses
WU Dexi,ZHU Yanpeng,LIANG Qingwei,ZHANG Gang,MING Qinghe
Abstract: To address the path planning and trajectory tracking problem for autonomous buses during active obstacle avoidance,this paper proposes an optimized algorithm based on decoupled longitudinal and lateral control.First,the algorithm establishes a two-degree-of-freedom vehicle dynamics model and transforms the global coordinate system into the Frenet coordinate system,thereby constructing a path tracking error model.On this basis,the lateral controller employs linear quadratic regulator(LQR)theory combined with feedforward compensation,while the longitudinal controller adopts an improved model-free adaptive control(MFAC)method.Furthermore,it utilizes a quintic polynomial for smooth path planning to generate feasible trajectories that satisfy motion boundary conditions.The real vehicle test results demonstrate that the pro-posed algorithm outperforms traditional PID and model predictive control(MPC)methods in path tracking accuracy,lateral stability,and speed control smoothness,while also exhibiting good real-time performance and robustness.
Key words: path planning; error model; LQR; feedforward; MFAC; quintic polynomial
