汽车工程 ›› 2026, Vol. 48 ›› Issue (3): 553-565.doi: 10.19562/j.chinasae.qcgc.2026.03.006

• • 上一篇    

基于iLQR的智能车辆换道轨迹规划算法研究

刘永涛1,亢浩宇1,纳林奇1,李智鹏1,朱屹晨2,陈轶嵩1()   

  1. 1.长安大学汽车学院,西安 710064
    2.长安大学都柏林国际交通学院,西安 710064
  • 收稿日期:2025-05-20 修回日期:2025-07-11 出版日期:2026-03-25 发布日期:2026-03-19
  • 通讯作者: 陈轶嵩 E-mail:chenyisong_1988@163.com
  • 基金资助:
    陕西省“两链”融合重点专项揭榜挂帅项目(2023JBGS-13);咸阳市重大科技创新专项项目(L2025-ZDKJ-ZDGG-RGZN-003);中央高校基本科研业务项目(300102223204)

Research on Lane-Change Trajectory Planning Algorithm for Intelligent Connected Vehicles Based on iLQR

Yongtao Liu1,Haoyu Kang1,Linqi Na1,Zhipeng Li1,Yichen Zhu2,Yisong Chen1()   

  1. 1.School of Automobile,Chang’an University,Xi’an 710064
    2.Dublin Intemational College of Transportation,Chang'an University,Xi’an 710064
  • Received:2025-05-20 Revised:2025-07-11 Online:2026-03-25 Published:2026-03-19
  • Contact: Yisong Chen E-mail:chenyisong_1988@163.com

摘要:

针对智能车辆换道过程中高效性、安全性和舒适性的协调兼顾问题,本文提出一种基于迭代线性二次调节器(iLQR)算法的换道轨迹规划方法。首先,在横向路径规划方面,于Frenet坐标系下采用改进的五次多项式生成初步横向轨迹;在纵向速度规划方面,利用引入启发信息的动态规划(DP)方法,快速生成满足车辆运动学约束的速度规划序列。其次,通过iLQR算法对初始轨迹进行二次优化,将碰撞风险、舒适性及控制约束纳入优化目标,得到高效、安全且平滑的最优换道轨迹。最后,通过CarSim、Matlab/Simulink及Prescan进行联合仿真验证。仿真结果表明,该方法较传统DP算法换道效率提升约20%,纵向加速度更平稳,显著提升了换道的安全性与舒适性,可为自动驾驶车辆在复杂交通环境下的换道决策提供高效、可靠的技术支撑。

关键词: 智能车辆, 迭代线性二次调节器, 改进五次多项式, 动态规划算法, 换道轨迹规划

Abstract:

For balancing efficiency, safety and comfort during lane-changing of intelligent vehicles, this paper proposes a lane-changing trajectory planning method based on the iterative linear quadratic regulator (iLQR) algorithm. Firstly, in the lateral path planning, an improved quintic polynomial is used in the Frenet coordinate system to generate an initial lateral trajectory. In the longitudinal speed planning, a dynamic programming (DP) method with heuristic information is utilized to quickly generate a speed planning sequence that meets the kinematic constraints of the vehicle. Secondly, the initial trajectory is further optimized by the iLQR algorithm, incorporating collision risk, comfort and control constraints into the optimization objective, to obtain an efficient, safe and smooth optimal lane-changing trajectory. Finally, the method is verified through a joint simulation using CarSim, Matlab/Simulink and Prescan. The simulation results show that this method improves the lane-changing efficiency by approximately 20% compared to the traditional DP algorithm, with a more stable longitudinal acceleration, and significant improvement in the safety and comfort of lane-changing, which can provide efficient and reliable technical support for lane-changing decision-making of autonomous vehicles in complex traffic environment.

Key words: intelligent vehicles, iLQR, improved fifth degree polynomials, DP algorithm, lane change trajectory planning