基于NSGA-Ⅱ的智能车辆换道轨迹规划与优化

Lane-Change Trajectory Planning and Optimization for Intelligent Vehicles Based on NSGA-Ⅱ

针对智能车辆三次B样条曲线换道轨迹规划算法中控制点位置难以确定的问题,提出一种基于NSGA-Ⅱ的换道轨迹优化方法。采用三次B样条曲线规划了智能车辆换道轨迹,在低、中、高车速工况下,以换道轨迹长度及轨迹平均曲率为优化目标,采用NSGA-Ⅱ多目标优化算法对三次B样条曲线换道轨迹的控制点位置进行优化。为了验证优化后轨迹的可行性,进行仿真与实车验证。结果表明,在3种不同车速工况下,优化后换道轨迹在平均曲率、轨迹长度均有所减小,纵向位移与平均曲率分别减少了12.5%和12.0%、12.5%和40.0%、8.3%和15.4%;在联合仿真场景中,在10 m/s、20 m/s的低速、中速工况下,优化后轨迹跟踪最大横向误差小于0.1 m,在30 m/s的高速工况下,优化后轨迹跟踪最大横向误差不超过0.3 m;在实车验证中,优化前轨迹跟踪最大横向误差接近0.5 m,优化后轨迹跟踪最大横向误差不超过0.4 m,较优化前横向误差降低了20%以上。

To address the challenge of determining control point locations in the cubic B-spline curve algorithm for intelligent vehicle lane-change trajectory planning,an optimization method based on NSGA-Ⅱ was proposed.Lane-changing trajectories for intelligent vehicles were planned using cubic B-spline curves.Under low,medium,and high-speed conditions,the NSGA-Ⅱ multi-objective optimization algorithm was applied to optimize the control point positions of these trajectories.The optimization focused on two key objectives,i.e.minimizing the length of lane-changing trajectories and reducing the average curvature of the trajectories.To verify the feasibility of the optimized trajectory,both simulations and real-vehicle tests were conducted.The results show that the mean curvature and trajectory length are reduced after optimization under three different speed conditions.Specifically,the longitudinal displacement and mean curvature are reduced by 12.5% and 12%,12.5% and 40%,8.3%and 15.4%for low,medium and high speeds,respectively.In the co-simulation scenario,the optimized trajectory tracking shows a maximum lateral error of less than 0.1 m under low and medium speeds of 10 m/s and 20 m/s,respectively.At high speed of 30 m/s,the maximum lateral error remains below 0.3 m.In the real vehicle tests,the maximum lateral error before optimization is approximately 0.5 m.After optimization,this error is reduced to under 0.4 m,reflecting an improvement of over 20%.