摘要
针对智能车参考路径平滑中起点曲率难以约束和未考虑曲率变化率的问题,提出一种基于多目标优化的路径处理方法。该方法首先对二维平面内离散点的曲率计算方式进行分析,通过离散点的几何关系构建二次项的优化目标函数,间接表示曲率连续性指标;其次通过新增离散点的方式,将非线性的起点曲率约束项转化为软约束;最后将带约束的多目标优化函数转化为二次规划型快速求解。仿真和实车试验结果表明:该方法能够适应不同大小的起点曲率约束,有效平滑从高精度地图中提取的离散路径点集,为智能车提供曲率连续的参考路径,优化方法平均耗时为6.82 ms,满足运动规划层的实时性要求。
To solve the problem that the starting point curvature is difficult to constrain and the curvature change rate is not considered in the reference path smoothing of intelligent vehicle,a path processing method based on multi-objective optimization is proposed in this paper.Firstly,the curvature calculation method of discrete points in two-dimensional plane is analyzed,and the optimization objective function of quadratic term is constructed through the geometric relationship of discrete points,which indirectly represents the curvature continuity index.Secondly,by adding discrete points,the nonlinear starting point curvature constraint is transformed into soft constraint.Finally,the multi-objective optimization function with constraints is transformed into quadratic programming for fast solution.The simulation and real vehicle experimental results show that this method can adapt to the starting point curvature constraints of different sizes,effectively smooth the discrete path point set extracted from the high-definition map,and provide a curvature continuous reference path for the intelligent vehicle.The average time of the optimization method is 6.82 ms,which meets the real-time requirements of the motion planning layer.
作者
齐尧
何滨兵
章永进
徐友春
QI Yao;HE Binbing;ZHANG Yongjin;XU Youchun(Army Military Transportation University,Tianjin 300161,China)
出处
《军事交通学报》
2022年第7期41-46,共6页
Journal of Military Transportation University
基金
军队重点学科专业建设项目
关键词
智能车
参考路径
多目标优化
二次规划
intelligent vehicle
reference path
multi-objective optimization
quadratic programming
作者简介
齐尧(1994-),男,博士研究生.
