基于组合代价函数的智能汽车轨迹规划
AUTONOMOUS VEHICLE TRAJECTORY PLANNING BASED ON COMBINED COST FUNCTION
朱凌云 1周宇1
作者信息
- 1. 重庆理工大学计算机科学与工程学院 重庆 400000
- 折叠
摘要
针对智能汽车在多车道公路场景行驶中的安全性、舒适性问题,提出一种基于组合代价函数的轨迹规划方法.利用Frenet坐标系分别求解车辆的横向和纵向运动轨迹,通过五次多项式拟合生成候选轨迹集;以危险程度和轨迹偏离程度为基准,以车辆在多车道公路行驶中的安全性、舒适性为目标,设计一种新的代价函数组合,通过求解代价函数从而选择最优轨迹.仿真结果表明:该方法能够使车辆在三车道动态公路场景中实现跟车与变道.
Abstract
Aiming at the safety and comfort of smart cars in multi-lane highway scenes,a trajectory planning method based on combined cost function is proposed.The Frenet coordinate system is used to solve the vehicle's lateral and longitudinal motion trajectories,and a set of candidate trajectories are generated through fifth-order polynomial fitting.Taking the degree of danger and the degree of trajectory deviation as the benchmark,and taking the safety and comfort of the vehicle on the multi-lane highway as the goal,a new cost function combination is designed,and the optimal trajectory is selected by solving the cost function.Simulation results show that the proposed new trajectory planning method enables vehicles to follow and change lanes in a three-lane dynamic highway scene.
关键词
自动驾驶/动态公路场景/轨迹规划/轨迹选择/代价函数Key words
Autonomous driving/Dynamic highway scene/Trajectory planning/Trajectory selection/Cost function引用本文复制引用
基金项目
重庆市巴南区科技计划项目(2018TJ02)
中国汽车工程研究院资助项目(2019Q109)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据