摘要
根据传统快速搜索随机树算法(rapidly random-exploring trees,简称RRT)搜索速度快、所需时间短,但随机性大以及约束不足等特点,建立了直道和弯道的期望路径模型,采用高斯分布描述随机采样点,并引入启发式搜索机制,改进RRT算法.与原算法仿真对比,结果表明:改进算法所规划的路径质量显著提高,规划时间缩短一倍.同时,在Prescan软件中搭建直道和弯道仿真场景,跟随规划路径,结果表明:改进后RRT算法所得路径具有很好的跟随效果,且侧向加速度在车辆稳定性要求范围内,说明采用改进后的RRT算法进行汽车局部路径规划可行实用.
The original Rapidly-exploring Random Trees(RRT) algorithm has a rapid exploring-speed and short required time in path planning though it has large randomness and lacks of constraints. Thus, an improved RRT was proposed where the expected paths were built in both straight and curved roads. The random points were accorded with normal distribution around the expected paths. Heuristic search method that led the random points to the goal with a certain probability was also used for improvement. Compared with the original RRT algorithm, it performs quite well in both time-efficient and path quality in the simulation. Meanwhile, the effectiveness of the improved RRT algorithm was verified in Prescan. The path can be followed well and the secure lateral acceleration was satisfied. In conclusion, the improved RRT is effective in the path planning for vehicle collision avoidance.
作者
宋晓琳
周南
黄正瑜
曹昊天
SONG Xiaolin ZHOU Nan HUANG Zhengyu CAO Haotian(Stake Key Laboratory of Advanced Design and Manufacturing for Vehicle Body, Hunan University,Changsha 410082,China)
出处
《湖南大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2017年第4期30-37,共8页
Journal of Hunan University:Natural Sciences
基金
国家自然科学基金资助项目(51175159
51575169)
National Natural Science Fundation of China(51175159
51575169)
关键词
快速搜索随机树
汽车局部路径规划
高斯分布
路径跟随
rapidly-exploring random trees
vehicle path planning
Gauss distribution
path following