摘要
针对RRT算法随机性大、收敛速度慢和偏差性的问题,采用双向随机树和多棵局部随机树的探索与合并。增加引力分量,使双向随机树朝着各自目标方向生长,减少了算法的随机性。基于障碍物周围均匀生成若干根节点,对根节点增加斥力分量,生成多棵局部随机树。快速寻找可通行的路径,减少扩展过程中对障碍物的检测时间,加快算法的收敛速度,改善了算法的偏差性。用MATLAB进行虚拟仿真,验证了该算法的正确性。
Aimed at the problems of large randomness,slow convergence rate and deviation of RRT algorithm,the exploration and merging of bidirectional random tree and multiple local random tree are proposed,which increases the gravitational component and makes bidirectional random tree grow in the direction of the respective target,reducing the randomness. Several root nodes are evenly generated around obstacles,and the repulsion component is added to the root node to generate multiple local random trees. Through this method,the accessible path can be quickly searched,the detection time of obstacles in the expansion process can be reduced,the convergence speed of the algorithm can be accelerated,and the deviation of the algorithm can be improved. The improved algorithm is simulated by MATLAB software,which verifies its correctness.
作者
施杨洋
杨家富
布升强
朱林峰
SHI Yang-yang;YANG Jia-fu;BU Sheng-qiang;ZHU Lin-feng(College of Mechanical and Electronic Engineering,Nanjing Forestry University,Nanjing,Jiangsu 210037,China)
出处
《计算技术与自动化》
2019年第4期81-86,共6页
Computing Technology and Automation
基金
国家公益性行业科研专项重大项目资助(201404402-03)
南京市科技创新项目资助(2015CG047)
关键词
智能车辆
快速搜索随机树
路径规划
障碍物斥力函数
intelligent vehicle
RRT(Rapidly-Exploring Random Tree)
path planning
obstacle repulsion function
