摘要
快速随机搜索树(RRT)算法是解决运动规划问题的典型算法。针对传统RRT算法采样效率低、不满足车辆运动学约束的缺陷,提出一种基于先验概率分布采样改进的RRT车辆运动规划算法。先利用环境信息生成初始解并根据其生成RRT采样的先验概率,再使用控制信号采样和航迹推演的方式进行搜索树扩展,最后用样条曲线连接目标。经仿真实验验证,提出的改进RRT算法能够高效规划出具有运动学可行性的轨迹。
Rapidly Random-exploring Trees(RRT) algorithm is a typical algorithm to solve motion planning problems. But traditional RRT algorithm have disadvantages of low sampling efficiency and not meeting the vehicle kinematics constraints.Considering these problems, an improved RRT motion planning algorithm based on prior probability is proposed in this paper.Firstly, the initial solution is calculated by using the map information, and the prior probability of RRT sampling is generated according to it. Then, the searching tree is expanded by using the methods of control signal sampling and dead reckoning. Finally,the target is connected by spline curve. The simulation results show that the improved RRT algorithm proposed in this paper can efficiently calculate the trajectory meeting the vehicle kinematics constraints.
作者
王圣懿
黄劲松
WANG Shengyi;HUANG Jinsong(School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China)
出处
《导航定位学报》
CSCD
2022年第2期85-92,共8页
Journal of Navigation and Positioning
关键词
自动驾驶
运动规划
快速随机探索树
先验概率
航迹推演
autonomous driving
motion planning
rapidly random-exploring trees
prior probability
dead reckoning