基于节点过滤及运动学约束改进的RRT算法

Improved intelligent vehicle path planning RRT algorithm based on node filtering and kinematic constraints

由于快速扩展随机树(RRT)算法存在规划路径曲折、收敛较慢,且无法被智能车辆直接跟踪等问题。为克服此缺陷,本文在基本RRT算法基础上,加入目标偏向性策略和密集节点过滤,以此提高规划速度;在选择父节点时考虑车辆运动学约束并根据车辆位置动态确定扩展步长,最后对所得路径进行修剪,并使用三次B样条曲线进行平滑,生成一条平滑可被追踪的曲线。仿真实验表明,改进的RRT算法生成路径更加合理、平滑,且符合车辆运动学特性。

Rapid Extended Random Tree(RRT)algorithm planning path twists and turns,convergence is slow,and can not be directly tracked by intelligent vehicles.To overcome this defect,an improved RRT algorithm is proposed in this paper.Based on the basic RRT algorithm,target bias strategy and dense node filtering are added to improve the planning speed.When selecting the parent node,the vehicle kinematics constraint is considered and the extended step size is determined dynamically according to the vehicle position.Finally,the path obtained is trimmed and smoothing by cubic B-spline curve to generate a smooth and tractable curve.The simulation results show that the path generated by the improved RRT algorithm is more reasonable and smooth,and conforms to the vehicle kinematics characteristics.