基于高斯采样的改进RRT*路径规划算法

An Improved RRT*Path Planning Algorithm Based on Gaussian Sampling

针对渐进最优快速扩展随机树(RRT*)算法随机性大,搜索效率低的问题,提出一种结合目标偏置概率策略与缩小相对状态空间的改进算法(G-RRT*)。该算法在保持RRT*原有的优势基础上以一定概率使采样点偏置为目标点,提高路径规划的导向性,并结合高斯分布缩减相对状态空间,限制随机树的扩展方向,从而加快路径规划的速度,保证以较短时间找到渐进最优路径。在MATLAB平台上分别对二维平面路径规划和三维机械臂路径规划测试。结果表明:G-RRT*算法有效减少了路径距离开销和计算时间,具有一定的应用价值。

Aiming at the problem of the asymptotically optimal rapid expansion random tree(RRT*)algorithm has large randomness and low search efficiency,an improved algorithm(G-RRT*)combining the target bias probability strategy and the relative state space reduction(G-RRT*)is proposed.The algorithm keeps the original advantages of RRT*and uses a certain probability to bias the sampling point as the target point,which improves the guidance of path planning,and combines the Gaussian distribution to reduce the relative state space,restricts the expansion direction of the random tree,and speeds up the path The speed of planning ensures that the gradual optimal path can be found in a short time.The two-dimensional planar path planning and the three-dimensional robotic arm path planning were tested on the MATLAB platform.The results show that the G-RRT*algorithm effectively reduces the path distance cost and calculation time,and has certain application value.