高斯混合模型优化的Informed-RRT?路径规划算法
Path planning algorithm of Informed-RRT? optimized by Gaussian mixture model
韩龙 1姜楠 1邓东江 1陈楚1
作者信息
- 1. 黑龙江科技大学 电气与控制工程学院,哈尔滨 150022
- 折叠
摘要
针对巡检机器人应用Informed-RRT∗算法路径规划时的无效采样、速度较慢以及路径不平滑等问题,提出了高斯混合模型优化的Informed-RRT∗路径规划算法.运用高斯分布函数获取障碍物附近无碰撞采样节点的样本集,训练生成的高斯混合模型将采样样本集中在更有效的区域中,增强采样目的性与准确性.采用三次B样条曲线对路径进行平滑处理,在不同二维栅格地图中进行仿真实验.结果 表明,改进算法与Informed-RRT∗算法相比,找到最优路径花费时间最高缩短了57.17%,需寻找的采样点最多减少57.86%,路径长度及生长转角均有较大改进,路径更平滑.搭建巡检机器人进行现场测试,改进算法能够满足巡检机器人路径规划的要求,证明该方法的有效性.
Abstract
This paper is aimed at addressing the invalid sampling,slow speed,and unsmooth paths for inspection robots using the Informed-RRT∗ algorithm on path planning and proposes an Informed-RRT∗path planning algorithm optimized with a Gaussian mixture model.The study includes obtaining a sample set of collision-free sampling nodes near obstacles by using Gaussian distribution function;con-centrating the sampling samples trained by Gaussian mixture model within more effective areas;enhancing the purpose and accuracy of the sampling;and smoothing the path by using cubic B-spline curves to con-duct the simulations test in different 2D grid maps.The results show that by comparison with the In-formed-RRT∗ algorithm,the improved algorithm reduces the time on searching the optimal path up to 57.17%,and decreases the number of sampling points up to 57.86%with greater improvement in path length,growth angle and smooth path.Live testing on the inspection robot demonstrates that the improved algorithm meets the path planning requirements,proving the effectiveness of this method.
关键词
巡检机器人/路径规划/Informed-RRT∗/路径优化Key words
inspection robot/path planning/Informed-RRT∗/path optimization引用本文复制引用
基金项目
黑龙省重点研发计划项目(JD2023SJ25)
黑龙江省省属高等学校基本科研业务费项目(2022-KYYWF-0553)
出版年
2024
国家科技期刊平台
NSTL
万方数据