马转转,陈丽,梁天.基于激光雷达的移动小车实时避障算法[J].高技术通讯(中文),2025,35(3):273~283 |
基于激光雷达的移动小车实时避障算法 |
LiDAR based real-time obstacle avoidance for mobile vehicles |
|
DOI:10. 3772 / j. issn. 1002-0470. 2025. 03. 005 |
中文关键词: 移动小车; 激光雷达; 实时避障; 速度矢量合成 |
英文关键词: mobile vehicle, light detection and ranging (LiDAR), real-time obstacle avoidance, composition of velocity vectors |
基金项目: |
作者 | 单位 | 马转转 | (上海工程技术大学航空运输学院上海 201620) | 陈丽 | | 梁天 | |
|
摘要点击次数: 143 |
全文下载次数: 98 |
中文摘要: |
本文提出了一种适用于移动小车在未知环境下实时避障的方法。首先,采用欧式聚类方法将2D激光雷达点云聚类为不同的簇,通过主成分分析法将每一簇障碍物描述为有向闭环矩形;其次,建立了方位K维(K-dimensional,KD)树,实现障碍物顶点存储和当前避障所需考虑的障碍物顶点的快速查找;进一步,通过几何法获得线性凸的避障速度区域,在该区间设计具有最小改变量的避障速度,并与目标速度矢量合成得到小车的指令跟踪速度。实验结果表明,该算法能够快速实现障碍物建模和所需障碍物点查找;基于几何规划获得的指令跟踪速度有明确的物理意义,在给定场景下规划的路径具有可重复性;同时该算法对传感器的要求低、计算量小,适用于低配置移动小车在动态环境的避障运动控制。 |
英文摘要: |
A real-time obstacle avoidance of mobile vehicles in unknown environments is proposed. The 2D light detection and ranging (LiDAR) point cloud is clustered into different clusters using the Euclidean clustering method, and each obstacle cluster is described as a directed closed-loop rectangle using principal component analysis. A directional K-dimensional (KD) tree is established to store obstacle vertices and quickly search for obstacle vertices that need to be considered for current obstacle avoidance. A linear convex obstacle avoidance velocity region is obtained through geometric planning, and an obstacle avoidance velocity with the minimum change is designed in this region. The target velocity vector and the obstacle avoidance velocity vector are synthesized to obtained the command tracking velocity. The experimental results confirmed that this method can achieve fast obstacle modeling and required obstacle point search. The tracking speed obtained based on geometric method has a clear physical meaning, and the path planned in a given scenario has repeatability. At the same time, this algorithm has low requirements for sensors and small computational complexity, making it suitable for obstacle avoidance motion control of low configuration mobile vehicles in dynamic environments. |
查看全文
查看/发表评论 下载PDF阅读器 |
关闭 |