一种激光三维点云动态障碍剔除算法框架

A framework on dynamic obstacle removal algorithm of laser 3D point cloud

基于激光点云地图的动态障碍剔除是同步定位与建图(simultaneous localization and mapping,SLAM)研究领域的难题之一。动态障碍的移动轨迹不仅会遮挡真实的静态环境信息,也会对移动机器人的定位和路径规划造成阻碍。针对激光点云地图中的动态障碍识别问题,该文提出一种基于卡尔曼滤波的运动障碍跟踪方法。首先,对原始点云预处理,使用欧式聚类算法,实现离散点云聚类。其次设计了基于卡尔曼滤波的运动障碍状态预估方程,并预测出动态障碍点云目标在下一时刻的位置。然后,使用匈牙利算法将预测位置与下一时刻真实位置进行匹配,实现对每一时刻动态障碍体的识别。最后,剔除后进行点云配准建图。在室内外环境下对提出的动态障碍剔除算法进行验证,并将剔除动态障碍后的点云地图可视化输出。实验结果表明该算法在室内外环境下对激光点云地图中动态障碍均能较好地识别与剔除。

Dynamic obstacle removal based on laser point cloud map is one of the difficult problems in the field of simultaneous localization and mapping(SLAM).The moving trajectory of the dynamic obstacle will not only block the real static environment information,but also hinder the positioning and path planning of the mobile robot.This paper proposes a new method for dynamic obstacle recognition and removal based on Kalman filter in laser point cloud map.Firstly,the original laser point cloud map is preprocessed by segmentation and clustering,and the objects in the point cloud map are separated into point cloud targets.Secondly,the Kalman filter algorithm is used to analyze the historical state of the point cloud target,and the point cloud map is identified and predicted.The position of the dynamic point cloud target is then matched by the Hungarian algorithm to the predicted position and the newly observed position.In this way,the recognition and storage of dynamic obstacle objects in the point cloud map at each moment is realized.Finally,the point cloud after screening out the dynamic target is registered and constructed.In this paper,an experimental platform is built to verify the effect of the dynamic obstacle removal algorithm indoors and outdoors,and the laser point cloud after removing the dynamic obstacle is matched,and the point cloud map is constructed and visualized.Experiments show that the algorithm can better identify and eliminate the dynamic obstacles on the ground in the laser point cloud map both indoors and outdoors.