激光雷达动态障碍物检测

Dynamic Obstacles Detection by Laser Radars

为了解决在未知环境情况下障碍物运动状态的检测问题,提出一种基于激光雷达自主动态障碍物的检测方法.通过近邻域法和规则分类器算法对激光雷达检测的数据进行聚类;在此基础上分析了聚类障碍物的特征参数,利用聚类障碍物数据的置信区间关联性分析确定了障碍物的类型,对同一类型障碍物进行相对坐标转化分析确定了障碍物的速度、航向.实验结果表明;该方法能够在25ms内实现对车辆通行区域内的障碍物进行有效的聚类,从而实现对同一障碍物速度和航向的检测.

In order to detect the motion state of an obstacle in an unknown environment,a method for detecting dynamic obstacles by using laser radar is put forward.The close neighborhood method and the rules classifier algorithm are used to cluster the data obtained by the laser radar.And then the characteristic parameters of the clustering obstacles are analyzed.The types of obstacles are determined by analyzing the confidence interval correlation of the clustering obstacle data,and the velocity and azimuth of the obstacles are determined by the relative coordinate transformation of the obstacles of the same type.The experimental results show that the method can effectively cluster the obstacles in the traffic region in 25 ms,so as to detect the velocity and azimuth of the same obstacle.