摘要
低矮道边检测是无人驾驶汽车环境感知的重要课题,其感知的效果直接影响着自动驾驶的性能。针对城市低矮道边检测问题提出了一种新的基于双多线激光雷达的道边检测与提取方法。方法首先通过梯度一致性特征将原始雷达点云分成路面点与障碍物点两类;其次,利用栅格地图对障碍物点进行过滤、聚类处理;然后利用路面点与栅格地图提取候选路边点;最后使用改进的RANSAC算法实现道边的提取。方法已成功运用在无人车上,实验表明,方法具有良好的检测效果。
Curb detection is an important issue in the environment perception of unmanned vehicle,the effects of perceptiondirectly affect the performance of automatic driving. A new curb detection algorithm based on double multi-beam LiDAR is pro-posed. By using the gradient consistency characteristics divide LiDAR points into surface points and obstacle points,then grid mapis applied to cluster and filter obstacle points,curb points is extracted with the grid map and surface points. Finally,the curb is ex-tracted by using improved RANSAC algorithm. The method has been successfully used in our unmanned ground vehicle,the applica-tion shows the good detection.
出处
《计算机与数字工程》
2017年第12期2368-2372,共5页
Computer & Digital Engineering
基金
国家自然基金青年项目(编号:61305134)
博士点基金项目(编号:20133219120035)资助
关键词
环境感知
激光雷达
点云分割
道边检测
environment perception
LiDAR
point segment
curb detection
