基于三维激光雷达的障碍物检测与跟踪

Obstacle detection and track based on 3D lidar

针对无人车对障碍物运动状态的检测问题,文章提出基于三维激光雷达的动态障碍物检测方法。通过使用PCL库中的欧式聚类法对激光雷达数据进行处理,在此基础上使用匈牙利算法对障碍物进行匹配,并对匹配上的数据运用扩展卡尔曼滤波方法进行融合,实现对障碍物的实时检测。实验结果表明,文章提出的算法能够对密集障碍物和较小的障碍物实现实时准确的跟踪。

For the problem object detector in autonomous vehicle system, this paper presents a dynamic obstacle detection method based on 3 D lidar. For real-time detection of obstacles, the Euclidean clustering method in the PCL library is used to process the lidar data, and the Hungarian algorithm is used to match the obstacles, and the extended kalman filter method is used to fuse the match obstacles. The experimental results show that the algorithm can track the dense and small obstacles accurately and real time.