Map Construction And Obstacle Detection Based On New Lidar

Lidar is wildly applied to intelligent vehicle because of its high precision and wide range detection.Establishing high-precision 3D map and detecting obstacle are the main directions of lidar in practical application.Based on the characteristics of the new lidar and the Ros data package,this paper will establish a high-precision 3D map and obstacle detection method in campus environment.The main work contents are as follows:The lidar is installed and calibrated correctly on the vehicle with a driver;On the establishment of high-precision map of campus environment part,some methods of eliminating bad points and extracting feature points are proposed based on the characteristics of the lidar;K-D tree is used for searching point quickly in point cloud registration,which makes the algorithm determine the edge and surface speed in the point cloud faster;Linear interpolation and segment processing are used to eliminate the fuzzy in motion compensation.By comparing the real map with 3D map,the advantages and disadvantages of the algorithm are analyzed.In the obstacle detection part,this paper introduces the definition of obstacles and the main detection objects.Several methods of ground segmentation are introduced and the segmentation method based on height is used to segment and remove the ground point cloud.The method achieves a good segmentation result;Introducing several common point cloud clustering methods,and Euclidean distance based clustering method is used to cluster obstacles with the help of K-D tree.And the constraints such as cluster size and reflection intensity are added in order to improve the clustering accuracy of this paper.PCA principal component analysis is used to analyze the cluster of point cloud to establish the barrier bounding box.Finally,the accuracy of obstacle identification is verified and the detection effect of the algorithm is tested.