This these is based on the three-dimensional environment reconstruction of two-dimensional lidar in the project of "research and development of standardized cattle manure cleaning device".The main content is to convert the two-dimensional point cloud collected by RPLIDAR A3M1 into three-dimensional data to realize the reconstruction,so as to provide the information of obstacles in front of the robot.In view of this design requirement,this these establishes a complete theoretical system of three-dimensional environment reconstruction,including data collection and preprocessing,multi-angle point cloud matching and three-dimensional reconstruction.This these mainly completed the following tasks:(1)Point cloud data collection and preprocessing.Since two-dimensional lidar is used in this design,an automatic rotating cradle head is designed to assist the lidar in collecting three-dimensional point cloud data.Euclidean algorithm is used to separate point clouds in different regions to extract the region of interest,Kd-tree algorithm of searching neighborhood is used to remove outlier noise points,and the minimum distance method isused to simplify the point cloud.(2)Multi-angle point cloud registration.In order to complete the object reconstruction,four lidar is used to scan the object from multiple angles in this design.Firstly,the feature parameters of point cloud surface including normal vector and curvature are calculated to match the data.Secondly,point cloud matching is divided into two steps:initial matching and Iterative Closest Point(ICP)matching,so as to match point clouds from multiple perspectives to the same coordinate system.(3)Three-dimensional reconstruction.After collection,preprocessing and matcging,the point cloud data has no topological relationship,nor information about faces and edges.Therefore,it is necessary to reconstruct the topological relationship between point clouds.The process of surface reconstruction is to form the final closed surface through three-dimensional point cloud data fitting.This these introduces two kinds of surface reconstruction algorithms,one is Delaunay triangulation algorithm and the other is Ball Pivoting Algorithm(BPA)algorithm.In the design,the BPA algorithm is used and gives the final effect chart.At last,how to apply the design to the project of "research and development of standardized cattle manure cleaning device" is described in detail.Two 2D Lidars and automatic rotating cradle head are carried on dashgo robot platform in the projects to reconstruct 3D environment,thereby gaining the obstacles in front of the robot information,including the size and shape of the barrier and the distance between the multiple obstacles,eventually lead to robots in apprehending cow dung in the process of obstacle avoidance effectively,and provide effective support for the project’s development. |