Motion Estimation And 3D Reconstruction Based On LIDAR For Space Non-cooperative Target

It is important to estimate the motion attitude and three-dimensional structure of no-cooperative objects in the field orbit service and space-based attack and defense in space.Since there is no information interaction between the non-cooperative target and the tracker,it is always a difficult problem to measure the motion characteristics and three-dimensional structure of the non-cooperative target.As a non-contact active detection technology,Lidar can obtain three-dimensional information of space target by launching laser beam.With the rapid development of Lidar technology,laser point cloud imaging technology has been applied in space on orbit service.The main work of this paper is as follows:(1)Firstly,the working process of Lidar is simulated.This paper analyzes the principle of Lidar ranging,establishes the transformation relationship between different coordinate systems,gives the scanning mode and main geometric parameters of Lidar,analyzes the formation of Lidar motion distortion,and constructs the Lidar scanning simulation system.The model of the non-cooperative target satellite is gridded,which can simulate the Lidar flying around the non-cooperative target in space and verify the simulation effect of the Lidar scanning simulation system.(2)Secondly,the research on the pose estimation of space non-cooperative target based on Lidar is carried out.Aiming at the problem that the traditional ICP algorithm is sensitive to the initial value,a point cloud matching algorithm based on ISS feature points and RANSAC is proposed,which provides a good initial value for the ICP algorithm.This method extracts the edge points of the point cloud by calculating the ISS characteristics of the points,and then estimates the motion characteristics between the point clouds by RANSAC method.(3)Finally,to reconstruction non-cooperative target in space,an algorithm of Lidar loop detection based on ICP and main direction distribution of point cloud is presented.When the relative rotation angle of Lidar is close to one cycle,the point cloud of the current frame and the loop frame is matched,and the angle between the main direction of two groups of point cloud after matching is used to determine whether the Lidar loop is completed,if the loop is completed According to the graph optimization algorithm,the pose transformation between the looping frames is optimized to realize the 3D reconstruction of non-cooperative objects.