Pose Measurement Of The Target Based On Planar Array 3D Imaging LIDAR

As a key technology of three-dimensional shape recognition and trajectory tracking,relative pose measurement of moving target is widely used in space on-orbit service,autonomous driving,intelligent robot and other fields.Non-cooperative targets cannot provide cooperative information that is convenient for pose identification,so it is a great opportunity and challenge to use hardware imaging system to capture useful information and to design better robust and efficient pose measurement algorithm.In order to solve the problems of lack of priori information,short detection range and low accuracy of long-range target ranging,through the improvement and innovation of key technologies,the research on non-cooperative moving target pose measurement technology based on planar array Three-dimensional(3D)imaging lidar is carried out.Firstly,based on the polarization-modulated laser 3D imaging method,the detailed theoretical technical scheme of high-resolution planar array 3D imaging system is established.By analyzing the polarization modulation characteristics of electro-optical crystal,a method of error calibration based on polarization 3D imaging is proposed.The problems of low imaging accuracy and limited imaging field-of-view are solved.The high precision and fast 3D imaging of long distance moving target is realized,which provides 3D data information for non-cooperative target pose measurement.Secondly,the traditional segmentation method of point cloud has the problem of large amount of data and slow convergence speed.Based on the unique imaging mechanism and system structure of the new planar array Lidar,this paper proposes the solution of multi-dimensional data fusion to separate the point cloud of the target from the complex environment background.This method reduces the dimension of threedimensional point cloud processing to two-dimensional image processing,and use the Particle Swarm Optimization algorithm to identify and segment the edge of target.The 3D point cloud of the segmented target is obtained by the mapping relation of pixel coordinates.This method not only significantly improves the segmentation speed,but also not affected by the noise in depth direction.Because of the high horizontal resolution of the surface array detector,the segmentation accuracy is guaranteed.For the measurement of non-cooperative target pose parameters,the Coherent Point Drift registration method with high algorithm efficiency and strong anti-noise capability is studied.The problem of point cloud registration is transformed into the problem of probability density estimation,covariance descriptors are used to expand the robust registration of CPD algorithm in complex scenarios.The algorithm can maintain good robustness and high registration accuracy in the presence of abnormal points and noises.It lays a foundation for the calculation of pose parameters of non-cooperative targets in unknown environment.On the basis of research results,the "Flash" 3D imaging capability of polarizationmodulated planar array Lidar to distant moving targets is verified by semi-simulation experiment and field measurement data experiment.Aiming at the special working mode of planar array Lidar,the performance of registration method is analyzed theoretically.The accuracy of pose measurement and its applicability under different conditions were evaluated from the aspects of target structure characteristics,differences of movement state and changes of environmental condition.In summary,the novel planar array 3D imaging Lidar proposed in this paper has the advantages of high transverse resolution,high imaging accuracy,fast imaging speed and long operating distance.At the same time,combining with the structure characteristics of the planar array Lidar system,the multi-dimensional data fusion idea is proposed,which provides a new solution for the algorithms of point cloud segmentation,registration and pose measurement.It provides a complete technical idea from hardware platform to software algorithm in the field of non-cooperative dynamic target pose measurement.