关键词: 空间非合作目标, 位姿估计, 点云配准, 截断最小二乘估计与半定松弛算法, 迭代最近点算法

Abstract: The pose estimation of space noncooperative targets based on point cloud is often affected by noise. In order to improve the accuracy and robustness of pose estimation for space noncooperative targets, a combination algorithm of truncated least squares estimation and semidefinite relaxation (TEASER) and iterative closest point (ICP) is proposed in this paper. The method includes two parts: coarse registration and fine registration. In coarse registration, the matching pair is found by the signature of histogram of orientation (SHOT) of local point cloud and model point cloud, and then the initial pose is solved by TEASER. In fine registration, ICP is used to optimize the pose estimation results. The Beidou satellite simulation experiment shows that when the noise standard deviation is 3 times the resolution of the point cloud, the translation error of the periodic key frame registration method based on TEASER is less than 3.33cm, and the rotation error is less than 2.18° in the pose estimation of continuous frames. Compared with the traditional ICP method, the average translation error and average rotation error are reduced. The results show that the proposed pose estimation method has good accuracy and robustness.

Key words: space noncooperative target, pose estimation, point cloud registration, truncated least squares estimation and semidefinite relaxation algorithm, iterative closest point algorithm