Estimation of Rotational Motion Parameters for Non-Cooperative Targets Using Continuous 3D Point Cloud Sequences

Aiming at the observation of the rotational motion process of non-cooperative targets in space missions, an estimation method of rotational motion parameters based on continuous 3D point cloud sequences is proposed. Firstly, based on the metric function of geometric feature, the point clouds of the adjacent two frames are respectively classified into the line point set, the surface point set and the ordinary point set. In the coarse registration, for the feature points selected from the line point set and the plane point set, the feature vectors are constructed for matching and the RANSAC algorithm is used to correct the matching results. In the fine registration, according to the category of point cloud, the corresponding loss function is constructed, and the fine registration matrix between adjacent frames is obtained by improving the ICP algorithm. Secondly, the motion trajectory of the selected point is obtained based on the registration matrix between adjacent frames, and the velocity information of the point is derived from the trajectory. Finally, the velocity information and the position information of the point are substituted into the motion equation to solve the angular velocity of the target, and the registration matrix of adjacent frames is employed to multiply to estimate the target's attitude. Ground experimental results demonstrate the effectiveness of the proposed method.