Real-time Pose Determination of Ultra-Close Non-Cooperative Satellite based on Time-of-Flight Camera

This paper proposes an ultra-close range real-time pose determination algorithm for space non-cooperative satellite capture missions. The method leverages a Time-of-Flight (ToF) camera to establish a comprehensive pose determination framework, encompassing both pose tracking and pose estimation. Initially, the Iterative Closest Point (ICP) algorithm is employed to estimate the pose between two consecutive point clouds, forming the basis for pose tracking. Keyframes are inserted during the tracking process. Subsequently, the paper introduces a pose estimation algorithm that integrates two-dimensional features and three-dimensional information for pose initialization and keyframe pose estimation. Feature matching between keyframes ensures consistent pose estimation across different keyframes. Finally, the pose estimation and pose tracking are merged, utilizing the pose estimation results from keyframes to mitigate accumulated tracking errors and achieve high-precision real-time pose determination. Simulation results demonstrate the effectiveness of the proposed algorithm, meeting the real-time requirement of space capture tasks with high accuracy in pose tracking.