6-DoF Pose Estimation of Uncooperative Space Object Using Deep Learning with Point Cloud

Autonomous rendezvous and approaching of spacecrafts and uncooperative space objects is the fundamental portion of future on-orbit satellite maintenance or asteroid exploration mission, while estimating the relative attitude and position of known uncooperative space objects on-board is still one of challenges in these tasks. Recently, the development of flash light detection and ranging sensors (LIDARs) and 3D imaging technology provides a feasible and reliable method for relative navigation. This paper proposes an end-to-end neural network based on Transformer to estimate 6-DoF attitude of the uncooperative space object with point cloud. The experiments are conducted to validate the performance and robustness of our network.