Dynamic pose estimation based on 3D Point Clouds

Pose estimation is an important step towards spacecraft docking with the space station, as it can make the spacecraft react to the pose change in real time and better accomplish the tracking mission. However, it is difficult to conduct the real rendezvous and docking practice due to the limitations of research conditions and the expenses. In order to facilitate the validation of the pose estimation of the space station in the docking process, a new advanced simulation system was established based on Gazebo and used to estimate the pose of the space station in this paper. The data obtained by laser range finders (LRF) forms the 3D point clouds. With the help of Point Cloud Library (PCL), it is convenient to process the point clouds data as there are many algorithms including filtering, segmentation and visualization. The code is written under the Robot Operating System (ROS) framework and the data is released by ROS topics in the simulation process. Experimental results show a high accuracy of pose estimation of the space station.