Research On 3D Shape And Pose Measurement Of Uncooperative Spacecraft Based On ToF-monocular Camera

With the development of space technology and the increasing of space exploration activities,an increasing number of mission objects are uncooperative targets,such as space debris,space junk,defective satellites,enemy spacecraft,etc.They are unable to proactively provide any effective cooperative information,which brings great challenges to on-orbit servicing(OOS),like manipulation,repairing,and debris removal.Therefore,the 3D shape measurement and pose estimation of uncooperative spacecraft is an important prerequisite for subsequent missions such as space exploration and situational awareness.Electro-Optical(EO)sensors have the advantages of being fully automatic,low cost,compact structure,etc.,and are important methods for the characteristic measurement of space targets.However,due to the lack of prior information,limited observation conditions,complex space background and environment,it is difficult to meet the mission requirements of the accurate measurement by relying on a single EO sensor.Aiming at the characteristics of space missions and the measurement requirements of uncooperative spacecraft,a 3D shape and pose measurement method of uncooperative spacecraft based on ToF(Time-of-Flight)and monocular cameras is proposed.A ToF and monocular camera measurement system is built,which combines the advantages of the ToF camera and the monocular camera.The ToF-monocular camera system is calibrated and completes the depth-texture image alignment.This paper realizes the shape measurement and pose estimation of uncooperative spacecraft,and the technical contributions can be summarized as follows:1.A 3D shape measurement method for uncooperative spacecraft based on depth super-resolution is proposed.This paper overcomes the low resolution and heavy noise problem of the ToF camera by making use of the texture information provided by the monocular camera.The high-resolution texture image in the same scene guides the low-resolution depth image recovery to reconstruct the fine textures and structures of uncooperative spacecraft.This paper presents an S-estimator-based iterative closest points(S-ICP)algorithm for shape reconstruction and completion by aligning point clouds from different views.Compared with the original point cloud obtained by the ToF camera,the number and density of the reconstructed 3D point cloud are significantly improved,with higher accuracy.2.A relative pose measurement method for uncooperative spacecraft based on2D-3D line correspondences is proposed.Aiming at the problems of complex background and occlusion of target feature points during pose measurement,based on the 3D reconstruction results of uncooperative spacecraft,a pose measurement method based on 2D-3D line correspondences is proposed.First,a ToF-monocular camera measurement system is used to take image sequences of the target.Then,establish2D-3D line correspondences,which effectively associate 2D images with the reconstructed 3D model.Finally,2D-3D line correspondences are used to minimize the error function to solve the optimal relative pose parameters.