Relative Navigation For Malfunctioned Satellite Based On Depth Information

To reduce the economic damage caused by malfunctioned satellites and the probability of orbital accidents raised by “space junk”,the technology of on-orbit service which focuses on satellite maintenance and junk removal has become one of the future development trends.The precondition of this technology is to measure the relative position and attitude of the malfunctioned satellites,which can't be obtained by themselves initially and have to be detected by the service satellites effectively in real time.Therefore,this paper proposes a measurement method based on the depth information to realize the relative navigation for malfunctioned satellites.The research aims to make a breakthrough in the technology of relative navigation and identification,and provide necessary theoretical and technical reserve for the future on-orbit operation of China.Firstly,an appropriate coordinate system is set to simplify the modeling of the relative position and attitude between malfunctioned satellite and service satellite,and the transformation of different coordinates is given.Then the imaging process of the depth camera is modeled,the measuring principle of depth camera is introduced,and the radial and tangential distortion of the lens is analyzed.Finally,the method of depth camera calibration is introduced based on the 2D target plane.On the basis of analyzing the geometric features of the malfunctioned satellite,this paper selects the rectangular as the characteristic object.After the rectangular feature plane is extracted with the point cloud processing algorithm,this paper presents a method to identify the rectangular feature plane with the geometrical moment of inertia.At first,the obtained point cloud data is processed with the filter to eliminate the background point cloud and the noise point cloud generated by the sensor,and furthermore compress the point cloud data.Then the three algorithms of plane extraction are analyzed and compared through simulation experiment,including LMedS,RANSAC,and MLESAC.In the end,the formula to identify the rectangular plane of the point cloud is derived based on the geometric moment of inertia.The simulation results verify its correctness,that is,the geometrical moment of inertia can be used to identify the rectangular surfaces.In this paper,a method of relative pose estimation is proposed based on the rectangular feature plane of malfunctioned satellite.In this method,it is not necessary to obtain the complete point cloud of the malfunctioned satellite.Instead,the position and attitude is estimated based on the rectangular surface of the satellite.At first,the relative position and attitude between the malfunctioned satellite and the service satellite are calculated with the point cloud rectangular feature plane,and the problem of multiple-solutions due to the symmetry of the rectangle is solved.Then the relative pose model is established,and the relative position parameter filter and the malfunctioned satellite attitude filter are designed.Finally,the correctness is verified by the simulation experiment.In order to verify the effectiveness of the proposed method,the semi physical simulation experiment platform based on the depth information is established.According to the working principle of the platform,the relative position filter and the relative attitude filter are designed,and the verification results are listed.