Relative Navigation For Malfunctioned Satellite Based On SLAM Using Structure Features

With the rapid development of space robots and rendezvous and docking technologies,on-orbit service technology represented by spacecraft life extension,malfunctioned maintenance,and rail garbage cleaning has become an important development direction of future space activities.Malfunctioned satellite relative pose measurement technology is the premise of on-orbit service.At present,spacecraft and space debris in orbit are non-cooperative targets,that is,there is no cooperation sign for auxiliary measurement and a handle for easy arrest.In order to efficiently obtain the relevant parameters of non-cooperative spacecraft,the service satellite is equipped with measurement sensors with non-contact,low energy consumption and high precision to measure the relative pose of the target.In this paper,the non-cooperative spacecraft pose determination algorithm is studied.A SLAM algorithm based on combined features for relative navigation of malfunctioned satellites is proposed,which provides theoretical and technical preparation for on-orbit operation.Lidars and cameras are commonly used sensors for relative measurement in orbit,and Kinect is used to simulate the aforementioned sensors.Based on the information acquisition principle of the depth camera and the RGB image imaging model,this paper studies the algorithm of merging two kinds of information into three-dimensional color point cloud,and verifies the correctness by experiments.At the same time,in order to ensure the accuracy of the malfunctioned satellite tracking measurement,a coordinate system is established to facilitate the relative pose and relative position modeling,and the conversion relationship between the coordinate systems is given.Malfunctioned satellites generally have regular geometric features.After analyzing the surface features of the malfunctioned satellite,this paper selects the rectangular surface and circular structure of the satellite body as the feature recognition object.After the point cloud data is acquired by the depth camera,the invalid background point cloud and the interference noise point cloud are eliminated by the filtering algorithm.A feature recognition scheme based on geometric size features and RGB information is proposed.The coordinate transformation relationship between the feature plane ontology coordinate system and the sensor measurement coordinate system is derived,and the point cloud plane is calculated in the feature plane ontology coordinate system.Geometric dimension,as the basis for identifying the rectangular surface;the feature circle is extracted by RANSAC algorithm,and the radius of the circle is used as the basis for identifying the circle feature.Eliminate the singularity caused by geometric symmetry plane features using RGB information.A SLAM algorithm for relative pose estimation using the combined features of faulty satellites is proposed.The method does not need to acquire the complete point cloud data of the malfunctioned satellite surface,and fully utilizes multiple rectangular surface features or circle features in the field of view to construct a combined feature body coordinate system.The EKF-SLAM algorithm is used to dynamically model the surface contour of the malfunctioned satellite while estimating the position and attitude of the faulty satellite.Firstly,the method of establishing the combined feature coordinate system is studied.The position of the origin of the combined feature coordinate system in the sensor measurement coordinate system and the attitude of the combined feature coordinate system in the sensor measurement coordinate system are taken as the observation values of the EKF-SLAM;The relative motion model of the satellite service star and the attitude kinematics model of the fault satellite are used as the state equation of the EKF-SLAM;the relative pose parameter filter based on the combined feature is designed.Simulation experiments verify the effectiveness of the combined feature SLAM relative pose measurement algorithm.The relative satellite physical simulation experimental platform was built by using the malfunctioned satellite 3D model,rotating table,translation stage and Kinect.The SLAM algorithm model of the physical simulation experiment platform was deduced,and the estimated performance of the filter was verified by experiments.