A monocular vision space non-cooperative target relative pose measurement method and system based on a feature constraint set
By using a monocular vision method based on feature constraint sets, multiple constraint relationships between target features are established and input into an extended Kalman filter, solving the problem of rapid determination of pose and inertial parameters of non-cooperative targets in space, and achieving efficient navigation and task execution.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- HARBIN INST OF TECH
- Filing Date
- 2023-07-13
- Publication Date
- 2026-05-12
AI Technical Summary
In space missions, existing technologies struggle to quickly determine the pose and inertial parameters of non-cooperative targets in space, especially during close-range relative navigation, which affects missions such as on-orbit maintenance, space debris removal, and the capture and destruction of enemy satellites.
A monocular vision method based on feature constraint sets is adopted. By defining a reference coordinate system, various constraint relationships between target features are established, a feature constraint set equation is constructed, and it is used as a pseudo-measurement input into the extended Kalman filter to improve the amount of observation information and the estimation accuracy.
It enables rapid determination of the pose and inertial parameters of non-cooperative targets in space, supports the effective implementation of on-orbit maintenance and acquisition tasks, and improves navigation accuracy and efficiency.
Smart Images
Figure CN116929305B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of spacecraft relative navigation and pose estimation technology, and in particular to a method and system for measuring the relative pose of a non-cooperative spatial target with monocular vision based on a feature constraint set. Background Technology
[0002] Spacecraft relative attitude and inertial parameter measurement technology is of great significance in space mission scenarios. When the model of a non-cooperative space target is unknown, the estimation of the attitude and inertial parameters of the non-cooperative space target plays an important role in the implementation of missions such as on-orbit maintenance of a faulty spacecraft, active removal of space debris, and destruction and capture of enemy satellites.
[0003] In relative pose and inertial parameter measurement, visual navigation is widely used in spacecraft relative navigation due to its relatively low cost. Compared to lidar, monocular cameras are smaller, consume less power, and are less expensive, and monocular vision systems are relatively simple to configure. Compared to stereo vision, monocular vision has a simpler structure, is easier to calibrate, and occupies less platform area. Based on these characteristics of visual navigation, scholars at home and abroad have conducted extensive research.
[0004] Currently, research on visual navigation technology abroad has made some progress. Italian scholars Vincenzo Pesce et al. used stereo vision measurement combined with filtering methods to estimate the pose and inertial parameters of non-cooperative targets in space. However, stereo vision, using binocular cameras, occupies a large platform space. American scholars Sean Augenstein et al. used a monocular vision SLAM algorithm for attitude tracking and shape reconstruction of unknown targets, but did not estimate the target's inertial parameters. Domestic scholars Ge Dongming et al. used stereo vision measurement methods, selecting three feature points on the target as measurement observations, and combined them with filtering algorithms to estimate the pose and inertial parameters of non-cooperative targets in space. However, this method has a slow convergence speed and cannot achieve rapid estimation of the pose and determination of the inertial parameters of non-cooperative targets in space. Summary of the Invention
[0005] The technical problem to be solved by this invention is:
[0006] To address the problem of close-range relative navigation during missions such as on-orbit maintenance of non-cooperative space targets, active space debris removal, and capture and destruction of enemy satellites, this invention provides a monocular vision method and system for measuring the relative pose of non-cooperative space targets based on feature constraint sets, which is necessary to quickly determine the pose estimation and inertia parameters of non-cooperative space targets during rendezvous.
[0007] The technical solution adopted by the present invention to solve the above-mentioned technical problems is as follows:
[0008] A method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set, the method comprising the following steps:
[0009] Define a reference coordinate system;
[0010] Modeling the relative kinematics and dynamics of the tracking spacecraft and the target spacecraft;
[0011] Based on a monocular vision measurement model, we identify different constraint relationships between target features and establish feature constraint set equations.
[0012] By establishing feature constraint set equations for different constraint relationships between target features and incorporating these constraint equations into the measurement equations, the amount of measurement information is increased, thereby improving the accuracy of target estimation.
[0013] Extended Kalman filters are designed based on different feature-constrained measurement models to estimate the target's motion parameters and inertial parameters.
[0014] Based on the target features, the pose parameters and inertial parameters of non-cooperative targets in space are filtered and estimated. Finally, a measurement system (model) for the pose and inertial parameters of non-cooperative targets in space is designed, thereby realizing the relative pose measurement of non-cooperative targets in space based on a monocular vision feature constraint set.
[0015] Furthermore, the space target is an unknown, non-cooperative entity.
[0016] Furthermore, the reference coordinate system is defined as follows:
[0017] Geocentric equatorial inertial coordinate system (I system): The origin is located at the Earth's center of mass, X I The axis points in the direction of the vernal equinox; Z I The axis points towards the Earth's North Pole; Y I The axis is determined according to the right-hand rule.
[0018] LVLH coordinate system (O system): The origin of the coordinate system is the center of mass of the tracking spacecraft, X O The axis points from the Earth's center to the direction of the spacecraft's center of mass; Y O The axis lies within the orbital plane and is perpendicular to X. O The axis and the angle between it and the velocity direction are acute; Z O The axis is determined according to the right-hand rule.
[0019] Tracking spacecraft body coordinate system (N system): The body coordinate system with the center of mass of the tracking spacecraft as the origin, and the coordinate axes of the tracking spacecraft body coordinate system coincide with the principal axes of inertia.
[0020] Target spacecraft body coordinate system (Γ system): With the target spacecraft's center of mass as the origin, this coordinate system is assumed to coincide with the target's principal inertial axes. Let the principal axis of minimum inertia be X. Γ The axis with the greatest moment of inertia is Z. Γaxis, Y Γ The axis is determined according to the right-hand rule.
[0021] Camera coordinate system (C-frame): with the camera optical center as the origin and the camera optical axis as the Z-axis. C Axis, X C axis, Y C The axis is parallel to the image plane.
[0022] Furthermore, the relative kinematics and dynamics of the tracking spacecraft and the target spacecraft are modeled in the inertial coordinate system, as follows:
[0023] According to Newton's second law, in the absence of disturbance, the equations of motion for the tracking spacecraft and the target spacecraft are as follows:
[0024]
[0025] Where μ is the Earth's gravitational constant, r L r T Let be the position vectors of the center of mass in the inertial frame, and be the position vectors of the center of mass of the target spacecraft and the tracking spacecraft, respectively. The magnitudes of these position vectors can be expressed as: a is the semi-major axis, e is the eccentricity, and θ is the true angle of approach. Define ρ = r L -r T =(xyz) T Then, in the orbital coordinate system, the relative translational dynamic equation of the target spacecraft relative to the tracking spacecraft is:
[0026]
[0027]
[0028]
[0029] in Let x, y, and z be the second derivatives, respectively. Let be the angular velocity of the orbital coordinate system in inertial space. Angular acceleration;
[0030] The relative attitude of the target spacecraft with respect to the tracking spacecraft is represented using Euler quaternions q0 q1 q2 q3, where attitude quaternion q = [q0 q1 q2 q3]. v ] T =[q0 q1 q2 q3] T D(q) is the rotation matrix from the target coordinate system to the tracking coordinate system, and D(q) is expressed as:
[0031]
[0032] Where q × For vector qv antisymmetric matrix,
[0033] The attitude kinematics of the target spacecraft relative to the tracking spacecraft are represented by quaternions:
[0034]
[0035] in: ω is the angular velocity vector of the target spacecraft relative to the tracking spacecraft.
[0036] In the orbital coordinate system, the relative angular velocity ω between the target spacecraft and the tracking spacecraft is expressed as:
[0037]
[0038] The rotation matrix from the target spacecraft's coordinate system to the tracking spacecraft's coordinate system. Then, the dynamic equation of the target spacecraft's rotation relative to the tracking spacecraft is:
[0039]
[0040] Furthermore, a measurement model based on monocular vision is established to extract target feature points, and feature constraint set equations are established for various constraint relationships between target features. Specifically:
[0041] (1) Assuming the projection center of the camera coincides with the center of mass of the tracking spacecraft, select the coordinates of n feature points on the target as the measured values of the observation, such as Figure 1 As shown. Then, within the system of tracking the spacecraft:
[0042]
[0043] (2) Assume the projection coordinates P of the target feature point on the image. i 'For [XY], the coordinates P of the target feature point in the target volume coordinate system. i T for The target feature point has coordinates P in the camera coordinate system i C for The target feature point vector in each camera coordinate system is then represented in the tracking spacecraft body coordinate system as follows: Let be the rotation matrix from the camera coordinate system to the tracking spacecraft's body system. The relationship between the target feature points and the image projection coordinates in the camera coordinate system is as follows: Figure 2 As shown, according to the principle of projective geometry, the i-th feature point P i C The coordinates in the image are:
[0044]
[0045] Image coordinates P i Expressed in pixels:
[0046]
[0047] Where f is the camera focal length, and (u0 v0) are the pixel coordinates of the principal point. x and s y These represent the horizontal and vertical pixel densities [pix / m], respectively.
[0048] (3) The system observation equation is Z i =h(x)+ε k , ε k To measure sensor noise, the equation 0 = g is used. i (ξ1,ξ2,L) represent i distinct pseudo-measurement equations. Then the constraint set observation equation can be expressed as:
[0049]
[0050] The various constraint relationships between the target features establish the feature constraint set equation. When some features in the feature points are collinear, the collinearity 0 = g1(ξ1,ξ2,…) constraint equation is established, as follows: Figure 3 As shown, assume o is the target centroid, and feature points P1, P2, ..., P i P k Collinear, the vector corresponding to the feature point is When 2 ≤ i < k, the collinearity constraint equation is:
[0051] The various constraint relationships between the target features establish the feature constraint set equation. When some features among the feature points are circular constraint equations, a concyclic 0 = g2(ξ1,ξ2,…) constraint equation is established, as follows: Figure 4 As shown, assume o is the target centroid, and feature points P1, P2, ..., P i They are coplanar and form a circle with radius R.
[0052] The equation for the circular constraint is:
[0053] The various constraint relationships between the target features establish the feature constraint set equation. When some features among the feature points are coplanar constraint equations, a coplanar 0 = g3(ξ1,ξ2,…) constraint equation is established. P1, P2, P3, and P4 are four coplanar feature points on the target. Any three of these four feature points form two planes, and n1 and n2 are the normal vectors of the two planes, respectively. Figure 5 As shown, the coplanar constraint equation is:
[0054]
[0055] For the various constraint relationships between the target features, a feature constraint set equation is established, and the feature constraint set observation equation can be expressed as:
[0056]
[0057] By establishing feature constraint set equations based on various constraint relationships between target features, and inputting the constraint set as a pseudo-measurement into the filter, the amount of observation information is increased, enabling rapid determination of motion and inertial parameters.
[0058] Furthermore, for non-cooperative unknown targets in space, the parameters to be estimated include the inertial parameterization of the non-cooperative target and the estimation of the inertial matrix. Specifically:
[0059] Assuming the target spacecraft's body coordinate system is aligned with its principal inertial axes, let the target spacecraft's moment of inertia be I. xx I yy I zz For the measurement model in this patent, it is assumed that the rigid body target is spatially unknown and non-cooperative. Estimating the inertial parameters requires prior knowledge of the target's inertial matrix. In the case of no torque on the target, scholar BE. Tweddle has demonstrated the observability of the inertial matrix, showing that only two of the three degrees of freedom are observable. Therefore, the logarithmic inertia ratio method is used. This method has two degrees of freedom corresponding to two random variables, let... The diagonal moment of inertia matrix is then simplified to obtain the expression for the scaling factor:
[0060]
[0061] in:
[0062] Furthermore, the design of extended Kalman filters based on different feature-constrained measurement models is as follows:
[0063] Define the state vector of the extended Kalman filter.
[0064] The relative motion dynamics state equations of the target spacecraft are discretized; the attitude kinematics equations and attitude dynamics equations of the target spacecraft are also discretized. Based on the extended Kalman filter theory, an extended Kalman filter is designed, with the following steps:
[0065] Based on the above nonlinear state equation and observation equation, it can be written as:
[0066]
[0067] Linearizing the nonlinear equations to first order respectively yields:
[0068] initialization: P0 = P(t0)
[0069] State prediction:
[0070] Covariance matrix prediction:
[0071] EKF gain matrix prediction:
[0072] The state is updated based on the state prediction and filtering gain. Through iteration, the relative position, velocity, attitude, angular velocity, and inertia ratio of the non-cooperative target can be estimated quickly and accurately.
[0073] Furthermore, by establishing feature constraint set equations based on different constraint relationships between target features, the constraint set is used as a pseudo-measurement input to the filter to improve the amount of observation information and estimation accuracy, thereby realizing the rapid determination of the pose estimation and inertial parameters of non-cooperative targets in space.
[0074] Based on the method of this invention, a space non-cooperative target pose measurement system was designed as shown in the attached figure. Figure 6 As shown.
[0075] The system comprises three parts: a measurement unit primarily using a monocular camera, a data processing unit, and a result output unit. The measurement unit employs a monocular camera as the measuring element, acquiring sequential spatial images of non-cooperative targets. The data processing unit extracts target features from the acquired images, establishes measurement equations, and models the target's relative attitude dynamics and kinematics. Filters are used to estimate the target's relative pose and inertial parameters. Finally, the result output unit outputs the target pose estimate and its accuracy. This system can rapidly process the relative pose and inertial parameters of non-cooperative targets.
[0076] The present invention has the following beneficial technical effects:
[0077] Building upon the research of domestic and international scholars mentioned in the background section, this invention proposes a monocular vision-based method and system for measuring the relative pose of non-cooperative targets in space, based on feature constraint sets. By establishing feature constraint set equations based on different constraint relationships between target features, and using the constraint set as a pseudo-measurement input to a filter, the invention improves the filter's observation information and estimation accuracy, thereby achieving rapid estimation of the pose and inertial parameters of non-cooperative targets in space. This invention defines a reference coordinate system; models the relative kinematics and dynamics of the tracking spacecraft and the target spacecraft; establishes a measurement model; extracts target feature points; establishes feature constraint equations based on the target feature constraint set; selects the motion and inertial parameters to be estimated; and designs an extended Kalman filter, ultimately achieving rapid estimation of the motion and inertial parameters of non-cooperative targets in space. Furthermore, this application designs a monocular vision-based system for measuring the pose of non-cooperative targets in space. This system can rapidly process the relative pose and inertial parameters of non-cooperative targets.
[0078] Compared to stereo vision, monocular vision has a simpler structure, is easier to calibrate, and occupies a smaller platform area. Therefore, this application adopts monocular vision measurement.
[0079] This invention effectively solves the problem of close-range relative navigation during missions such as on-orbit maintenance of non-cooperative targets in space, active removal of space debris, and capture and destruction of enemy satellites. When prior information such as target motion and structure is lacking, rapid estimation of the pose, motion, and inertial parameters of non-cooperative targets plays an important role in mission implementation.
[0080] This application addresses close-range relative navigation during on-orbit maintenance and capture of non-cooperative space targets. It utilizes monocular vision measurement, employing a monocular camera as the measurement sensor to extract target feature points. Multiple constraint relationships (various constraint relationships) are established between these target features to form a feature constraint set. This constraint set is then used as a pseudo-measurement input to a filter to increase the amount of observational information. Ultimately, this enables the rapid determination of the motion and inertial parameters of non-cooperative space targets, thus better facilitating the implementation of on-orbit missions involving non-cooperative space targets. Therefore, this invention is of great significance for the rapid estimation of the pose and inertial parameters of non-cooperative space targets during missions such as on-orbit maintenance, active space debris removal, capture of non-cooperative target satellites, and capture and destruction of enemy satellites. Attached Figure Description
[0081] To more clearly illustrate the key technical points of this application, the following will provide accompanying drawings to illustrate the key technical points, so as to facilitate a better understanding of the technical solution of this application.
[0082] Figure 1 This is a schematic diagram of the coordinate system of the tracking star and the target star system in this invention; Figure 2This is a schematic diagram of monocular camera measurement in this invention; Figure 3 This is a schematic diagram illustrating the collinearity constraint of some feature points among the target feature points in this invention; Figure 4 This is a schematic diagram illustrating the circular constraints on some feature points within the target feature points of this invention; Figure 5 This is a schematic diagram illustrating the coplanar constraint of some feature points among the target feature points in this invention; Figure 6 This is a schematic diagram of the spatial non-cooperative target pose and inertial parameter measurement system in this invention. Figure 7 This is a flowchart of the method described in this invention. Detailed Implementation
[0083] To more clearly describe the implementation method, process, and innovative aspects of this application, the following will be combined with the appendix. Figure 1-7 A complete process description of the embodiments of the present invention is provided.
[0084] First, a method for measuring the relative pose of non-cooperative targets in monocular vision space based on feature constraint sets defines a reference coordinate system.
[0085] Step 201: In the orbital coordinate system (O frame), the relative translational dynamics of the target spacecraft relative to the tracking spacecraft are modeled as follows:
[0086]
[0087]
[0088]
[0089] in Let x, y, and z be the second derivatives, respectively. Let be the angular velocity of the orbital coordinate system in inertial space. ω is angular acceleration; x, y, z: position vectors of the target relative to the tracking spacecraft in the inertial frame; μ: Earth's gravitational constant; r L : To track the orbital radius of a spacecraft.
[0090] Step 202: The relative attitude of the target spacecraft with respect to the tracking spacecraft is represented using Euler quaternions q0 q1 q2 q3, where attitude quaternion q = [q0 q1 q2 q3]. v ] T =[q0 q1 q2 q3] T D(q) is the rotation matrix from the target coordinate system to the tracking coordinate system, and D(q) is expressed as:
[0091]
[0092] Where q × For vector q v antisymmetric matrix,
[0093] The attitude kinematics equations of the target spacecraft relative to the tracking spacecraft are represented by quaternions:
[0094]
[0095] in: ω is the angular velocity vector of the target spacecraft relative to the tracking spacecraft.
[0096] Step 203: In the orbital coordinate system, the relative angular velocity ω between the target spacecraft and the tracking spacecraft is expressed as: in The rotation matrix from the target spacecraft coordinate system to the tracking spacecraft coordinate system. The dynamic equations of the target spacecraft's rotation relative to the tracking spacecraft:
[0097]
[0098] I T I L : These are the rotational inertia matrices of the target spacecraft and the tracking spacecraft, respectively;
[0099] N T N L These are the external torques of the target spacecraft and the tracking spacecraft, respectively.
[0100] ω T|Γ : The angular velocity of the target spacecraft in the body coordinate system;
[0101] ω L|O Track the angular velocity of a spacecraft in its orbital coordinate system;
[0102] ω: angular velocity of the target spacecraft relative to the tracking spacecraft.
[0103] Step 301: Establish a measurement model based on monocular vision, extract target feature points, and establish feature constraint set equations for different constraint relationships between target features. Specifically:
[0104] (1) Assuming the projection center of the camera coincides with the center of mass of the tracking spacecraft, select the coordinates of n feature points on the target as the measured values of the observation, as shown in the attached figure. Figure 1 As shown.
[0105]
[0106] (2) Assume the projection coordinates P of the target feature point on the image. i 'xy' represents the coordinates of the target feature point in the target volume coordinate system. for The target feature point has coordinates P in the camera coordinate system iC for The target feature point vector in each camera coordinate system is then represented in the tracking spacecraft body coordinate system as follows: This represents the rotation matrix from the camera coordinate system to the tracking spacecraft's body system. The relationship between the target feature points and the image projection coordinates in the camera coordinate system is shown in the appendix. Figure 2 As shown:
[0107] According to the principles of projective geometry, the i-th feature point P i C The coordinates in the image are:
[0108]
[0109] Image coordinates P i Expressed in pixels:
[0110]
[0111] Where f is the camera focal length, and (u0 v0) are the pixel coordinates of the principal point. x and s y These represent the horizontal and vertical pixel densities [pix / m], respectively.
[0112] (3) The system observation equation is Z i =h(x)+ε k , ε k To measure sensor noise, the equation 0 = g is used. i (ξ1,ξ2,…) represent i distinct pseudo-measurement equations. Then the constraint set observation equation can be expressed as:
[0113]
[0114] Step 302: Establish feature constraint set equations for different constraint relationships between target features. When some features in the feature points are collinear, establish the collinearity 0 = g1(ξ1,ξ2,…) constraint equation, as shown in the appendix. Figure 3 As shown, assume o is the target centroid, and feature points P1, P2, ..., P i P k Collinear, the vector corresponding to the feature point is When 2 ≤ i < k, the collinearity constraint equation is:
[0115] Step 303: Establish feature constraint set equations for different constraint relationships between target features. When some features in the feature points have circular constraint equations, establish concyclic 0 = g2(ξ1,ξ2,…) constraint equations, as shown in the appendix. Figure 4 As shown, assume o is the target centroid, and feature points P1, P2, ..., P iThe elements are coplanar and form a circle with radius R. Then the equation of the circle constraint is:
[0116] Step 304: Establish feature constraint set equations based on different constraint relationships between target features. When some features among the feature points are coplanar constraint equations, establish a coplanar 0 = g3(ξ1,ξ2,…) constraint equation. P1, P2, P3, and P4 are four coplanar feature points on the target. Take any three of the four feature points to form two planes, and n1 and n2 are the normal vectors of the two planes, respectively. (See attached diagram) Figure 5 As shown, the coplanar constraint equation is:
[0117] Step 305: Establish the feature constraint set equation based on the various constraint relationships between target features. Then, the feature constraint set observation equation is:
[0118]
[0119] By establishing feature constraint set equations based on various constraint relationships between target features, and inputting the constraint set as a pseudo-measurement into the filter, the amount of observation information is increased, enabling rapid determination of motion and inertial parameters.
[0120] Step 401: Select the parameters to be estimated, including the motion parameters and inertial parameters of the non-cooperative target. Details are as follows:
[0121] Assuming the target spacecraft's body coordinate system is aligned with its principal inertial axes, let the target spacecraft's moment of inertia be I. xx I yy I zz The equations of motion for the target spacecraft relative to the tracking spacecraft are:
[0122]
[0123] Assume the external torque N acting on the target spacecraft T If the value is 0, then from the above formula we can obtain:
[0124]
[0125] Step 501: For the measurement model of this patent, it is assumed that the target is a rigid body, the target is in an unknown space and is non-cooperative, and the principal axes of inertia are inherent properties of the rigid body. Estimating the inertia parameters requires prior knowledge of the inertia matrix. In the case of no torque motion of the target, the inertia matrix is not completely observable; only two of the three degrees of freedom are observable. Therefore, the logarithmic inertia ratio method is used. This method has two degrees of freedom corresponding to two random variables, let... The diagonal moment of inertia matrix is then simplified to obtain the expression for the scaling factor:
[0126]
[0127] in:
[0128] Step 601: Design of Extended Kalman Filters Based on Different Feature-Constrained Measurement Models. Details are as follows:
[0129] Define the state vector of the extended Kalman filter.
[0130] Discretize the relative kinematics and dynamic state equations of the target spacecraft; discretize the attitude dynamics equations and attitude kinematics equations of the target spacecraft. Based on the extended Kalman filter theory, design the extended Kalman filter, as follows:
[0131] Based on the above nonlinear state equation and observation equation, it can be written as:
[0132]
[0133] Linearizing the nonlinear equations to first order respectively yields:
[0134] initialization:
[0135] State prediction:
[0136] Covariance matrix prediction:
[0137] EKF gain matrix prediction:
[0138] Then, the state is updated based on the state prediction and filter gain. Through iteration, the relative position, velocity, attitude, angular velocity, inertia ratio and other information of the non-cooperative target can be estimated quickly and accurately.
[0139] Step 701: By establishing feature constraint set equations based on different constraint relationships between target features, the constraint set is used as a pseudo-measurement input to the filter to improve the amount of observation information and estimation accuracy, thereby realizing the rapid determination of the pose estimation and inertia parameters of non-cooperative targets in space. Furthermore, this application designs a non-cooperative target pose measurement system for space as shown in the attached figure. Figure 6 As shown, the system includes a measurement unit with a monocular camera as the main component, a data processing unit, and a result output unit. This system can quickly process the relative pose and inertial parameters of non-cooperative targets.
[0140] Step 702: To achieve real-time estimation of the relative pose and inertial parameters of non-cooperative targets in space, the measurement system is divided into three parts: a measurement unit, a data processing unit, and an output unit. The measurement unit uses a monocular camera as the measurement element to acquire a sequence of images of non-cooperative targets in space. The data processing unit extracts target features from the acquired images, establishes measurement equations, and models the relative attitude dynamics and kinematics. The relative pose and inertial parameters of the target are estimated by a filter. Finally, the result output unit outputs the target pose estimate and the estimation accuracy.
[0141] This invention addresses the challenges of performing on-orbit maintenance, proactive space debris removal, and sabotage / capture of enemy satellites on non-cooperative spacecraft when the model of the non-cooperative space target is unknown. It enables accurate estimation of the attitude and inertial parameters of the non-cooperative space target, which is crucial for mission implementation and completion.
Claims
1. A method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set, the method comprising the following steps: Define the reference coordinate system: the geocentric equatorial inertial coordinate system. System: The origin is located at the Earth's center of mass. The axis points in the direction of the vernal equinox; The axis points towards the Earth's North Pole; The axis is determined according to the right-hand rule; LVLH coordinate system System: The origin of the coordinate system is the center of mass of the tracking spacecraft. The axis points from the Earth's center of gravity towards the center of mass of the tracking spacecraft. The axis lies within the track plane and is perpendicular to it. The axis forms an acute angle with the direction of velocity; The axis is determined according to the right-hand rule; Tracking spacecraft body coordinate system System: A body coordinate system with the center of mass of the tracking spacecraft as its origin, and the coordinate axes of the tracking spacecraft's body coordinate system coincide with the principal axes of inertia; Target spacecraft body coordinate system Coordinate system: With the target spacecraft's center of mass as the origin, assume this coordinate system coincides with the target's principal inertial axes; let the principal axis of minimum inertia be... The axis with the maximum moment of inertia is axis, The axis is determined according to the right-hand rule; Camera coordinate system System: with the camera's optical center as the origin and the camera's optical axis as the coordinate system. axis, axis, The axis is parallel to the image plane; Perform dynamic modeling of the relative motion between the tracking spacecraft and the target spacecraft; Construct a measurement model, extract target feature points, construct target feature constraints to establish feature constraint equations, and based on a monocular vision measurement model, find multiple constraint relationships between target features to establish feature constraint set equations. Based on the multiple constraint relationships between target features, a feature constraint set equation is established. The constraint equation is then added to the measurement equation to increase the amount of measurement information, thereby improving the target estimation accuracy. Select parameters to be estimated, including non-cooperative target pose parameters and inertial parameters; design extended Kalman filters based on different feature constraint measurement models to estimate the target pose parameters and inertial parameters; Based on the target features, the pose parameters and inertial parameters of the non-cooperative target in space are filtered and estimated. Finally, a measurement model for the pose and inertial parameters of the non-cooperative target in space is designed, thereby realizing the relative pose measurement of the non-cooperative target in space based on the feature constraint set using monocular vision. A measurement model based on monocular vision is established, target feature points are extracted, and feature constraint set equations are established for various constraint relationships between target features, as detailed below: (1) Assuming the projection center of the camera coincides with the center of mass of the tracking spacecraft, select the target on Using the coordinates of each feature point as the measured value of the observation, then within the tracking spacecraft's own system: （7） In the formula: : The position vector of the target feature point in the tracking spacecraft's body coordinate system; : Tracking the vector from the spacecraft's center of mass to the target's center of mass in the spacecraft's body coordinate system; : The position vector of the target feature point in the target volume coordinate system; Target feature points; (2) Assume the projected coordinates of the target feature points on the image for The coordinates of the target feature points in the target volume coordinate system for The coordinates of the target feature points in the camera coordinate system for Then, the target feature point vector in each camera coordinate system, when transformed to the tracking spacecraft body coordinate system, is represented as: , Let be the rotation matrix from the camera coordinate system to the tracking spacecraft's body system; the relationship between the target feature points and the image projection coordinates in the camera coordinate system is: According to the principles of projective geometry, the first... Feature points The coordinates in the image are: （8） Image coordinates Expressed in pixels: （9） in For camera focal length, These are the pixel coordinates of the principal point. and These represent the horizontal and vertical pixel densities, respectively. ; (3) The system observation equation is , To measure sensor noise; use the equation expression represent Given several different pseudo-measurement equations, the constraint set observation equation can be expressed as: （10） The feature constraint set equation is established by defining the various constraint relationships between the target features: when some features in a feature point are collinear, the collinear constraint is established. Constraint equations, assumptions For the target centroid, feature points , … , Collinear, the vector corresponding to the feature point is , ,when hour, The collinearity constraint equation is: (11) The various constraint relationships between the target features establish the feature constraint set equation. When some features among the feature points are circular constraint equations, a concyclic constraint equation is established. Constraint equations, assumptions For the target centroid, feature points , … They are coplanar and form a circle with a radius of . ; The equation for the circular constraint is: (12) The feature constraint set equation is established by considering the various constraint relationships between the target features: when some features in a feature point are coplanar, a coplanar constraint equation is established. constraint equations , , , Given four coplanar feature points on the target, take any three of these feature points to form two planes. , These are the normal vectors of the two planes, respectively; The coplanar constraint equation is: （13） From equation (10-13), the observation equation for the characteristic constraint set is: （14） By establishing feature constraint set equations based on various constraint relationships between target features, and using the constraint set as a pseudo-measurement input to the filter, the amount of observation information can be increased, enabling rapid determination of motion and inertial parameters.
2. The method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set as described in claim 1, characterized in that: The target is a non-cooperative spacecraft.
3. The method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set as described in claim 1, characterized in that: In the inertial coordinate system, the relative motion between the tracking spacecraft and the target spacecraft is dynamically modeled as follows: According to Newton's second law, in the absence of disturbance, the equations of motion for the tracking spacecraft and the target spacecraft are as follows: （1） in : is the Earth's gravitational constant. , Let be the position vectors of the center of mass in the inertial frame, and be the position vectors of the center of mass of the target spacecraft and the tracking spacecraft, respectively. The magnitudes of these position vectors can be expressed as: , For the semi-major axis, For eccentricity, A true near angle; defined Then, in the orbital coordinate system, the relative translational dynamic equation of the target spacecraft relative to the tracking spacecraft is: （2） in They are respectively The second derivative, Let be the angular velocity of the orbital coordinate system in inertial space. Angular acceleration; The relative attitude transformation matrix of the target spacecraft relative to the tracking spacecraft uses Euler quaternions. This indicates that the attitude quaternion is... ; The rotation matrix from the target coordinate system to the tracking coordinate system. Represented as: （3） in For vectors antisymmetric matrix, ; The attitude kinematics of the target spacecraft relative to the tracking spacecraft are represented by quaternions: （4） in: , The angular velocity vector of the target spacecraft relative to the tracking spacecraft; In the orbital coordinate system, the relative angular velocity between the target spacecraft and the tracking spacecraft Represented as: （5） The rotation matrix from the target spacecraft coordinate system to the tracking spacecraft coordinate system; then the dynamic equation of the target spacecraft's rotation relative to the tracking spacecraft is: （6） In the formula: , : These are the rotational inertia matrices of the target spacecraft and the tracking spacecraft, respectively; , These are the external torques of the target spacecraft and the tracking spacecraft, respectively. : The angular velocity of the target spacecraft in the body coordinate system; Track the angular velocity of a spacecraft in its orbital coordinate system; Angular velocity of the target spacecraft relative to the tracking spacecraft.
4. The method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set as described in claim 1, characterized in that: For non-cooperative unknown targets in space, the parameters to be estimated include the inertial parameterization of the non-cooperative target and the estimated inertial matrix, as detailed below: Assuming the target spacecraft's body coordinate system is aligned with its principal inertial axes, let the target spacecraft's moment of inertia be... , , Assuming the target is a rigid body, and since the target is spatially unknown and non-cooperative, prior knowledge of the target's inertia matrix is required for inertia parameter estimation. In the case of no torque motion of the target, the inertia matrix is not completely observable; only two of the three degrees of freedom are observable. Therefore, the logarithmic inertia ratio method is used. This method has two degrees of freedom corresponding to two random variables, let... , The diagonal moment of inertia matrix is then simplified to obtain the scaling factor expression: （15） in: .
5. The method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set as described in claim 4, characterized in that: The design of the extended Kalman filter based on different feature-constrained measurement models is as follows: Define the state vector of the extended Kalman filter. Discretize the relative motion dynamics state equation (2) of the target spacecraft; discretize the attitude kinematics equation and attitude dynamics equation (4) and (6) of the target spacecraft. Based on the extended Kalman filter theory, design the extended Kalman filter. The steps are as follows: Based on the above nonlinear state equation and observation equation, it can be written as: （16） Linearizing the nonlinear equations to first order respectively yields: , initialization: , State prediction: (17) Covariance matrix prediction: (18) EKF gain matrix prediction: (19) The state is updated based on the state prediction and filtering gain. Through iteration, the relative position, velocity, attitude, angular velocity, and inertia ratio of the non-cooperative target can be estimated quickly and accurately.
6. A method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set as described in claim 1 or 5, characterized in that: By establishing feature constraint set equations based on different constraint relationships between target features, and using the constraint set as a pseudo-measurement input to the filter to improve the amount of observation information and estimation accuracy, the pose estimation and inertial parameters of non-cooperative targets in space can be rapidly determined.
7. A monocular vision spatial non-cooperative target relative pose measurement system based on feature constraint sets, characterized in that: The system has a program module corresponding to the steps of any one of the claims 1-6 above, and executes the steps in the above-described method for measuring the relative pose of a non-cooperative target in monocular vision space based on a feature constraint set when it is run.
8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program configured to, when invoked by a processor, implement the steps of any one of claims 1-6, a monocular vision spatial non-cooperative target relative pose measurement method based on a feature constraint set.