A relative pose measurement method for non-cooperative spacecraft based on multi-source information fusion

Abstract

The present invention relates to a non-cooperative spacecraft relative posture measurement method based on multi-source information fusion, and the steps are as follows: first, a non-cooperative spacecraft model is designed, and a simulation model of a binocular camera and a laser radar is constructed; then, considering that the binocular camera can obtain all the information of the non-cooperative spacecraft, but the accuracy of the restored depth information is affected by the configuration calibration of the binocular camera, and the laser radar imaging can directly obtain the accurate depth information of the non-cooperative spacecraft, but the laser radar imaging point cloud is sparse, combining the advantages of the two types of sensors, a sensor data fusion structure is designed to perform non-cooperative spacecraft relative posture measurement; finally, based on the measured target relative posture, the laser radar data is used to perform iterative calculations to improve the measurement accuracy of the non-cooperative spacecraft relative posture. The present invention realizes high-precision identification of non-cooperative spacecraft relative posture information through non-cooperative spacecraft relative posture measurement based on binocular camera and laser radar data fusion.

Description

Technical Field

[0001] The invention belongs to the field of non-cooperative spacecraft navigation design, and relates to a non-cooperative spacecraft relative posture measurement method based on multi-source information fusion. Background Art

[0002] At present, the mission requirements of spacecraft tend to be diversified and complex, including various space operations, such as on-orbit refueling, on-orbit maintenance and assembly, on-orbit module replacement and other close-range operation tasks; there is a large amount of space debris in space, and there is a risk of collision when spacecraft are operating in space. Cleaning up space debris can ensure the safety of spacecraft; spacecraft rendezvous and docking requires accurate knowledge of the relative posture information of the main and auxiliary spacecraft; and so on. With the rapid development of space technology and the increasing strategic status of space, the research on space missions of spacecraft has been increasingly valued by major space powers in the world, with broad market application prospects and important military application value, and relative navigation is the prerequisite for realizing the above technologies. For the mission of spacecraft, whether it is on-orbit operation or space debris removal, the objects are mostly non-cooperative spacecraft. Therefore, the exploration of relative navigation methods for non-cooperative spacecraft has broad prospects and important scientific research value.

[0003] For observations of non-cooperative spacecraft, the relative position and pose information of the target is generally measured through the output of various sensitive components, such as cameras or lidar. However, a single sensor has certain limitations in navigation. The accuracy of visual navigation is easily affected by changes in ambient lighting; the resolution of lidar data is low, the point cloud is sparse, and it is impossible to obtain texture information of non-cooperative spacecraft... Different single sensors have their own advantages and disadvantages in navigation applications. How to combine their advantages, effectively fuse information from different sensors, and obtain more accurate and robust target relative position and pose is particularly important for space missions targeting non-cooperative spacecraft.

[0004] Chinese patent CN202010324661.5 considers using only binocular cameras to measure the relative pose of non-cooperative spacecraft, but this method is based on binocular cameras to measure the relative pose of non-cooperative spacecraft, and uses two image processing boards to process binocular images respectively, which makes parameter calibration complicated. If affected by sunlight, the measurement accuracy will be greatly reduced; Chinese patent CN201610557070.6 considers using only laser radar to measure the relative pose of non-cooperative spacecraft, but this method requires the establishment of a precise point cloud model library of non-cooperative spacecraft, and the accuracy of coarse registration depends on the selection accuracy of the model library. Therefore, when performing high-precision relative pose measurement of non-cooperative spacecraft, it is greatly affected by the point cloud model library. It can be seen from this that the robustness and accuracy of the relative pose measurement of non-cooperative spacecraft based on a single sensor are greatly affected by the outside world, so there is an urgent need for a robust and accurate relative pose measurement method for non-cooperative spacecraft.

[0005] Fusion technology is considered to be the most appropriate method to determine the relative position of non-cooperative spacecraft. Sensor information fusion has many advantages: reducing uncertainty, improving accuracy, and reducing costs. Therefore, in order to achieve a certain level of accuracy, integrating multiple sensors can obtain more ideal results, because the limitations of a single sensor can be supplemented by another sensor. For example, images captured by a camera can be used to correct anomalies of inertial sensors. Fusion of information from different sensors, complementary advantages, and use of limited information to obtain high-precision, fast-response non-cooperative spacecraft information has become the mainstream solution for current navigation tasks.

[0006] In summary, the relative posture measurement of non-cooperative spacecraft in the existing technology is easily affected by the external environment, resulting in poor measurement robustness and low accuracy. This patent proposes to use multi-source information to measure the relative posture of non-cooperative spacecraft, and use multiple data sources and multiple data processing methods to measure the relative posture of spacecraft to improve calculation robustness. Summary of the invention

[0007] In order to solve the problem that a single sensor is unable to accurately and quickly obtain the relative pose information of a non-cooperative spacecraft due to the influence of the environment and the characteristics of the sensor itself, the present invention is based on two of the most commonly used sensors for space navigation: a binocular camera and a lidar, and provides a non-cooperative spacecraft relative pose measurement method based on multi-source information fusion, which overcomes the defect of insufficient accuracy of measuring the relative pose of a non-cooperative spacecraft by relying on a single sensor, establishes a binocular camera and a lidar data fusion structure to measure the relative pose of a non-cooperative spacecraft, and ensures the measurement accuracy of the relative pose of the non-cooperative spacecraft.

[0008] According to one aspect of the present invention, a method for high-precision identification of relative pose of non-cooperative spacecraft based on multi-source information fusion is provided, comprising the following steps:

[0009] (1): Design a non-cooperative spacecraft model and build simulation models of binocular cameras and lidar to prepare for acquiring multi-source information;

[0010] (2): Based on the simulation model constructed in step (1), the binocular image and point cloud data of the non-cooperative spacecraft are obtained through the simulation models of the binocular camera and the lidar, and an information fusion scheme of the binocular image and point cloud data is designed to measure the relative pose of the non-cooperative spacecraft, thereby obtaining a rough relative pose of the non-cooperative spacecraft;

[0011] (3): Based on the relative position and pose of the non-cooperative spacecraft obtained in step (2), iterative calculations are performed using the point cloud data to obtain a high-precision relative position and pose of the non-cooperative spacecraft, thereby completing the relative position and pose measurement of the non-cooperative spacecraft.

[0012] Furthermore, the non-cooperative spacecraft model designed in step (1) is:

[0013] X = {x′ i |x′ i =Rx i +t,i=1,2,…,n} (1)

[0014] Among them, x i =[x1 x2 x3] T and x′ i =[x′1 x′2 x′3] T They respectively represent the coordinates of the non-cooperative spacecraft in the world coordinate system before and after rotation and translation, R represents the attitude rotation matrix of the non-cooperative spacecraft, t represents the translation vector of the non-cooperative spacecraft in the world coordinate system, n represents the number of spatial points that constitute the non-cooperative spacecraft model; the superscript T represents the transposed matrix.

[0015] The binocular camera simulation model built in step (1) is:

[0016]

[0017] Among them, [u l v l 1] T and[u r v r 1] T They represent the pixel coordinates of the non-cooperative spacecraft projected onto the left and right pixel planes of the binocular image, is the camera's internal parameter matrix, which represents the camera's internal parameters. 1 / dx and 1 / dy represent the camera's horizontal and vertical resolutions, respectively. f represents the camera's focal length. (c x ,c y ) represents the center position of the image plane, R′ lRepresents the posture rotation matrix of the left camera in the world coordinate system, R′ r Represents the attitude rotation matrix of the right camera in the world coordinate system, Represents the attitude rotation matrix of the left camera relative to the right camera, t′ r Represents the spatial coordinates of the right camera in the world coordinate system, t′ l represents the spatial coordinates of the left camera in the world coordinate system, and Δt′ represents the relative position of the left and right cameras;

[0018] The laser radar simulation model built in step (1) is:

[0019] The laser radar optical axis points to the positive z-axis direction of the world coordinate system, and the simulation model of the non-cooperative spacecraft in the laser radar pointing area is:

[0020]

[0021] Where S = {t Lj ,j=1,…,m} represents the point cloud set of lidar imaging, t Lj =[x Lj y Lj z Lj ] T Represents the position data of the jth point cloud in the laser radar imaging data, t Wj =[x Wj y Wj z Wj ] T represents the spatial position of the i-th point in the point set data constituting the non-cooperative spacecraft, θ i =[θ 1i θ 2i ] T represents the pointing information of the i-th point constituting the non-cooperative spacecraft in the lidar coordinate system, θ 1i ,θ 2i Respectively represent the angle between the i-th target point and the vertical symmetry plane and the horizontal symmetry plane of the laser radar installation, t LiDAR =[x LiDAR y LiDAR z LiDAR ] T represents the spatial coordinates of the lidar in the world coordinate system, θ LiDAR ={[θ LiDAR1 θ LiDAR2 ] T |θ LiDAR1 ∈{k1dθ x ,k1=1,…,p},θ LiDAR2 ∈{k2dθ y ,k2=1,…,q}}, where dθx represents the horizontal scanning interval of the laser radar, dθ y represents the vertical scanning interval of the lidar, p represents the number of horizontal scanning lines of the lidar, which is affected by the scanning interval and the horizontal field of view, q represents the number of vertical scanning lines of the lidar, m represents the number of lidar point clouds obtained, and n represents the number of spatial points that constitute the non-cooperative spacecraft model.

[0022] Further, the specific process of step (2) is as follows:

[0023] First, the binocular image and point cloud data of the non-cooperative spacecraft designed in step (1) are obtained through the binocular camera and lidar simulation model constructed in step (1), and the line features of the non-cooperative spacecraft in the binocular image are extracted. Three lines L1, L2, and L3 are found to satisfy L1⊥(L2||L3), where (·) indicates that the part in the brackets is parallel to the part outside the brackets. The position relationship of the three lines projected into the binocular image and the vanishing point are used to solve the attitude rotation matrix of the non-cooperative spacecraft:

[0024] R compute =[r1 r2 r1×r2] T (4)

[0025] Where r1 = K -1 v1, r2 = K -1 v2, v2 = Kr2 = l2 × l3, null(·) means to find the null space, l1, l2, l3 are the homogeneous coordinates of the three lines L1, L2, L3, r1×r2=[r 12 r 23 -r 13 r 22 r 13 r 21 -r 11 r 23 r 11 r 22 -r 12 r 21 ] T .

[0026] After obtaining the relative posture of the non-cooperative spacecraft, the relative position of the non-cooperative spacecraft is calculated by combining the binocular image data obtained by the binocular camera and the lidar with the radar point cloud data; the Harris feature points of the non-cooperative spacecraft are extracted from the binocular image, and the feature points of the non-cooperative spacecraft extracted from the binocular image are matched with each other, and the matched Harris feature points are marked as p lm =[u lm v lm 1] T and prm =[u rm v rm 1] T , and then use the disparity information of the binocular image to restore the 3D coordinates of the matching points:

[0027]

[0028] Where R′=R′ l =R′ r Represents the attitude rotation matrix of the binocular camera, t′ l ,t′ r Represent the spatial coordinates of the left and right cameras respectively. Use the above formula to solve the three-dimensional point set S corresponding to all binocular image matching points match ={[X Wm Y Wm Z Wm ] T ,m=0,…n match}, the calculated three-dimensional point set is combined with the radar point cloud set S to obtain a new three-dimensional point set S fusion =S∪S match Since the surface point cloud of the non-cooperative spacecraft close to the laser radar is dense, and the surface point cloud far away from the laser radar is sparse, different weights are assigned to each point according to different distances. The closer to the laser radar, the smaller the weight; the fusion point set S fusion The centroid position of is obtained by calculating the weighted average of the point coordinates and comparing it with the main body position. The centroid coordinates of the fused point set are written as:

[0029]

[0030] Among them, X c represents the centroid coordinates of the fused point set, n fusion represents the point set capacity, ρ i represents the weight of the i-th point, then the relative position of the non-cooperative spacecraft can be expressed as t compute =X c -X0, where X0 represents the body position.

[0031] Using the calculated R compute ,t compute Get the rough relative position of non-cooperative spacecraft.

[0032] Further, the specific process of step (3) is as follows:

[0033] Based on the rough relative pose of the non-cooperative spacecraft measured in step (2), it is used as an ideal initial value, and further iterative calculations are performed using the lidar point cloud data to improve the accuracy of the relative pose acquisition of the non-cooperative spacecraft. The lidar point cloud data is represented by tLi ,i=1,…,mThe data of the non-cooperative spacecraft point set is represented by x j , j = 1, ..., n, calculate the centroid position of the lidar point cloud data and the non-cooperative spacecraft point set and record them as t Lave ,x ave , and then calculate the centroid coordinates of each point in the two sets of data: q i =t Li -t Lave ,q′ j =x j -x ave , and then calculate the rotation matrix according to the following optimization problem:

[0034]

[0035] Among them, ||·|| represents the Euclidean norm of the orientation vector, and the optimized rotation matrix R is obtained by continuous iterative calculation. end , and finally calculate the corresponding displacement vector t end =t Lave -R end x ave .

[0036] The R obtained by the above method end and t end , and obtain the high-precision relative position and pose of non-cooperative spacecraft.

[0037] Beneficial effects of the present invention:

[0038] (1) In order to identify cooperative spacecraft in space, it is proposed to extract the line features of non-cooperative spacecraft in binocular images to measure the rough attitude information of non-cooperative spacecraft;

[0039] (2) Establishing sensor simulation models for binocular cameras and lidar provides a way to process information without physical sensors, and provides a basis for verifying the relative pose algorithm of non-cooperative spacecraft;

[0040] (3) In order to overcome the defects of insufficient measurement accuracy and weak robustness of a single sensor, the present invention establishes a data fusion scheme based on the data features of binocular images and lidar, and uses binocular camera and lidar data to calculate a relatively rough relative pose of the non-cooperative spacecraft. The lidar data is then further iteratively calculated to obtain the pose of the non-cooperative spacecraft with sub-millimeter accuracy. Using a single sensor to measure relative pose, the accuracy can only reach millimeter level. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a principle block diagram of the non-cooperative spacecraft relative posture measurement method based on multi-source information fusion of the present invention;

[0042] Figure 2 It is a flowchart of the non-cooperative spacecraft relative posture measurement method based on multi-source information fusion of the present invention;

[0043] Figure 3 A non-cooperative spacecraft model involved in the present invention;

[0044] Figure 4 A diagram defining the relevant coordinate system involved in the present invention;

[0045] Figure 5 Result diagram of relative position and attitude measurement error of non-cooperative spacecraft according to the method proposed in the present invention, (a) is the relative position error of non-cooperative spacecraft, and (b) is the relative attitude error of non-cooperative spacecraft. DETAILED DESCRIPTION

[0046] The present invention will be further described below in conjunction with the accompanying drawings.

[0047] like Figure 1 As shown, the sensor data fusion structure design of the present invention is aimed at the structural characteristics of the non-cooperative spacecraft, and comprehensively considers the data characteristics of the binocular camera and the lidar sensor, extracts the line features of the non-cooperative spacecraft in the binocular image to perform low-precision target relative posture solution, and then extracts the point features of the non-cooperative spacecraft in the binocular image, after restoring its depth information, it is combined with the radar point cloud data to obtain the low-precision relative position of the target, and finally uses the obtained relative posture of the non-cooperative spacecraft as the initial value, and uses the radar point cloud data for iterative calculation to further improve the measurement accuracy of the relative posture of the non-cooperative spacecraft.

[0048] like Figure 2 As shown, the non-cooperative spacecraft relative posture measurement method based on multi-source information fusion of the present invention comprises the following steps:

[0049] (1) Design a non-cooperative spacecraft model and build simulation models of the binocular camera and lidar;

[0050] like Figure 3 As shown in the figure, for the non-cooperative spacecraft to be identified in space, the simulation model is designed as a cuboid with a length of 0.2 cm, a width of 0.1 cm, and a height of 0.3 cm. No texture is added to the surface of the non-cooperative spacecraft, but 4 small cubes with a side length of 0.03 cm are added to one of the surfaces, and 3 small cubes with the same side length are added to the opposite surface. The overall structure is asymmetric, and the ratio of the model to the actual object is assumed to be 1:400. The non-cooperative spacecraft model is represented by a point set as:

[0051] X = {x′ i |x′ i =Rx i+t,i=1,2,…,n} (1)

[0052] Among them, x i =[x1 x2 x3] T and x′ i =[x′1 x′2 x′3] T They respectively represent the coordinates of the non-cooperative spacecraft in the world coordinate system before and after rotation and translation, R represents the attitude rotation matrix of the non-cooperative spacecraft, t represents the translation vector of the non-cooperative spacecraft in the world coordinate system, n represents the number of spatial points that constitute the non-cooperative spacecraft model; the superscript T represents the transposed matrix.

[0053] The constructed binocular camera simulation model is:

[0054]

[0055] Among them, [u l v l 1] T and[u r v r 1] T They represent the pixel coordinates of the non-cooperative spacecraft projected onto the left and right pixel planes of the binocular image, is the camera's internal parameter matrix, which represents the camera's internal parameters. 1 / dx and 1 / dy represent the camera's horizontal and vertical resolutions, respectively. f represents the camera's focal length. (c x ,c y ) represents the center position of the image plane, R′ l Represents the posture rotation matrix of the left camera in the world coordinate system, R′ r Represents the attitude rotation matrix of the right camera in the world coordinate system, Represents the attitude rotation matrix of the left camera relative to the right camera, t′ r Represents the spatial coordinates of the right camera in the world coordinate system, t′ l represents the spatial coordinates of the left camera in the world coordinate system, Δt′ represents the relative position of the left and right cameras; considering the designed non-cooperative spacecraft model parameters, the camera parameters are set to f = 0.161, dx = dy = 0.000014, c x =c y =512, R′ r =diag(1,1,1), R′ l =diag(1,1,1), t′ r =[0.24 0 0] T , t′ l =[0 0 0] T , Δt′=[0.24 0 0]T .like Figure 4 As shown in the figure, the optical center position and attitude of the left and right eye cameras are represented by O1-X1Y1Z1 and O3-X3Y3Z3, O-XYZ represents the coordinate system fixed to the service spacecraft body, o-xyz represents the coordinate system fixed to the non-cooperative spacecraft, and R,t represents the corresponding rotation and translation of the non-cooperative spacecraft.

[0056] The laser radar’s optical axis points to the positive z-axis direction of the world coordinate system. The simulation model of the non-cooperative spacecraft in the laser radar’s pointing area is:

[0057]

[0058] Where S = {t Lj ,j=1,…,m} represents the point cloud set of lidar imaging, t Lj =[x Lj y Lj z Lj ] T Represents the position data of the jth point cloud in the laser radar imaging data, t Wj =[x Wj y Wj z Wj ] T represents the spatial position of the i-th point in the point set data constituting the non-cooperative spacecraft, θ i =[θ 1i θ 2i ] T represents the pointing information of the i-th point constituting the non-cooperative spacecraft in the lidar coordinate system, θ 1i Characterizes the angle between the i-th target point and the vertical symmetry plane where the laser radar is installed, θ 2i Represents the angle between the i-th target point and the horizontal symmetry plane of the laser radar installation, t LiDAR =[x LiDAR y LiDAR z LiDAR ] T Represents the spatial coordinates of the laser radar in the world coordinate system. The simulated laser radar model has 64 lines, θ LiDAR ={[θ LiDAR1 θ LiDAR2 ] T |θ LiDAR1 ∈{k1dθ x ,k1=1,…,p},θ LiDAR2 ∈{k2dθ y ,k2=1,…,q}},dθ x =0.01° represents the horizontal scanning interval of the laser radar, dθ y=0.625° represents the vertical scanning interval of the laser radar. The horizontal and vertical fields of view of the simulated laser radar are fov x =120°,fov y =40°, p=fov x / dx=12000 indicates the number of horizontal scan lines of the laser radar, which is affected by the scanning interval and the horizontal field of view, q=fov y / dy=64 represents the number of vertical scanning lines of the laser radar, m represents the number of laser radar point clouds obtained, and n represents the number of spatial points that constitute the non-cooperative spacecraft. In order to fuse the acquired data from different sensors, the laser radar is installed in the middle of the binocular camera. The position and attitude of the laser radar are as follows: Figure 4 As shown in O2-X2Y2Z2.

[0059] (2) Based on the non-cooperative spacecraft model and sensor simulation model constructed in step (1), the advantages of binocular camera and lidar sensor are combined to design a sensor data fusion structure to measure the relative pose of non-cooperative spacecraft;

[0060] First, the binocular image and point cloud data of the non-cooperative spacecraft are obtained respectively through the binocular camera and lidar simulation models constructed in step (1), and the line features of the non-cooperative spacecraft in the binocular image are extracted, and three lines L1, L2, and L3 are found to satisfy L1⊥(L2||L3), where (·) indicates that the part in the brackets is parallel to the part outside the brackets. The positional relationship between the three lines projected into the image and the vanishing point are used to solve the attitude rotation matrix of the non-cooperative spacecraft:

[0061] R compute =[r1 r2 r1×r2] T (4)

[0062] Where r1 = K -1 v1, r2 = K -1 v2, v2 = Kr2 = l2 × l3, null(·) means to find the null space, l1, l2, l3 are the homogeneous coordinates of the three lines L1, L2, L3 respectively, K represents the camera projection matrix, r1×r2=[r 12 r 23 -r 13 r 22 r 13 r 21 -r 11 r 23 r 11 r 22 -r 12 r 21 ] T .

[0063] After obtaining the relative posture of the non-cooperative spacecraft, the relative position of the non-cooperative spacecraft is calculated by combining the binocular image data obtained by the binocular camera and the lidar with the radar point cloud data; the Harris feature points of the non-cooperative spacecraft are extracted from the binocular image, and the feature points of the non-cooperative spacecraft extracted from the binocular image are matched with each other, and the matched Harris feature points are marked as p lm =[u lm v lm 1] T and p rm =[u rm v rm 1] T , and then use the disparity information of the binocular image to restore the 3D coordinates of the matching points:

[0064]

[0065] Where R′=R′ l =R′ r Represents the attitude rotation matrix of the binocular camera, t′ l ,t′ r Represent the spatial coordinates of the left and right cameras respectively. Use the above formula to solve the three-dimensional point set S corresponding to all binocular image matching points match ={[X Wm Y Wm Z Wm ] T ,m=0,…n match}, the calculated three-dimensional point set is combined with the radar point cloud set S to obtain a new three-dimensional point set S fusion =S∪S match Since the surface point cloud of the non-cooperative spacecraft close to the laser radar is dense, and the surface point cloud far away from the laser radar is sparse, different weights are assigned to each point according to different distances. The closer to the laser radar, the smaller the weight; the fusion point set S fusion The centroid position of is obtained by calculating the weighted average of the point coordinates and comparing it with the main body position. The centroid coordinates of the fused point set are written as:

[0066]

[0067] Among them, X c represents the centroid coordinates of the fused point set, n fusion Represents the point set capacity, represents the weight of the i-th point, then the relative position of the non-cooperative spacecraft can be expressed as t compute =X c -X0, where X0 represents the body position.

[0068] Using the calculated R compute ,t compute , and obtain the rough relative position of the non-cooperative spacecraft.

[0069] (3) Based on the relative position and attitude of the non-cooperative spacecraft measured in step (2), iterative calculations are performed using lidar data to improve the measurement accuracy of the relative position and attitude of the non-cooperative spacecraft.

[0070] Based on the rough relative pose of the non-cooperative spacecraft measured in step (2), it is used as an ideal initial value, and further iterative calculations are performed using the lidar point cloud data to improve the accuracy of the relative pose acquisition of the non-cooperative spacecraft. The lidar point cloud data is represented by t Li ,i=1,…,mThe data of the non-cooperative spacecraft point set is represented by x j , j = 1, ..., n, calculate the centroid position of the lidar point cloud data and the non-cooperative spacecraft point set and record them as t Lave ,x ave , and then calculate the centroid coordinates of each point in the two sets of data: q i =t Li -t Lave ,q′ j =x j -x ave , and then calculate the rotation matrix according to the following optimization problem:

[0071]

[0072] Among them, ||·|| represents the Euclidean norm of the orientation vector, and the optimized rotation matrix R is obtained by continuous iterative calculation. end , and finally calculate the corresponding displacement vector t end =t Lave -R end x ave ;

[0073] The R obtained by the above method end and t end , and obtain the high-precision relative position and pose of the non-cooperative spacecraft.

[0074] like Figure 5 As shown, (a) is the relative position error of the non-cooperative spacecraft, and (b) is the relative attitude error of the non-cooperative spacecraft. The present invention has a high accuracy in measuring the relative position and attitude of the non-cooperative spacecraft, and the relative attitude measurement accuracy of the non-cooperative spacecraft is within 0.5°, and the relative position measurement accuracy of the non-cooperative spacecraft is within 0.1mm.

[0075] Through the above-mentioned solution based on binocular camera and lidar data fusion, and the measurement simulation of the relative posture of non-cooperative spacecraft, it can be known that the present invention realizes the high-precision calculation of the relative posture of non-cooperative spacecraft.

[0076] The contents not described in detail in the specification of the present invention belong to the prior art known to the professional and technical personnel in this field.

[0077] The above embodiments are only intended to illustrate the main implementation methods of the invention, and cannot be used to limit the protection scope of the invention. All technical solutions and modifications based on the invention fall within the protection scope of the claims of the invention.

Claims

1. A non-cooperative spacecraft relative posture measurement method based on multi-source information fusion, characterized in that: The following steps are involved: (1) Design a non-cooperative spacecraft model and build simulation models of the binocular camera and lidar; (2) Based on the simulation model described in step (1), the binocular image and point cloud data of the non-cooperative spacecraft are obtained through the simulation models of the binocular camera and the lidar, and an information fusion scheme of the binocular image and the point cloud data is designed to measure the relative pose of the non-cooperative spacecraft, thereby obtaining a rough relative pose of the non-cooperative spacecraft; (3) Based on the relative position and posture of the non-cooperative spacecraft in step (2), iterative calculation is performed using the point cloud data to obtain a high-precision relative position and posture of the non-cooperative spacecraft, thereby completing the relative position and posture measurement of the non-cooperative spacecraft; In step (2), the designed binocular camera and lidar data fusion scheme is: First, the binocular image and point cloud data of the non-cooperative spacecraft designed in step (1) are obtained through the binocular camera and lidar simulation model constructed in step (1), and the line features of the non-cooperative spacecraft in the binocular image are extracted to find three lines satisfy ,in Indicates that the part in the brackets is parallel to the part outside the brackets. The attitude rotation matrix of the non-cooperative spacecraft is solved as follows: (4) in, , , , represents the search for the null space, There are three lines respectively The homogeneous coordinates of ; is the camera’s intrinsic parameter matrix, representing the camera’s internal parameters. Respectively represent the horizontal and vertical resolution of the camera imaging, represents the focal length of the camera, Indicates the center position of the image plane; Then, the Harris feature points of the non-cooperative spacecraft are extracted from the binocular image, and the feature points of the non-cooperative spacecraft extracted from the binocular image are matched with each other, and the matched Harris feature points are marked as and , and then use the disparity information of the binocular image to restore the 3D coordinates of the matching points: (5) in, Represents the attitude rotation matrix of the binocular camera, Represent the spatial coordinates of the left and right cameras respectively. Use equation (5) to solve the 3D point set corresponding to all binocular image matching points , the calculated three-dimensional point set and the radar point cloud set The new three-dimensional point set is obtained by combining , fusion point set The centroid position of is obtained by calculating the weighted average of the point coordinates and comparing it with the main body position. The centroid coordinates of the fused point set are written as: (6) in, represents the centroid coordinates of the fused point set, Represents the point set capacity, Indicates The weights of the points, then the relative positions of the non-cooperative spacecraft are expressed as ,in Indicates the body position; Finally, using the calculated , and obtain the rough relative position of the non-cooperative spacecraft.

2. The non-cooperative spacecraft relative posture measurement method based on multi-source information fusion according to claim 1 is characterized in that: In step (1), the non-cooperative spacecraft model is: (1) in, and They represent the coordinates of the non-cooperative spacecraft in the world coordinate system before and after rotation and translation, represents the non-cooperative spacecraft attitude rotation matrix, represents the translation vector of the non-cooperative spacecraft in the world coordinate system, represents the number of spatial points constituting the non-cooperative spacecraft model; superscript represents the transposed matrix; The constructed binocular camera simulation model is: (2) in, and They represent the pixel coordinates of the non-cooperative spacecraft projected onto the left and right pixel planes of the binocular image, is the camera’s intrinsic parameter matrix, representing the camera’s internal parameters. Respectively represent the horizontal and vertical resolution of the camera imaging, represents the focal length of the camera, represents the center position of the image plane, Represents the posture rotation matrix of the left camera in the world coordinate system, Represents the attitude rotation matrix of the right camera in the world coordinate system, Represents the attitude rotation matrix of the left camera relative to the right camera, Represents the spatial coordinates of the right camera in the world coordinate system. Represents the spatial coordinates of the left camera in the world coordinate system. Indicates the relative position of the left and right cameras; The constructed laser radar simulation model is: The laser radar’s optical axis points to the world coordinate system. In the positive direction of the laser radar axis, the simulation model of the non-cooperative spacecraft in the laser radar pointing area is: (3) in, represents a collection of point clouds imaged by LiDAR, Represents the first The location data of the point cloud, Indicates the point set data that constitutes the non-cooperative spacecraft The spatial position of a point, The first The pointing information of a point in the laser radar coordinate system, , Respectively characterize the The angle between the target point and the vertical symmetry plane and horizontal symmetry plane of the laser radar installation, represents the spatial coordinates of the laser radar in the world coordinate system, , represents the horizontal scanning interval of the laser radar, represents the vertical scanning interval of the laser radar, Indicates the number of horizontal scan lines of the laser radar, which is affected by the scanning interval and the horizontal field of view. Indicates the number of vertical scan lines of the laser radar, Indicates the number of LiDAR point clouds obtained, Represents the number of space points that constitute a non-cooperative spacecraft.

3. The non-cooperative spacecraft relative posture measurement method based on multi-source information fusion according to claim 1 is characterized in that: The specific implementation of step (3) is as follows: Based on the rough relative pose of the non-cooperative spacecraft measured in step (2), it is used as an ideal initial value and further iterative calculations are performed using the lidar point cloud data to improve the accuracy of the relative pose acquisition of the non-cooperative spacecraft. The lidar point cloud data is expressed as , The data representation of the non-cooperative spacecraft point set is , , calculate the centroid positions of the two sets of point sets of the lidar point cloud data and the non-cooperative spacecraft point set and record them as , and then calculate the centroid coordinates of each point in the two sets of data: , and then calculate the rotation matrix according to the following optimization problem: (7) in, Represents the Euclidean norm of the orientation vector, and continuously iterates to obtain the optimized rotation matrix , and finally calculate the corresponding displacement vector ; represents the non-cooperative spacecraft attitude rotation matrix; The solution obtained by the above method and , and obtain the high-precision relative position and pose of the non-cooperative spacecraft.

Images