A method for extracting association information of flexible attachment nodes of small celestial bodies

Abstract

The present invention discloses a method for extracting node-related observation information, which belongs to the field of deep space exploration technology. The implementation method of the present invention is as follows: according to the characteristics that the relative pointing information between nodes and the relative position vector information between nodes satisfy the multi-view geometric constraints between nodes, the opportunity features in the common observation area of ​​the node optical navigation camera are matched, and the relative pointing information between nodes is solved by the epipolar constraint equations of multiple groups of features; the morphological features with known absolute positions are searched in the common observation area, and if such features exist, the laser rangefinder information is combined to obtain the relative position vector information between nodes by optical measurement geometric relationship; an evaluation function for selecting node-related information is constructed based on the Sampson distance, and information with a smaller Sampson distance is selected as the node-related information according to the evaluation function, so as to provide accurate node-related information for flexible attachment state collaborative navigation. The node state is corrected according to the accurate node-related information to improve the accuracy of flexible attachment state collaborative navigation.

Description

Technical Field

[0001] The invention relates to a method for extracting association information of flexible attachment nodes of a small celestial body, belonging to the technical field of deep space exploration. Background Art

[0002] The detection of small celestial body attachment has important scientific value in answering questions such as the origin and evolution of the solar system and the origin and evolution of life. It is also an effective means to develop and verify new space technologies. Considering that the gravitational field of small celestial bodies is weak and irregular, the environmental disturbance is complex, and the surface morphology is rugged, the rigid landing mode is likely to cause the lander to overturn and rebound on the surface of the small celestial body, resulting in mission failure. Therefore, it is urgent to develop and create a flexible attachment mode for small celestial bodies, enhance the adaptability of the lander to complex environments and the tolerance to landing state deviations, and achieve highly reliable and highly adaptable attachment to the surface of small celestial bodies. However, the dynamic characteristics of flexible landers are complex, and it is difficult to design an autonomous navigation and control system for its attachment process. In the previous technology, an approximate model of a flexible lander was established on the premise of retaining the flexible characteristics. The mass aggregation area with specific functions and rigid characteristics on the flexible lander was called a node, and a multi-node collaborative navigation scheme was designed to realize the flexible attachment state estimation. In the above-mentioned flexible attachment collaborative navigation scheme, the node association information is used to correct the node state, improve the node state estimation accuracy of the flexible attachment process, and thus improve the estimation accuracy of the position, velocity and attitude of the flexible lander. For this flexible attachment collaborative navigation scheme, it is necessary to develop a node correlation information extraction method to provide accurate inter-node correlation information and provide a feasible technical approach for multi-node collaborative navigation of flexible landers. Summary of the invention

[0003] Aiming at the problem of multi-node collaborative navigation of small celestial bodies with flexible attachment, the main purpose of the present invention is to provide a method for extracting node-related observation information, based on the morphological features in the common observation area of ​​the node optical navigation camera, extracting the relative pointing information between nodes or the relative position vector information between nodes, constructing an evaluation function for selecting node-related information based on the Sampson distance, selecting information with a smaller Sampson distance as the node-related information according to the evaluation function, and providing accurate node-related information for collaborative navigation of flexible attachment states. The node state is corrected according to the accurate node-related information to improve the accuracy of collaborative navigation of flexible attachment states.

[0004] The purpose of the present invention is achieved through the following technical solutions.

[0005] The present invention discloses a method for extracting node-related observation information, which matches the opportunity features in the common observation area of ​​the node optical navigation camera, and solves the relative pointing information between nodes by the epipolar constraint equations of multiple sets of features; searches for morphological features with known absolute positions in the common observation area, and if such features exist, combines the laser rangefinder information to obtain the relative position vector information between nodes by optical measurement geometric relationship; constructs an evaluation function for selecting node-related information based on the Sampson distance, and selects the information with a smaller Sampson distance as the node-related information according to the evaluation function, so as to provide accurate node-related information for flexible attachment state collaborative navigation. The node state is corrected according to the accurate node-related information, so as to improve the accuracy of flexible attachment state collaborative navigation.

[0006] The method for extracting association information of small celestial body flexible attachment nodes disclosed in the present invention comprises the following steps:

[0007] Step 1: Match the opportunity features in the common observation area of ​​the node optical navigation camera, and jointly establish the epipolar constraint equations of multiple sets of features to obtain the relative pointing information between nodes.

[0008] Match the opportunity characteristics of the common observation area of ​​the node cameras and establish the epipolar constraint equation

[0009]

[0010] In the above formula, E ij is the essential matrix, and is the pixel image line of the opportunity feature P under the camera system of node j and node i, K j and K i is the camera intrinsic parameter matrix.

[0011] The opportunity features include corner points, SIFT feature points, and SURF feature points.

[0012] In order to extract information quickly, equation (1) is transformed into a linear equation and obtained

[0013] A P e ij =0 (2)

[0014] Among them, e ij =[e1…e9] T is the essential matrix E ij The elements in A P for and The vector formed by the elements in is in the form of

[0015]

[0016] Combine multiple sets of characteristic epipolar constraint equations to solve the essential matrix. Perform SVD decomposition to obtain relative pointing information between nodes Its form is

[0017]

[0018] in, is the unit relative position vector from node i to node j in the attachment point coordinate system, that is, the relative orientation between nodes.

[0019] Step 2: Search for morphological features with known absolute positions in the common observation area. If such features exist, combine the laser rangefinder information to obtain the relative position vector information between nodes by optically measuring the geometric relationship, which can reduce the amount of calculation required to obtain the relative position vector information between nodes.

[0020] Search for morphological features with known absolute positions in the common observation area. When there is a feature L with known absolute positions in the common observation area, the laser rangefinder information is combined to obtain the geometric relationship of optical measurement.

[0021]

[0022]

[0023]

[0024] in, and The pixel image line of feature L in the images of node 1 and node 2, f is the focal length of the node camera, R 1 and R 2 is the attitude transformation matrix from the node camera system to the fixed connection system of the attachment point, r L,z is the height of feature L when it is fixed at the attachment point, r 1,z and r 2,z is the height of node 1 and node 2 when they are fixedly connected at the attachment point, measured by a laser rangefinder. Thus, the relative position vector information between the nodes is obtained. Its form is

[0025]

[0026] in, is the relative position from node i to node j in the attachment point coordinate system.

[0027] Step three: According to the relative pointing information between nodes and the relative position vector information between nodes obtained in steps one and two, an evaluation function for selecting node association information is constructed based on the Sampson distance. According to the evaluation function, information with a smaller Sampson distance is selected as the node association information for the flexible lander multi-node collaborative navigation, providing accurate node association information for the flexible attachment state collaborative navigation.

[0028] Remember R ij is the attitude transformation matrix from the camera system of node i to the camera system of node j, is the relative orientation information between nodes and the relative position vector information between nodes obtained in step 1 and step 2. When the information is extracted accurately, the feature m observed by node j j (including the chance features in step 1 and the morphological features with known absolute positions in step 2) should be located on their corresponding polar lines Therefore, the distance from the feature to its corresponding epipolar line can represent the accuracy of information extraction. Sampson distance is an indicator to measure the distance from the feature to its corresponding epipolar line, which is defined as

[0029] d Sampson =g(JJ T ) -1 g (9)

[0030] in,

[0031] g=(m j ) T l j (10)

[0032]

[0033] Comprehensively combining all the matching features in the common observation area, the evaluation function for selecting node association information is constructed as follows:

[0034]

[0035] Among them, ι is the label of the matchable feature.

[0036] Evaluate the two types of information through formula (12) and select The smaller information serves as the node association information for the flexible lander's multi-node collaborative navigation, providing accurate node association information for the flexible attachment state collaborative navigation.

[0037] The method further includes step 4: correcting the node state according to the node association information obtained in step 3 to improve the accuracy of the flexible attachment state collaborative navigation.

[0038] Considering the flexible attachment state constraint, the least squares estimation problem of the flexible lander node state is:

[0039]

[0040] in, is the estimated result after correction at node i, is the independent estimation result of node i state, is the independent estimation error covariance matrix of node i state, l ij is the nominal relative distance between node i and node j, Δl ij is the maximum deformation between the nodes of the flexible body, r 1 ,r 2 ,r 3 is the position of node 1, node 2 and node 3, r 12 =r 2 -r 1 ,r 13 =r 3 -r 1 is the relative position vector between nodes, n L =[0,0,1] T is the landing plane normal vector.

[0041] Introducing node association information into the above-mentioned least squares problem of node state estimation, when the relative pointing information between nodes is used as node association information, the index augmentation J' of the least squares problem of node state estimation is:

[0042]

[0043] Among them, H r =[I3 0 3×3 ], is the weight.

[0044] When the relative position vector information between nodes is used as the node association information, the index augmentation J" of the node state least squares estimation problem is

[0045]

[0046] in, is the weight.

[0047] Beneficial effects:

[0048] 1. The present invention discloses a method for extracting node association observation information. According to the characteristics that the relative pointing information between nodes and the relative position vector information between nodes satisfy the multi-view geometric constraints between nodes, an evaluation function for selecting node association information is constructed based on the Sampson distance. According to the evaluation function, information with a smaller Sampson distance is selected as the node association information for the flexible lander multi-node collaborative navigation, thereby providing accurate node association information for the flexible attachment state collaborative navigation.

[0049] 2. The present invention discloses a method for extracting node-related observation information, which obtains relative position vector information between nodes by optically measuring geometric relationships, can reduce the amount of calculation for obtaining relative position vector information between nodes, thereby improving the real-time performance of obtaining node-related information.

[0050] 3. A node association observation information extraction method disclosed in the present invention, on the basis of achieving beneficial effects 1 and 2, corrects the node state according to the acquired accurate node association information, thereby improving the accuracy of flexible attachment state collaborative navigation. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of the method for extracting the association information of the flexible attachment nodes of small celestial bodies disclosed in the present invention;

[0052] Figure 2 Optical measurement of geometrical relationships of topographical features with known absolute positions;

[0053] Figure 3 is the relative pointing information extraction error between nodes for 50 simulations;

[0054] Figure 4 The relative position vector information extraction error between nodes for 50 simulations;

[0055] Figure 5 are the two types of information evaluation index values ​​for 50 simulations;

[0056] Figure 6 The node association information selection results for 50 simulations. DETAILED DESCRIPTION

[0057] In order to better illustrate the purpose and advantages of the present invention, the invention is further described below with reference to the accompanying drawings and examples.

[0058] Embodiment 1:

[0059] In order to verify the feasibility of the method, the surface attachment autonomous navigation simulation calculation is carried out by taking the attachment mission for small celestial bodies as an example. The diameter of the flexible lander is 3m, the number of measurement nodes is 3, the three measurement nodes are distributed at 120° in the nominal state, the distance between the node and the center of the "equivalent surface" of the flexible lander is 0.75m, and the position of node 1 at a certain moment in the attachment process is [-4.25, 5, 50]m, the position of node 2 is [-5.38, 5.65, 50]m, and the position of node 3 is [-5.38, 4.35, 50]m. The parameters of the sensors of each node are consistent, the camera field of view is 40°, the camera focal length is 0.02m, the imaging resolution is 1024×1024, the observation errors of each sensor are Gaussian white noise, the standard deviation of the error of opportunity feature extraction is 0.1 pixel, the standard deviation of the error of morphological feature extraction with known absolute position is 1 pixel, and the standard deviation of the error of the laser rangefinder is 0.05m. Without loss of generality, it is assumed that the coordinate system of each node sensor coincides with the node itself, and the posture of the node in the attachment point coordinate system is [1°, 1°, 1°].

[0060] like Figure 1 As shown, the method for extracting the associated information of the small celestial body flexible attachment nodes disclosed in this embodiment has the following specific implementation steps:

[0061] Step 1: Match the opportunity features in the common observation area of ​​the node optical navigation camera, and jointly establish the epipolar constraint equations of multiple sets of features to obtain the relative pointing information between nodes.

[0062] Match the opportunity features of the common observation area of ​​the node camera, such as corner points, SIFT feature points, SURF feature points, etc. Assuming that there are 8 pairs of matchable opportunity features in the common observation area of ​​node 1 and node 2, the epipolar constraint equation is established

[0063]

[0064] In the above formula, E 12 is the essential matrix, and is the pixel image line of the opportunity feature P in the camera system of node 2 and node 1, K1 and K2 are the camera internal parameter matrices, and their specific forms are:

[0065]

[0066] Combine multiple sets of characteristic epipolar constraint equations to solve the essential matrix. Perform SVD decomposition to obtain relative pointing information between nodes Its form is

[0067]

[0068] in, It is the unit relative position vector from node 1 to node 2 in the attachment point coordinate system, that is, the relative orientation between nodes.

[0069] Step 2: Search for morphological features with known absolute positions in the common observation area. If such features exist, combine the laser rangefinder information to obtain the relative position vector information between nodes by optically measuring the geometric relationship, which can reduce the amount of calculation required to obtain the relative position vector information between nodes.

[0070] Search for morphological features with known absolute positions in the common observation area. Assume that there is a morphological feature with known absolute positions in the common observation area of ​​nodes 1 and 2, such as Figure 2 As shown, combined with the laser rangefinder information, the optical measurement geometry is obtained

[0071]

[0072]

[0073]

[0074] in, and The pixel image line of feature L in the images of node 1 and node 2, f is the focal length of the node camera, R 1 and R 2 is the attitude transformation matrix from the node camera system to the fixed connection system of the attachment point, r L,z is the height of feature L when it is fixed at the attachment point, r 1,z and r 2,z is the height of node 1 and node 2 when they are fixedly connected at the attachment point, measured by a laser rangefinder. Thus, the relative position vector information between the nodes is obtained. Its form is

[0075]

[0076] in, is the relative position from node 1 to node 2 in the attachment point coordinate system.

[0077] Step three: According to the relative pointing information between nodes and the relative position vector information between nodes obtained in steps one and two, an evaluation function for selecting node association information is constructed based on the Sampson distance. According to the evaluation function, information with a smaller Sampson distance is selected as the node association information for the flexible lander multi-node collaborative navigation, providing accurate node association information for the flexible attachment state collaborative navigation.

[0078] The simulation scenario was simulated 50 times, and the relative pointing information extraction error between nodes was as follows: Figure 3As shown in the figure, the relative position vector information extraction error between nodes is as follows: Figure 4 As shown. 12 is the attitude transformation matrix from the camera system of node 1 to the camera system of node 2, n=1,2 is the relative pointing information between nodes and the relative position vector information between nodes obtained in step 1 and step 2, The Sampson distance is used to measure the distance from a feature to its corresponding epipolar line. By integrating all the matching features in the common observation area, the information evaluation function is constructed as follows:

[0079]

[0080] Among them, ι is the label of the matching feature, n=1,2 is the relative orientation information and relative position vector information between nodes obtained in step 1 and step 2, (·) [1] and(·) [2] Represents the first and second elements in a vector.

[0081] The two types of information are evaluated by formula (20). The evaluation index values ​​of the two types of information are as follows: Figure 5 As shown, the red line represents the evaluation index value of the relative pointing information between nodes, and the blue line represents the evaluation index value of the relative position vector information between nodes.

[0082] Select The smaller information is used as the node association information of the flexible lander multi-node collaborative navigation scheme, and the selection results are as follows Figure 6 As shown, the relative pointing information between nodes is selected as the node association information 34 times, the relative position vector information between nodes is selected as the node association information 16 times, and the node association information is used to collaboratively correct the node status, providing a feasible technical approach for multi-node collaborative navigation of flexible landers.

[0083] Step 4: Correct the node status according to the node association information obtained in step 3 to improve the accuracy of flexible attachment state collaborative navigation.

[0084] Considering the flexible attachment state constraint, the least squares estimation problem of the flexible lander node state is:

[0085]

[0086] in, is the estimated result after correction at node i, is the independent estimation result of node i state, is the independent estimation error covariance matrix of node i state, l ij is the nominal relative distance between node i and node j, Δl ijis the maximum deformation between the nodes of the flexible body, r 1 ,r 2 ,r 3 is the position of node 1, node 2 and node 3, r 12 =r 2 -r 1 ,r 13 =r 3 -r 1 is the relative position vector between nodes, n L =[0,0,1] T is the landing plane normal vector.

[0087] Introducing node association information into the above-mentioned least squares problem of node state estimation, when the relative pointing information between nodes is used as node association information, the index augmentation J' of the least squares problem of node state estimation is:

[0088]

[0089] Among them, H r =[I3 0 3×3 ], is the weight.

[0090] When the relative position vector information between nodes is used as the node association information, the index augmentation J" of the node state least squares estimation problem is

[0091]

[0092] in, is the weight.

[0093] The specific description above further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for extracting association information of flexible attachment nodes of small celestial bodies, characterized by: The following steps are included: Step 1: Match the opportunity features in the common observation area of ​​the optical navigation cameras of the nodes, and jointly establish the epipolar constraint equations of multiple sets of features to obtain the relative pointing information between the nodes; Step 2: Search for morphological features with known absolute positions in the common observation area. If such features exist, combine the laser rangefinder information to obtain the relative position vector information between nodes by optically measuring the geometric relationship, which can reduce the amount of calculation for obtaining the relative position vector information between nodes. Step 3: According to the relative pointing information between nodes and the relative position vector information between nodes obtained in Step 1 and Step 2, an evaluation function for selecting node association information is constructed based on the Sampson distance, and information with a smaller Sampson distance is selected as the node association information for the flexible lander multi-node collaborative navigation according to the evaluation function, so as to provide accurate node association information for the flexible attachment state collaborative navigation; The implementation method of step three is: Remember R ij is the attitude transformation matrix from the camera system of node i to the camera system of node j, n=1,2 is the relative pointing information and relative position vector information between nodes obtained in step 1 and step 2; when the information is extracted accurately, the feature m observed by node j j The chance features in step 1 and the morphological features whose absolute positions are known in step 2 should be located on their corresponding polar lines. Therefore, the distance from the feature to its corresponding epipolar line can characterize the accuracy of information extraction; Sampson distance is an indicator to measure the distance from the feature to its corresponding epipolar line, which is defined as <h2 style=";text-align:left;direction:ltr">d<h2 style=";text-align:left;direction:ltr"> Sampson <h2 style=";text-align:left;direction:ltr"> (g(JJ<h2 style=";text-align:left;direction:ltr"> T <h2 style=";text-align:left;direction:ltr"> )<h2 style=";text-align:left;direction:ltr"> -1 <h2 style=";text-align:left;direction:ltr"> g (9) in, g=(m j ) T l j (10) Comprehensively combining all the matching features in the common observation area, the evaluation function for selecting node association information is constructed as follows: Where ι is the number of the matching feature; Evaluate the two types of information through formula (12) and select The small information is used as the node association information for the flexible lander's multi-node collaborative navigation, providing accurate node association information for the flexible attachment state collaborative navigation.

2. The method for extracting association information of small celestial body flexible attachment nodes according to claim 1, characterized in that: The method further includes step 4, wherein the node state is corrected according to the node association information obtained in step 3, so as to improve the accuracy of the flexible attachment state collaborative navigation.

3. The method for extracting association information of small celestial body flexible attachment nodes according to claim 1 or 2, characterized in that: The implementation method of step one is: Match the opportunity characteristics of the common observation area of ​​the node cameras and establish the epipolar constraint equation In the above formula, E ij is the essential matrix, and is the pixel image line of the opportunity feature P under the camera system of node j and node i, K j and K i is the camera intrinsic parameter matrix; The opportunity features include corner points, SIFT feature points, and SURF feature points; In order to extract information quickly, equation (1) is transformed into a linear equation and obtained TO P And ij =0 (2) Among them, e ij =[e1…e9] T is the essential matrix E ij The elements in A P for and The vector formed by the elements in is in the form of Combine the epipolar constraint equations of multiple sets of characteristics to solve the essential matrix; Perform SVD decomposition to obtain relative pointing information between nodes Its form is in, is the unit relative position vector from node i to node j in the attachment point coordinate system, that is, the relative orientation between nodes.

4. The method for extracting association information of small celestial body flexible attachment nodes as claimed in claim 3, characterized in that: The implementation method of step 2 is: Search for morphological features with known absolute positions in the common observation area; when there is a feature L with known absolute positions in the common observation area, combine the laser rangefinder information to obtain the geometric relationship of optical measurement in, and The pixel image line of feature L in the images of node 1 and node 2, f is the focal length of the node camera, R 1 and R 2 is the attitude transformation matrix from the node camera system to the fixed connection system of the attachment point, r L,z is the height of feature L when it is fixed at the attachment point, r 1,z and r 2,z is the height of node 1 and node 2 when they are fixedly connected at the attachment point, measured by a laser rangefinder; thus, the relative position vector information between the nodes is obtained Its form is in, is the relative position from node i to node j in the attachment point coordinate system.

5. The method for extracting association information of small celestial body flexible attachment nodes as claimed in claim 4, characterized in that: The implementation method of step 4 is: Considering the flexible attachment state constraint, the least squares estimation problem of the flexible lander node state is: in, is the estimated result after correction at node i, is the independent estimation result of node i state, is the independent estimation error covariance matrix of node i state, l ij is the nominal relative distance between node i and node j, Δl ij is the maximum deformation between the nodes of the flexible body, r 1 ,r 2 ,r 3 is the position of node 1, node 2 and node 3, r 12 =r 2 -r 1 ,r 13 =r 3 -r 1 is the relative position vector between nodes, n L =[0,0,1] T is the landing plane normal vector; Introducing node association information into the above-mentioned least squares problem of node state estimation, when the relative pointing information between nodes is used as node association information, the index augmentation J' of the least squares problem of node state estimation is: Among them, H r =[I3 0 3×3 ], is the weight; When the relative position vector information between nodes is used as the node association information, the index augmentation J" of the node state least squares estimation problem is in, is the weight.

Images