An ESP-based modeling and solving method for flight trajectory adjustment problem of spacecraft
By modeling the composition adjustment problem of the aerial camera as an Euclidean shortest path problem, the shortest path of the aerial camera is directly solved, which solves the problem that it is difficult to obtain an exact solution in the existing technology and realizes efficient determination of the target position of the aerial camera.
Patent Information
- Application Number
- CN202310239755.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-03-13
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2043-03-13
AI Technical Summary
Existing technologies cannot effectively provide accurate closed-form solutions for the target position of flying cameras, resulting in time-consuming or multiple solutions during the composition adjustment process, which causes problems for practical applications.
The problem of adjusting the composition of the flying camera is modeled as a Euclidean shortest path problem. Through geometric analysis and region partitioning, the shortest path from the flying camera to the target torus is directly solved, avoiding obstacles and providing an accurate closed-form solution.
This method achieves an accurate solution for the target position of a flying camera in three-dimensional space without iteration, improving the efficiency and accuracy of composition adjustment and avoiding the problems of long computation time and multiple solutions in existing methods.
Smart Images

Figure CN116595707B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of path planning technology, and more specifically, relates to a modeling and solution method for the flight phase diagram adjustment problem based on ESP. Background Technology
[0002] Automatically adjusting the position of a flying camera so that a real-world object of interest is imaged at a specific location in the photograph is a complex and challenging problem. For ease of explanation, this problem will be referred to simply as the "composition adjustment" problem.
[0003] Considering the complexity of the image adjustment problem and the nonlinearity of the camera projection equations, almost all existing methods model the original problem as a nonlinear optimization problem. They embed the target position of the flying camera into the objective or constraint functions of the optimization problem, thus physically expressing the original problem as a search problem in the solution space of the target position. Therefore, existing methods all revolve around how to improve the search algorithm for the solution space.
[0004] To date, there is no universal method to completely solve the composition adjustment problem: these existing methods either provide an approximate solution for the target position through iterative algorithms (which often consume a lot of computation time) or provide a range solution for the target position through non-iterative algorithms (multiple solutions bring trouble to practical applications), and neither can effectively provide an exact closed-form solution for the target position. Summary of the Invention
[0005] In view of the above-mentioned defects or improvement needs of the existing technology, the present invention provides a modeling and solution method for the flight camera pattern adjustment problem based on ESP, the purpose of which is to obtain an accurate closed-form solution of the target position of the flight camera.
[0006] To achieve the above objectives, according to one aspect of the present invention, a method for modeling the flight phase diagram adjustment problem based on ESP is provided, comprising:
[0007] S1. Determine the three-dimensional coordinates (Q) of the center of the spherical obstacle i in the world coordinate system. i Sphere radius r i The initial position C0 of the flight camera, and the target torus Ω determined by Q1, Q2, and ω; where ω is the initial position of the flight camera at the target position C0. * Object of interest Q after adjusting posture i The image can fall at the desired position P in the image. i The expected line-of-sight angle of the flight camera at that time; i = 1, 2;
[0008] S2. In a three-dimensional Euclidean space with two spherical obstacles i, find the shortest path from the initial position C0 to the target torus Ω, which does not intersect with the interior of any spherical obstacle; the endpoint of the shortest path is located on the intersection arc A of the half-plane ∏ determined by the three points Q1, Q2 and C0 and the target torus Ω.
[0009] The present invention also provides a solution method based on the above modeling method, comprising:
[0010] S1. Determine the area where the initial position C0 of the flight camera is located:
[0011] If C0 is in ray O A P aits,1 Take O A Rotate along arc A to ray O. A P aits,2 The area traversed, minus the tangent τ i , X-rays O A P aits,1 O A P aits,2 And the region formed after the area enclosed by the obstacle circle i, then C0 is determined to be in region 0; O A Let τ be the center of arc A. i For the arc center O A To ⊙Q i tangent, ⊙Q i Let spherical obstacle i be intersected by a circle by a half-plane π; P aits,i Let's consider arc A and circle Q. i The intersection;
[0012] If C0 is not in region 0 and is located inside arc A, then C0 is determined to be in region 1.
[0013] If C0 is not in region 0, but is located outside of arc A and close to obstacle circle 1, then determine that C0 is in region 2-1;
[0014] If C0 is not in region 0, but is located outside of arc A and close to obstacle circle 2, then determine that C0 is in region 2-2.
[0015] S2. Calculate the shortest path based on the region where C0 is located:
[0016] When the initial position C0 of the flight camera is in region 0, ray O will be... A The intersection of C0 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0C. * ;
[0017] When the initial position C0 of the flying camera is in region 1, if tangent τ1 exists and tangent τ2 does not exist, the intersection of tangent τ1 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0P. ctgt,1 arc segment P ctgt,1 P otgt,1 and line segment P otgt,1 C * The shortest path is formed by connecting the points sequentially. If tangent τ1 does not exist, but tangent τ2 does, the intersection of tangent τ2 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0P. ctgt,2 arc segment P ctgt,2 P otgt,2 and line segment P otgt,2 C * It is formed by connecting them sequentially; if the tangent τ i Both exist; the smaller of the two paths is taken as the shortest path; P otgt,i Tangent τ i The point of tangency with the obstacle circle i, P ctgt,1 P is the point of tangency from C0 to ⊙Q1; ctgt,2 Let C0 be the point of tangency to ⊙Q2;
[0018] When the initial position C0 of the flight camera is in region 2-i, the intersection of arc A and obstacle circle i is taken as the endpoint C of the shortest path. * The corresponding shortest path is: if line segment C0C * Not with ⊙Q i If they intersect, the shortest path is the straight line segment C0C. * If line segment C0C * With ⊙Q i Intersecting lines, the shortest path is formed by the straight line segment C0P. ctgt,i and arc segment P ctgt,i C * It is formed by connecting the links.
[0019] Furthermore, step S1 specifically includes,
[0020] 01. Construct an auxiliary coordinate system F A :
[0021] Find the center O of the arc A of the intersection of the half-plane ∏ and the target torus Ω. A As F A The origin;
[0022] Let F be the unit direction vector pointing from Q1 to Q2. A The unit direction vector pointing positively upwards along the x-axis;
[0023] Let F be the unit direction vector of the perpendicular bisector of line segment Q1Q2 in the half-plane ∏.A The unit direction vector pointing positively upwards along the y-axis;
[0024] Based on the unit direction vectors pointing positively upwards along the x and y axes, F is determined according to the right-hand rule. A The unit direction vector pointing positively upwards along the z-axis;
[0025] 02. Calculate the initial positions C0 and P of the flight camera. aits,i P otgt,i Q i In the half-plane Π, according to xO A The polar coordinates corresponding to the y-coordinate system; where P otgt,i Tangent τ i The point of tangency with the obstacle circle i;
[0026] When C0 satisfies Or simultaneously satisfy and Or simultaneously satisfy and When C0 is in region 0, determine that the region is region 0.
[0027] When C0 is not in region 0 and is inside the torus, i.e. When C0 is in region 1, determine that the region is region 1.
[0028] When C0 is not in region 0 and is outside the torus, i.e. At that time, according to Determine that the region where C0 is located is region 2-i, i = 1, 2.
[0029] This invention also provides a modeling system for flight phase diagram adjustment problems based on ESP, comprising:
[0030] The spatial structure information loading module is used to determine the three-dimensional coordinates Q of the center of the spherical obstacle i in the world coordinate system. i Sphere radius r i The initial position C0 of the flight camera, and the target torus Ω determined by Q1, Q2, and ω; where ω is the initial position of the flight camera at the target position C0. * Object of interest Q after adjusting posture i The image can fall at the desired position P in the image. i The expected line-of-sight angle of the flight camera at that time; i = 1, 2;
[0031] The ESP model building module is used to solve for the shortest path from the initial position C0 of the flying camera to the target torus Ω in a three-dimensional Euclidean space with two spherical obstacles i, without intersecting the interior of any spherical obstacle; the endpoint of the shortest path is located on the intersection arc A of the half-plane ∏ determined by the three points Q1, Q2 and C0 and the target torus Ω.
[0032] The present invention also provides a solution system based on the above-described modeling system, comprising:
[0033] The region determination module is used to determine the region where the initial position C0 of the flight camera is located.
[0034] If C0 is in ray O A P aits,1 Take O A Rotate along arc A to ray O. A P aits,2 The area traversed, minus the tangent τ i , X-rays O A P aits,1 O A P aits,2 And the region formed after the area enclosed by the obstacle circle i, then C0 is determined to be in region 0; O A Let τ be the center of arc A. i For the arc center O A To ⊙Q i tangent, ⊙Q i Let spherical obstacle i be intersected by a circle by a half-plane π; P aits,i Let's consider arc A and circle Q. i The intersection;
[0035] If C0 is not in region 0 and is located inside arc A, then C0 is determined to be in region 1.
[0036] If C0 is not in region 0, but is located outside of arc A and close to obstacle circle 1, then determine that C0 is in region 2-1;
[0037] If C0 is not in region 0, but is located outside of arc A and close to obstacle circle 2, then determine that C0 is in region 2-2.
[0038] The shortest path planning module is used to plan the shortest path based on the different regions where C0 is located.
[0039] When the initial position C0 of the flight camera is in region 0, ray O will be... A The intersection of C0 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0C. * ;
[0040] When the initial position C0 of the flying camera is in region 1, if tangent τ1 exists and tangent τ2 does not exist, the intersection of tangent τ1 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0P. ctgt,1 arc segment P ctgt,1 P otgt,1 and line segment P otgt,1 C * The shortest path is formed by connecting the points sequentially. If tangent τ1 does not exist, but tangent τ2 does, the intersection of tangent τ2 and arc A is taken as the endpoint C of the shortest path. * The corresponding shortest path is the straight line segment C0P. ctgt,2 arc segment P ctgt,2 P otgt,2 and line segment P otgt,2 C * It is formed by connecting them sequentially; if the tangent τ i Both exist; the smaller of the two paths is taken as the shortest path; P otgt,i Tangent τ i The point of tangency with the obstacle circle i, P ctgt,1 P is the point of tangency from C0 to ⊙Q1; ctgt,2 Let C0 be the point of tangency to ⊙Q2;
[0041] When the initial position C0 of the flight camera is in region 2-i, the intersection of arc A and obstacle circle i is taken as the endpoint C of the shortest path. * The corresponding shortest path is: if line segment C0C * Not with ⊙Q i If they intersect, the shortest path is the straight line segment C0C. * If line segment C0C * With ⊙Q i Intersecting lines, the shortest path is formed by the straight line segment C0P. ctgt,i and arc segment P ctgt,i C * It is formed by connecting the links.
[0042] Overall, the above-described technical solutions conceived by this invention can achieve the following beneficial effects compared with the prior art.
[0043] (1) Unlike existing nonlinear optimization problem modeling programs, this invention proposes a novel ESP-based modeling method for composition adjustment problems involving two objects of interest. Therefore, the target position solution of the flying camera in the composition adjustment problem only needs to be solved by finding the shortest path solution of the ESP problem, thus making a solution method based on geometric intuition possible.
[0044] (2) Unlike existing iterative solution space search algorithms, this invention proposes a novel solution method based on the ESP problem for the composition adjustment problem involving two objects of interest. Although the ESP problem is NP-hard in general three-dimensional cases, the method of this invention can provide an accurate closed-form solution for the target position of the flying camera without iteration. Attached Figure Description
[0045] Figure 1 This is a schematic diagram of the shortest path planning problem for a flying camera studied in this invention. In sub-figure (a), when the flying camera is at the target position C, the images of the two objects of interest Q1 and Q2 fall on the desired positions P1 and P2 in the image, respectively, where ω is the line-of-sight angle at this time. In sub-figure (b), a torus Ω can be obtained by taking the line containing Q1Q2 as the axis of rotation and the arc with the circumferential angle ω subtended by the chord Q1Q2 as the generatrix. The target position C of the flying camera is located on this torus. The two objects of interest are represented by spherical obstacles i containing themselves, with the center of the sphere being Q. i The radius of the sphere is r i (i = 1, 2), we need to find the shortest path from the initial position C0 of the flight camera to the target torus Ω, avoiding all obstacles; at this time, C * It is the endpoint of the shortest path to be found, that is, the optimal target position in the target position C of the flying camera located on the target torus Ω, which is the shortest path and does not intersect with obstacles.
[0046] Figure 2 This is a three-dimensional schematic diagram of the shortest path planning method for a flying camera in obstacle space proposed in this invention; where C0 is the initial position of the flying camera, C * The optimal target position for the flying camera is given by Ω, where Q1 and Q2 are the centers of the two spherical obstacles, r1 and r2 are the radii of the two spherical obstacles, Ω is the target torus, and ∏ is the half-plane containing C0 with the line Q1Q2 as its boundary. It is the unit normal vector of the half-plane π. Let A be the unit direction vector of the perpendicular bisector of line segments Q1Q2 within the half-plane π, and let O be the intersecting arc of the half-plane π and the torus Ω. A r A These are the center and radius of arc A, respectively. In the diagram, the green area represents region 0, the red area represents region 1, the blue area near obstacle 1 represents region 2-1, and the blue area near obstacle 2 represents region 2-2. As shown in this diagram, the initial position C0 of the flying camera is located in region 0.
[0047] Figure 3 This is a two-dimensional schematic diagram of the shortest path planning method for a flying camera in obstacle space proposed in this invention, which can be regarded as follows: Figure 2The diagram shows a top view of the 3D structure about the half-plane π (the unit normal vector of the half-plane π). (vertically pointing towards the observer), where the meanings of each symbol and color area are... Figure 2 Same; the sub-image below is an enlarged view of the area within the black rectangle in the image, where P aits,i (i = 1, 2) represents the intersection of arc A and circle Q. i The intersection point, σ i O A To P aits,i Rays, τ i O A To ⊙Q i tangent, P otgt,i It is the tangent τ i With ⊙Q i The point of tangency, P oits,i It is the tangent τ i The subgraph shows the case where i=1, and the case where i=2 can be deduced by analogy.
[0048] Figure 4 Is it like this? Figure 3 The special case of the two-dimensional structure shown, in which the meanings of each symbol and color region are... Figure 2 , 3 All are the same, at this time the radius r1 of the spherical obstacle 1 is greater than the radius r of the circular arc A. A O A It falls inside the spherical obstacle 1, and therefore there is no tangent τ1.
[0049] Figure 5 This is a schematic diagram of the proof of Lemma 1 in the appendix; in it, the two shaded spheres represent two spherical obstacles. On the path shown by the red curve, there can always be a point P that falls in free space (i.e., Euclidean space excluding the interior of the obstacles), and the sphere with P as its center and a positive radius falls completely in free space. At this time, we only consider the two positions A and B where the path enters and leaves the sphere. As long as we replace the original red curve segment with the green line segment connecting the two points, the total path length can be shortened.
[0050] Figure 6 , Figure 7 The diagram above is a proof of Conclusion 1 in Appendix, showing the situation when the initial position C0 of the flight camera is visible to the toroid Ω and is located outside and inside the toroid.
[0051] Figure 8 , Figure 9 The diagram above is a proof of conclusion 2 in the appendix, showing the cases where the initial position C0 of the flight camera is invisible to the toroidal surface Ω and is located outside and inside the toroidal surface, respectively.
[0052] Figure 10This is a visualization of the shortest path calculated according to the method proposed in this invention for four initial positions C0 of the flight camera in a simulation experiment; in which, the straight line segments that make up the shortest path are represented in black, and the arc segments are represented in crimson. Subgraphs (a)-(d) correspond to the cases where C0 falls in region 0, region 1, region 2-1 and region 2-2, respectively.
[0053] Figure 11 This is a difference heatmap designed and calculated in a simulation experiment to verify the method proposed in this invention. The three-dimensional coordinates of each dot correspond to the three-dimensional coordinates of the initial position C0 of the flight camera. The color of each dot indicates the magnitude of the difference: from cool to warm colors, the difference increases. Sub-figures (a)-(d) respectively represent the complete heatmap, a partial slice with Z coordinate in the interval (0.05, 0.1), a partial slice with X coordinate in the interval (0.2, 0.25), and a partial slice with Y coordinate in the interval (-0.55, -0.5). Detailed Implementation
[0054] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.
[0055] The problem of composition adjustment for two objects of interest stems from the intelligent aerial photography requirements of flying cameras: when using flying cameras mounted on drones for aerial photography, it is often desirable for the flying cameras to automatically adjust their positions to meet the composition requirements for two objects of interest in the real world.
[0056] Unlike existing methods, this invention directly addresses the physical meaning of the original problem. By analyzing the geometric properties of the solution space for the target position of the flying camera, it models the original problem as an Euclidean Shortest Path (ESP) problem in computational geometry. Therefore, the solution for the target position of the flying camera in the original problem can be obtained simply by solving the shortest path solution of this ESP problem.
[0057] The ESP problem in 3D refers to finding the shortest path from a starting point to an ending point in 3D Euclidean space, avoiding all polyhedral obstacles. Considering only the exact (not approximate) solution of the shortest path, the ESP problem is NP-hard in the general 3D case; the fastest known algorithm currently requires exponential time. However, if restrictions are placed on this general problem, polynomial-time solutions may exist: for example, when the obstacles are determined to be k disjoint convex polyhedra with a total of n faces, the time complexity of the algorithm becomes O(n^2).o(k) ).
[0058] Therefore, this invention proposes a novel ESP problem modeling and solution method for the composition adjustment problem involving two objects of interest in a typical case. This method can provide an accurate closed-form solution for the target position of the flying camera without iteration.
[0059] Specifically, it is necessary to solve for the target position C of the flying camera. * This allows the flight camera to adjust its attitude at this location so that the object of interest Q is... i The image can fall at the desired position P in the image. i (i = 1, 2), at this point the desired line-of-sight angle ω of the flying camera can be obtained, such as Figure 1 As shown in (a).
[0060] By the inscribed angle theorem, the solution space of the target position of the flying camera is a surface of revolution in three-dimensional Euclidean space. This surface of revolution is uniquely determined by Q1, Q2, and ω (the axis of rotation is the line Q1Q2, and the generatrix is the arc of the circle with an inscribed angle of ω subtended by the chord Q1Q2), denoted as the torus Ω, as follows. Figure 1 As shown in (b).
[0061] Therefore, existing methods construct a nonlinear optimization problem to determine the target position C of the flying camera. * As variables in the optimization or constraint functions, the aim is to find the target position C of the flying camera that satisfies the mapping requirements through a search algorithm in the solution space. * An approximate solution or a solution within a certain range.
[0062] However, simply performing search optimization within the solution space is insufficient; the geometric properties of the solution space must also be fully considered: the solution space is the aforementioned torus Ω, with straight lines as the axes of rotation and circular arcs as the generatrices, thus exhibiting good symmetry. Furthermore, other constraints of the graph adjustment problem must also be fully considered: in practical applications, on the one hand, the flight path should be as short as possible to improve efficiency and save energy; on the other hand, the flight path should avoid the two objects of interest mentioned above to ensure safety.
[0063] Therefore, the composition adjustment problem for two objects of interest can be modeled as solving the shortest path problem from the initial position C0 of the flying camera to the target torus Ω, while avoiding the two objects of interest. The endpoint of the shortest path is the target position C of the flying camera. * To make the modeling method more general and the solution method more convenient, both objects of interest with arbitrary shapes can be represented as spherical obstacles, then Q i Let the center of the sphere be the center, and introduce the sphere radius parameter r based on the actual size of the object. i (i = 1, 2).
[0064] A modeling method for flight phase diagram adjustment problem based on ESP, comprising:
[0065] S1. Load spatial structure information: including the three-dimensional coordinates of the center of the spherical obstacle i in the world coordinate system. i Sphere radius r i (i = 1, 2), the initial position of the flight camera C0 and the target torus Ω.
[0066] S2. Construct the ESP problem model: In a three-dimensional Euclidean space with two spherical obstacles i (i = 1, 2), find the shortest path from the initial position C0 of the flying camera to the target torus Ω, which does not intersect with the interior of any spherical obstacle.
[0067] A solution method based on the above modeling method includes:
[0068] S1. Preprocessing: Solution space dimensionality reduction, that is, reducing the three-dimensional torus to a two-dimensional circular arc.
[0069] To reduce the complexity of the problem solution, it is necessary and worthwhile to perform dimensionality reduction on the solution space beforehand. Based on symmetry, it is easy to determine that the shortest path solution to the ESP problem must lie within the half-plane ∏ defined by points Q1, Q2, and C0. The endpoint of the shortest path is the target position C of the flying camera. * It must lie on the intersecting arc A of the half-plane ∏ and the target torus Ω, such as... Figure 2 As shown.
[0070] The above conclusion can be derived from the shortest subpath theorem in computational geometry (the subpath of the shortest path is still the shortest path).
[0071] S2. Categorical Solution: To determine the target position C of the flying camera on arc A. * It is only necessary to find the distance from the initial position C0 to the target position C * The shortest path is determined by the position of C0 in the half-plane ∏, and obtaining this shortest path requires classification and discussion based on the position of C0 in the half-plane ∏.
[0072] According to the execution steps, the two important processes involved in the classification solution are described below: region division and judgment, and path planning and solution.
[0073] S3. Region Division and Judgment: Divide the plane into four regions on the half-plane ∏ and determine the region where the initial position C0 of the flight camera is located.
[0074] According to the shortest path theory in computational geometry, the specific path planning result is directly related to the position of C0. Therefore, it is necessary to divide the half-plane ∏ into regions and determine the region where the initial position C0 of the flight camera is located.
[0075] like Figure 2 As shown, the circle intercepted by the spherical obstacle i (i = 1, 2) by the half-plane ∏ (the plane containing the obstacle) is denoted as ⊙Q. i Calculate the intersection of arc A and circle Q. i intersection point P aits,i And the center of the circle containing arc A (i.e., the center of the arc) O. A To ⊙Q i tangent τ i The point of tangency P with the obstacle circle i otgt,i .
[0076] It is worth mentioning that, such as Figure 4 As shown, when O A It falls inside the spherical obstacle i (because the radius r of the spherical obstacle i is...) i The radius r is greater than that of arc A A When τ does not exist, there is no tangent line. i Even P otgt,i However, for the sake of consistency in the following text, P will be used here. aits,i Assign the coordinates to P otgt,i And the tangent τ i Consider as ray O A P aits,i (denoted as ray σ) i )coincide.
[0077] This allows us to divide the half-plane Π into four regions (excluding the interior of the obstacle circle):
[0078] Region 0 is ray O A P aits,1 (denoted as ray σ1) with O A Rotate along arc A (the part outside the obstacle circle) to ray O. A P aits,2 The region traversed by (denoted as ray σ2), minus the tangent τ i , σ rays i And the region formed after the area enclosed by the obstacle circle i (i = 1, 2), such as Figure 2 , 3 As shown in the green area in section 4;
[0079] Region 1 is the region located inside arc A after excluding region 0, such as... Figure 2 , 3 As shown in the red area in section 4;
[0080] Region 2-1 is the region outside arc A and close to obstacle circle 1 after excluding region 0, as shown in the blue region on the left in 2, 3, and 4;
[0081] Region 2-2 is the region outside arc A and close to obstacle circle 2 after excluding region 0, as shown in the blue region on the right in 2, 3, and 4;
[0082] This allows us to determine the area where the initial position C0 of the flight camera is located.
[0083] Preferably, to simplify the calculation, a region determination algorithm based on polar coordinates is presented here.
[0084] First, construct an auxiliary coordinate system F on the half-plane Ώ. A Its origin is the center of the arc O. A Its unit direction vector O pointing positively upwards along the x-axis A x is the same as the unit direction vector pointing from Q1 to Q2, and its y-axis points positively upwards along the unit direction vector O. A The unit direction vector of the perpendicular bisector of line segment Q1Q2 in the half-plane π is the same as that of the line segment Q1Q2, and its unit direction vector O is positively upward along the z-axis. A z can be uniquely determined by the unit direction vectors of the two axes above according to the right-hand rule.
[0085] Secondly, calculate the initial position C0 of the flying camera and other key points P. aits,i P otgt,i Q i (i = 1, 2) In the half-plane Ώ, according to xO A The polar coordinates corresponding to the y-coordinate system, for example, P aits,1 polar coordinates The other points are notated in the same way.
[0086] Finally, based on the above polar coordinates, the region is determined: when C0 satisfies... Or simultaneously satisfy and Or simultaneously satisfy and When C0 is in region 0, determine that the region where C0 is located is region 0; when C0 is not in region 0 and is located inside the torus, i.e. When C0 is in region 1, it is determined that the region where C0 is located is region 1; when C0 is not in region 0 and is located outside the torus, it is determined that the region where C0 is located is region 1. At that time, according to Determine that the region where C0 is located is region 2-i (i = 1, 2).
[0087] S4. Path Planning and Solution: Based on the different areas where C0 is located, the shortest path is planned, and the endpoint of the shortest path is the target position C of the flying camera. * .
[0088] (1) When the initial position C0 of the flight camera is in region 0
[0089] Flight camera target position C *That is, ray O A The intersection of C0 and arc A is the shortest path, which is the straight line segment C0C. * .
[0090] The proof is as follows:
[0091] To prove line segment C0C * The shortest path is found by choosing any point on the torus Ω that does not intersect C. * Find the point C′ that coincides with the line segment, and prove that the length of the line segment ||C0C * ||<||C0C′|| is sufficient:
[0092] (i) When C0 is located outside the torus (e.g.) Figure 6 As shown),
[0093] At ΔC′C0C * In the middle, there is ∠C′C * C0 > 90°, therefore ||C0C * ||<||C0C′||;
[0094] (ii) When C0 is located inside the torus (e.g.) Figure 7 (as shown),
[0095] In ΔO A C′C * In the middle, there is ||C0C′||+||O A C0||>||O A C′||;
[0096] And because ||O A C′||≥r A =||O A C0||+||C0C γ ||, thus we can obtain ||C0C γ ||<||C0C′||.
[0097] Proof obtained.
[0098] (2) When the initial position C0 of the flight camera is in region 1
[0099] Flight camera target position C * There are two possibilities for the shortest path, which is O. A To ⊙Q i tangent τ i The existence of different cases is discussed in categories:
[0100] (i) Tangent τ1 exists, tangent τ2 does not exist: target position C * That is, P oits,1 (P oits,1That is, the intersection of tangent τ1 and arc A; the shortest path is the straight line segment C0P. ctgt,1 (P ctgt,1 That is, the point of tangency from C0 to ⊙Q1), and the arc segment P. ctgt,1 P otgt,1 and line segment P otgt,1 C * Formed by connecting them sequentially, such as Figure 9 As shown.
[0101] (ii) Tangent τ1 does not exist, tangent τ2 exists: target position C * That is, P oits,2 (P oits,2 That is, the intersection of tangent τ2 and arc A; the shortest path is the straight line segment C0P. ctgt,2 (P ctgt,2 That is, the point of tangency from C0 to ⊙Q2, and the arc segment P. ctgt,2 P otgt,2 and line segment P otgt,2 C * It is formed by connecting them sequentially.
[0102] (iii) If the tangent τ i For cases (i = 1, 2), both exist: calculate the shortest path length for each of the two cases, and take the smaller one as the shortest path in this case; for example, when the shortest path in case (i) is less than the shortest path in case (ii), the target position C in this case is... * The same applies to the shortest path (i), and vice versa (ii).
[0103] It should be noted that the tangent τ i The case where (i = 1, 2) does not exist does not exist in itself. Therefore, the discussion of the above three cases has fully explained the target position C of the flying camera. * And the two possibilities of the shortest path. The proof of the above conclusion is the same as the following case, and will not be discussed here.
[0104] (3) When the initial position C0 of the flight camera is in region 2-i (i = 1, 2),
[0105] Flight camera target position C * That is, P aits,i (The intersection of arc A and obstacle circle i), at this point, there are two possibilities for the shortest path:
[0106] (i) If line segment C0C * Not with ⊙Q i When they intersect, the shortest path is the straight line segment C0C. * ;
[0107] (ii) If line segment C0C * With ⊙Q iFor intersections, the intermediate node of the shortest path, i.e., from C0 to ⊙Q, needs to be calculated. i tangent τ c,i With ⊙Q i tangent point P ctgt,i At this point, the shortest path is the straight line segment C0P. ctgt,i and arc segment P ctgt,i C * It is formed by connecting the links.
[0108] The above conclusion can be naturally derived from the RubberBand Algorithm (RBA) in computational geometry. (Reference) Figure 8 The shortest path can be considered (here, the straight line segment C0P). ctgt,1 and arc segment P ctgt,1 C * (Connected together) is formed by any path (using polyline segments C0A, AB, and BP) ctgt,1 (Taking as an example) it is formed by infinitely approaching the spherical obstacle 1. The specific proof process is too lengthy to be elaborated here.
[0109] Simulation experiment:
[0110] 1. The method proposed in this invention was simulated using MATLAB programming, and a brute-force traversal method was used as a reference to verify the correctness and effectiveness of the proposed method.
[0111] Without loss of generality, a fixed scene can be randomly designed, with the following information about the spherical obstacles and the target torus:
[0112] Q1 = (0, 0, 0),
[0113] Q2 = (1, 0, 0),
[0114] r1 = 0.4,
[0115] r2 = 0.28,
[0116] ω = 43.6°;
[0117] To verify the completeness of the method, different initial positions C0 of the flight camera were set sequentially, so that C0 falls within region 0, region 1, region 2-1, and region 2-2 respectively:
[0118] ①C0 = (0.8, 0.5, 1.5), located in region 0;
[0119] ②C0 = (0.45, 0.1, 0.2), which is in region 1;
[0120] ③C0 = (-0.45, 0.02, 0.05), which is in region 2-1;
[0121] ④C0 = (1.35, 0.01, 0.04), located in region 2-2;
[0122] 2. Calculate the shortest path and length of the flying camera according to the method proposed in this invention.
[0123] Referring to the specific implementation method, the shortest path and shortest path length of the flying camera were calculated by taking the above four different values of C0 as the only variable, and the results are shown in Table 1. Figure 10 The shortest paths above are visualized in (a)-(d) respectively.
[0124] Table 1 shows the shortest path and length results provided by the method proposed in this invention.
[0125]
[0126] 3. Using a brute-force traversal method as a reference, the correctness and effectiveness of the proposed method are verified.
[0127] In this simulation experiment, the programming implementation of the target torus Ω is discretized: the rotation axis constituting the target torus is the straight line Q1Q2, and the generatrix is the arc of the circle with an inscribed angle ω subtended by the chord Q1Q2, as shown below. Figure 1 As shown in (b), the rotation angle (with a magnitude of 2π) around the axis Q1Q2 is divided into 100 equal parts, resulting in 100 of the above-mentioned arcs. These arcs are then divided into 100 equal parts according to their subtended central angle (with a magnitude of 2π-2ω). The entire target torus can then be discretized into 10,000 points, which can be used as a discrete approximation set of the original target torus to a certain extent.
[0128] Given the initial position C0 of the flying camera, for each discrete point C on the target torus Ω k Both can be used to calculate C0 to C k Shortest path S k (k = 1, ..., 10) 4 To verify the correctness and effectiveness of the method proposed in this invention, it is only necessary to determine all S k All are greater than or equal to the shortest path length S of the flying camera under the method proposed in this invention. min That is, it is only necessary to prove S min ≤S k k = 1, ..., 10 4 Therefore, we only need to find the set. The minimum value or a certain lower bound S′ (i.e., for any S) k For all, S′≤S k ), by verifying S min The statement that ≤S′ holds true indicates that S min ≤S k k = 1, ..., 10 4 .
[0129] In practice, for each C k We can first calculate C0 to C k Euclidean distance When line segment C0C k When it does not intersect with the spherical obstacle, C0 to C k shortest path And when line segment C0C k When intersecting with spherical obstacles (whether one or two), C0 to C k shortest path therefore, It is a set A lower bound; however, it is easy to imagine that a specific point C on the torus... k The Euclidean distance to C0 will always be less than the shortest path length S of the flying camera under the method proposed in this invention. min At this time, line segment C0C k It only intersects with a single spherical obstacle; in this case, point C needs to be calculated. k The exact shortest path S to C0 k For the method, refer to Lemma 1.
[0130] In summary, sets The lower bound S′ can be selected as follows:
[0131] (1) Traverse each C k Calculate C0 to C k Euclidean distance
[0132] (2) If remember like Then calculate point C k The exact shortest path S to C0 k And record
[0133] (3) Calculation
[0134] Following these steps, the calculation results and comparative analysis are shown in Table 2.
[0135] Table 2 shows the results and comparative analysis of the lower bound of the shortest path length given by the brute-force traversal method.
[0136]
[0137] Furthermore, to make the above comparative analysis more complete and intuitive, the initial position C0 of the flight camera can be distributed as evenly as possible in three-dimensional Euclidean space (excluding the interiors of the two spherical obstacles). Then, the above experimental operation is repeated for each C0, and S′-S is calculated. min Finally, if all these differences are greater than or equal to 0, the method proposed in this invention passes the test; otherwise, it fails.
[0138] Therefore, in this experiment, the midpoint of line segment Q1Q2, i.e., (0.5, 0, 0), is set as the origin of the spherical coordinate system, and the radius is set to 1. The azimuth interval (size 2π), elevation interval (size 2π), and distance interval (size 1) are divided into 50, 50, and 20 equal parts, respectively, thus discretizing 50,000 points in the spherical coordinate system. Among these points, excluding some points located inside the spherical obstacle, all other points will be successively captured by the initial position C0 of the flight camera.
[0139] like Figure 11 As shown in (a), S′-S can be calculated for each C0. min These differences are illustrated using a heatmap. The three-dimensional coordinates of each dot correspond to the three-dimensional coordinates of C0, and the color of each dot indicates the magnitude of the difference: from cool to warm colors, the difference increases. The heatmap shows that the minimum difference is 0, and the maximum difference is 1.64152 * 10^6. -2 That is, all differences are greater than or equal to 0, and the method proposed in this invention can be verified.
[0140] Figure 11 (b)-(d) are slices of the complete heatmap shown in (a), respectively, showing the portion of the complete heatmap where the Z coordinate is in the range of (0.05, 0.1), the X coordinate is in the range of (0.2, 0.25), and the Y coordinate is in the range of (-0.55, -0.5).
[0141] In summary, the correctness and effectiveness of the method proposed in this invention can be demonstrated.
[0142] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A solution method based on a modeling method for flight phase diagram adjustment problems, characterized in that, The flight phase diagram adjustment problem modeling method is an ESP-based flight phase diagram adjustment problem modeling method, which includes: S1. Identify obstacles on the sphere. The three-dimensional coordinates of the center of the sphere in the world coordinate system Sphere radius Initial position of the flying camera and by , and Determined target torus ;in, For the flying camera at the target location Object of interest after adjusting posture The image can fall at the desired position in the image. The expected line-of-sight angle of the flight camera at that time; ; S2. In an obstacle with two spheres In three-dimensional Euclidean space, solve for the flight camera from its initial position. To the target torus The shortest path that does not intersect the interior of any spherical obstacle; the endpoint of the shortest path is located at the point where... , and A half-plane defined by three points With the target torus Intersecting arcs superior; Solution methods include: S1. Determine the initial position of the flight camera. Location: like Under radiation by Centered along the arc Rotate to ray The area traversed, minus the tangent. ,ray , and obstacle circles The area formed after the enclosed area is then judged. Located in region 0; where, For arc The center of the arc, For the center of the arc arrive tangent, Sphere obstacle Half-plane The circle cut off; For arc and The intersection point; like Not in region 0, but located on an arc. Internally, then judge Located in region 1; like Not in region 0, but located on an arc. The region outside and close to obstacle circle 1 is then judged. In the range of 2-1; like Not in region 0, but located on an arc. The area outside and close to obstacle circle 2 is then judged. In 2-2; S2. According to Shortest path planning is performed based on the different areas they are located in: When the flight camera is in its initial position When in region 0, the ray With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path is a straight line segment. ; When the flight camera is in its initial position When in region 1, if the tangent Existence, tangent Does not exist, tangent line With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path consists of a straight line segment. arc segment and line segments Formed by connecting them sequentially; if tangent No, tangent Existence, will the tangent With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path consists of a straight line segment. arc segment and line segments Formed by connecting them sequentially; if tangent If both exist, the smaller of the two paths is taken as the shortest path. Tangent With the obstacle circle The point of tangency, for arrive The point of tangency; for arrive The point of tangency; When the flight camera is in its initial position Located in Zone 2- At that time, the arc With the obstacle circle The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path is: if line segment Not with If they intersect, the shortest path is a straight line segment. If line segment and When two lines intersect, the shortest path is formed by a straight line segment. and arc segment It is formed by connecting the links.
2. The solution method according to claim 1, characterized in that, Step S1 specifically includes, (01) Constructing an auxiliary coordinate system : half-plane With the target torus Intersecting arcs arc center As The origin; Will be point to The unit direction vector is used as The unit direction vector pointing positively upwards along the axis; Line segment In the half-plane The unit direction vector of the perpendicular bisector within is used as The unit direction vector pointing positively upwards along the axis; based on axis, The unit direction vector pointing positively upwards along the axis is determined by the right-hand rule. The unit direction vector pointing positively upwards along the axis; (02) Calculate the initial position of the flight camera and , , In the half-plane Internal basis Polar coordinates corresponding to the coordinate system , , , ;in, Tangent With the obstacle circle The point of tangency; when satisfy or simultaneously satisfy and or simultaneously satisfy and At that time, make a judgment The region it is located in is region 0; when Not in region 0 and located inside the torus At that time, make a judgment The region it is located in is region 1; r A half-plane With torus Intersecting arcs The radius of the arc; when Not in region 0 and located outside the torus At that time, according to judge The area it is located in is Region 2- , .
3. A solution system for a modeling system of flight phase diagram adjustment problems, characterized in that, The flight phase diagram adjustment problem modeling system is an ESP-based system for modeling flight phase diagram adjustment problems. This modeling system includes: The spatial structure information loading module is used to determine obstacles on the sphere. The three-dimensional coordinates of the center of the sphere in the world coordinate system Sphere radius Initial position of the flying camera and by , and Determined target torus ;in, For the flying camera at the target location Object of interest after adjusting posture The image can fall at the desired position in the image. The expected line-of-sight angle of the flight camera at that time; ; The ESP model building module is used for obstacles with two spheres. In three-dimensional Euclidean space, solve for the flight camera from its initial position. To the target torus The shortest path that does not intersect the interior of any spherical obstacle; the endpoint of the shortest path is located at the point where... , and A half-plane defined by three points With the target torus Intersecting arcs superior; The solution system includes: The area determination module is used to determine the initial position of the flying camera. Location: like Under radiation by Centered along the arc Rotate to ray The area traversed, minus the tangent. ,ray , and obstacle circles The area formed after the enclosed area is then judged. Located in region 0; For arc The center of the arc, For the center of the arc arrive tangent, Sphere obstacle Half-plane The circle cut off; For arc and The intersection point; like Not in region 0, but located on an arc. Internally, then judge Located in region 1; like Not in region 0, but located on an arc. The region outside and close to obstacle circle 1 is then judged. In the range of 2-1; like Not in region 0, but located on an arc. The area outside and close to obstacle circle 2 is then judged. In 2-2; The shortest path planning module is used to calculate the shortest path based on the path. Shortest path planning is performed based on the different areas they are located in: When the flight camera is in its initial position When in region 0, the ray With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path is a straight line segment. ; When the flight camera is in its initial position When in region 1, if the tangent Existence, tangent Does not exist, tangent line With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path consists of a straight line segment. arc segment and line segments Formed by connecting them sequentially; if tangent No, tangent Existence, will the tangent With arc The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path consists of a straight line segment. arc segment and line segments Formed by connecting them sequentially; if tangent If both exist, the smaller of the two paths is taken as the shortest path. Tangent With the obstacle circle The point of tangency, for arrive The point of tangency; for arrive The point of tangency; When the flight camera is in its initial position Located in Zone 2- At that time, the arc With the obstacle circle The intersection point is taken as the endpoint of the shortest path. The corresponding shortest path is: if line segment Not with If they intersect, the shortest path is a straight line segment. If line segment and When two lines intersect, the shortest path is formed by a straight line segment. and arc segment It is formed by connecting the links.
4. A modeling device for flight phase diagram adjustment problems based on ESP, characterized in that, include: processor; The processor is coupled to a memory for storing computer programs or instructions, and the processor is used to execute the computer programs or instructions in the memory, so that the solution method of claim 1 is executed.
5. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein the computer program, when executed by a processor, controls the device containing the storage medium to perform the solution method as described in claim 1.
Citation Information
Patent Citations
Aircraft route planning method and system based on identification point visual navigation and SINS (strap-down inertial navigation system)
CN108426576A
