Urban unmanned aerial vehicle anisotropic integrity risk modeling and path planning method

Abstract

The application belongs to the technical field of unmanned aerial vehicle navigation and path planning, and specifically discloses a kind of city unmanned aerial vehicle anisotropy integrity risk modeling and path planning method.The method establishes position error covariance submatrix according to space observation geometric matrix and linear signal quality factor, and projects the space relationship between the current position of unmanned aerial vehicle and the obstacle, introduces conservative Gaussian crossing envelope model to calculate anisotropic collision risk;Then three kinds of risk information are fused to generate anisotropic integrity risk field, and a risk constraint three-dimensional search graph is constructed based on the risk field, a vertical relaxation coefficient is introduced to scale the vertical displacement cost, and a multiplication transfer cost function is combined to search the safe path of unmanned aerial vehicle in three-dimensional space;Finally, the trajectory of the safe path of unmanned aerial vehicle is smoothed to generate a continuous three-dimensional trajectory that meets the kinematic and dynamic constraints.

Description

Technical Field

[0001] This invention belongs to the field of UAV navigation and path planning technology, specifically relating to a method for modeling and planning the anisotropic integrity risk of urban UAVs. Background Technology

[0002] In recent years, with the booming development of the "low-altitude economy," drones have been increasingly widely used in urban logistics, emergency rescue, and high-altitude inspection. Supported by the Global Navigation Satellite System (GNSS), drones can achieve autonomous track tracking and spatial positioning. In open airspace, GNSS positioning typically provides extremely high accuracy and integrity, forming the foundation for the safe operation of low-altitude drones. However, in the "urban canyon" environment of densely packed high-rise buildings, such as in core urban business districts, the safe navigation of drones faces extremely severe challenges. The complex urban building topology severely obstructs the line of sight to space satellites, causing severe non-line-of-sight (NLOS) obstruction, multipath reflection, and edge diffraction effects during GNSS signal propagation. These severe electromagnetic interferences cause the drone's underlying positioning error to exhibit significant nonlinearity, thick tail distribution, and spatial anisotropy, easily leading to positioning solution divergence and track loss, resulting in catastrophic accidents such as drone collisions with building facades.

[0003] To address the risk avoidance planning problem for drones in complex urban environments, existing technical solutions typically employ a two-stage processing mechanism: First, an environmental risk map is constructed using the geometrical precision factor (GDOP) or the method of expanding the safe radius of physical obstacles; then, a path search is performed in three-dimensional space based on graph search algorithms (such as the A* algorithm) or swarm intelligence heuristic algorithms (such as ant colony and wolf pack algorithms), with the cost function often being a linear weighted combination of physical flight distance and risk penalty.

[0004] However, these existing technologies have the following obvious limitations when dealing with drone risk avoidance planning in complex electromagnetic environments: First, existing risk assessment models often use the isotropic assumption, ignoring the "anisotropic" error distribution caused by unilateral building shading, which can easily generate fatal safety illusions in narrow streets; Second, existing planning algorithms usually perform a simple linear weighting of physical distance and risk probability when constructing cost functions, resulting in inconsistent dimensions of multi-objective optimization; Third, traditional algorithms have a "geotropic" trap in three-dimensional space, preferring to blindly detour in areas with strong interference at the bottom rather than spontaneously evolving an active risk avoidance strategy towards safe airspace above. Summary of the Invention

[0005] The purpose of this invention is to propose an anisotropic integrity risk modeling and path planning method for urban UAVs. By introducing a conservative Gaussian overbound envelope model to calculate anisotropic collision risk, introducing a power-order shaping operator when fusing the risk field, and introducing a vertical relaxation coefficient in the A* algorithm, the method enhances the UAV's risk avoidance capability and improves the safety of path planning.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for anisotropic integrity risk modeling and path planning of urban unmanned aerial vehicles (UAVs) includes the following steps: Step 1. Construct a 3D environmental map of the urban canyon in the target flight area and establish a spatial observation geometry matrix between the UAV and visible satellites; Step 2. Calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and establish the position error covariance submatrix based on the spatial observation geometry matrix and the linear signal quality factor; Step 3. Based on the spatial relationship between the current position of the UAV and the obstacle, perform directional projection on the position error covariance submatrix to obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian overbound envelope model to calculate the anisotropic collision risk. Step 4. Calculate the visibility degradation risk and multipath interference risk, and fuse the visibility degradation risk, multipath interference risk and anisotropic collision risk to generate a comprehensive anisotropic integrity risk field; Step 5. Construct a three-dimensional search graph of risk constraints based on the comprehensive anisotropic integrity risk field. In the process of expanding adjacent nodes, introduce a vertical relaxation coefficient to scale the vertical displacement cost, and combine it with the multiplicative transfer cost function to search for the safe path of the UAV in three-dimensional space. Step 6. Perform trajectory smoothing on the UAV's safe path to generate a continuous three-dimensional trajectory that satisfies kinematic and dynamic constraints.

[0007] Furthermore, based on the aforementioned method for modeling and path planning the anisotropic integrity risk of urban UAVs, this invention also proposes a corresponding system for modeling and path planning the anisotropic integrity risk of urban UAVs, which adopts the following technical solution: A system for modeling the anisotropic integrity risk and planning the path of urban unmanned aerial vehicles (UAVs) includes the following modules: The spatial observation geometric model building module is used to construct a three-dimensional environmental map of the urban canyon in the target flight area and to establish a spatial observation geometric matrix between the UAV and visible satellites. The positioning error establishment module is used to calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and to establish the position error covariance sub-matrix based on the spatial observation geometry matrix and the linear signal quality factor. The anisotropic collision risk model building module is used to perform directional projection on the position error covariance submatrix based on the spatial relationship between the current position of the UAV and the obstacle, obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian overbound envelope model to calculate the anisotropic collision risk. The integrated anisotropic integrity risk field generation module is used to calculate visibility degradation risk and multipath interference risk, and integrates visibility degradation risk, multipath interference risk and anisotropic collision risk to generate an integrated anisotropic integrity risk field. The UAV safe path generation module is used to construct a risk constraint three-dimensional search graph based on the comprehensive anisotropic integrity risk field. During the expansion of adjacent nodes, a vertical relaxation coefficient is introduced to scale the vertical displacement cost, and the UAV safe path is obtained by searching in three-dimensional space in combination with the multiplicative transfer cost function. The trajectory smoothing module is used to smooth the trajectory of the UAV for safe travel, generating a continuous three-dimensional trajectory that meets kinematic and dynamic constraints.

[0008] Furthermore, based on the aforementioned method for modeling and planning the anisotropic integrity risk of urban UAVs, this invention also proposes a computer device comprising a memory and one or more processors. Executable code is stored in the memory. When the processor executes the executable code, it implements the steps of the aforementioned method for modeling and planning the anisotropic integrity risk of urban UAVs.

[0009] Furthermore, based on the aforementioned method for modeling and path planning the anisotropic integrity risk of urban UAVs, this invention also proposes a computer-readable storage medium storing a program that, when executed by a processor, implements the steps of the aforementioned method for modeling and path planning the anisotropic integrity risk of urban UAVs.

[0010] The present invention has the following advantages: As described above, this invention proposes a method for anisotropic integrity risk modeling and path planning for urban UAVs. This method constructs a positioning error covariance matrix and extracts its first three rows and three columns to obtain a three-dimensional position error covariance submatrix. Then, it uses a quadratic projection of the unit threat direction vector to obtain the edge positioning error variance facing the nearest obstacle surface point, thus achieving accurate modeling of GNSS directional positioning errors in urban canyon environments. This invention can effectively describe the error ellipsoid stretching phenomenon caused by unilateral building occlusion in urban canyon environments, making the UAV risk assessment results more consistent with the real GNSS propagation characteristics, thereby significantly improving the spatial realism and navigation reliability of path planning in complex environments. Meanwhile, existing path planning methods typically use distance thresholds, obstacle inflation, or empirical penalty functions to describe risks, which lack probabilistic statistical significance and make it difficult to achieve quantitative analysis of navigation integrity. This invention establishes an edge positioning error variance oriented towards the nearest obstacle surface point, which is further combined with environment-driven alarm limits and a conservative Gaussian over-boundary envelope model to calculate anisotropic collision risks with clear statistical significance, transforming UAV collision risk from a traditional geometric distance problem into a statistical probability problem. This invention realizes a probabilistic expression of collision risk in complex urban canyon environments, giving the risk assessment results clear mathematical significance and engineering interpretability, and improving the safety of path planning. Furthermore, existing risk fields typically employ a simple linear weighting method to fuse risk indicators, resulting in issues such as inconsistent dimensions, high parameter sensitivity, and discontinuous risk gradients. This invention uses a nonlinear power-law shaping operator to suppress low-risk background noise and enhance the boundaries of high-risk areas, forming a comprehensive anisotropic integrity risk field that is more effective in distinguishing between safe and high-risk areas. During the path planning phase, this comprehensive anisotropic integrity risk field participates in the expansion of adjacent nodes as a risk amplification term in the multiplicative transfer cost function, actively amplifying the equivalent travel distance of high-risk nodes. This guides the UAV to avoid high-risk areas and prioritize searching for low-risk routes. Traditional A* algorithms often exhibit significant geotropism in 3D urban environments, easily causing UAVs to linger in low-altitude high-risk areas for extended periods, lacking the ability to actively escape to high-altitude safe areas. This invention introduces a vertical relaxation coefficient into the A* algorithm to scale the vertical displacement cost, enabling the UAV to prioritize searching for high-altitude low-risk areas. This invention effectively enhances the UAV's active vertical risk avoidance capability, reduces the probability of high-risk passage through complex urban canyons, and improves the safety margin of path planning. Attached Figure Description

[0011] Figure 1 This is a flowchart of the method for modeling and path planning the anisotropic integrity risk of urban UAVs in an embodiment of the present invention. Detailed Implementation

[0012] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Example 1 This embodiment 1 describes a method for anisotropic integrity risk modeling and path planning for urban unmanned aerial vehicles (UAVs), such as... Figure 1 As shown, the method includes the following steps: Step 1. Construct a 3D environmental map of the urban canyon in the target flight area and establish a spatial observation geometry matrix between the UAV and visible satellites.

[0013] First, large-scale three-dimensional elevation data and building outline data of the target flight area are acquired. The building outlines, height information and flight airspace boundaries are uniformly converted to the same geographic coordinate system. The continuous physical space is then processed into voxels according to the preset spatial resolution to obtain a three-dimensional urban canyon environment map composed of passable voxels, obstacle voxels and boundary voxels.

[0014] Simultaneously, GNSS satellite ephemeris parameters are acquired, and the positions of each visible satellite in the geocentric-ground-fixed coordinate system are calculated based on Kepler orbital parameters and the WGS84 ellipsoid model. In the simulation verification scenario, a geometric satellite distribution with regular spatial distribution can also be generated based on preset orbital parameters to construct the spatial observation relationship between the UAV and visible satellites.

[0015] Subsequently, based on the aforementioned GNSS satellite ephemeris parameters and the UAV's current position, the UAV's position was calculated up to the [number missing]. elevation angle of visible satellites With azimuth The calculation formula is as follows: ; ; in Indicates the first The line-of-sight vector of the visible satellite relative to the current position of the UAV in the northeast-northeast celestial coordinate system of the station center. , , , They represent the first The visible satellites are located relative to the drone's current position in the east, north, and sky directions. express The Euclidean norm.

[0016] Based on this, construct a system containing Space observation geometry matrix of visible satellites Its specific form is: ; Among them, the spatial observation geometric matrix The first three columns of components constitute the unit direction vector pointing to each visible satellite in the station-centered coordinate system, and the fourth column constant 1 corresponds to the clock error state of the receiver.

[0017] Step 2. Based on the spatial relationship between the satellite line-of-sight rays determined by 3D ray tracing and the building edges, building facades, and obstacle objects, calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and establish the position error covariance submatrix based on the spatial observation geometry matrix and the linear signal quality factor.

[0018] To overcome the isotropic limitations of traditional GDOP models and approximate the impact of building edges on GNSS signal propagation, a three-dimensional ray tracing method was used to detect the spatial relationship between each satellite line-of-sight ray and building edges, building facades, and obstacle elements in the three-dimensional urban canyon environment map. The clearance height of building edges relative to the line connecting the satellite transmitter and the UAV receiver was also calculated. Then the clearance height Substituting into the Fresnel-Kirchhoff edge diffraction model, calculate the first... Linear signal quality factor of visible satellite observation channels The calculation formula is as follows: ; in Indicates the first The linear signal quality factor for each visible satellite observation channel is used to characterize the impact of urban buildings on the observation quality of a single channel. When visible satellite signals are near the edge of a building or are blocked, As the linear signal quality factor decreases, the variance of the observation noise increases accordingly. It should be noted that this embodiment uses this linear signal quality factor as an engineering approximation for the path planning stage, rather than a complete characterization of the complex urban electromagnetic propagation process.

[0019] This indicates the additional loss from Fresnel-Kirchhoff edge diffraction.

[0020] ; in These are Fresnel-Kirchhoff diffraction parameters. . The value can be positive, zero, or negative. When the edge of a building encroaches on the line-of-sight propagation path between the transmitter and receiver, or the first Fresnel zone, Positive values ​​are typically used, indicating enhanced shading and diffraction effects; however, when the building edge is below the line-of-sight propagation channel and has a certain clearance, the value is less significant. Negative values ​​are acceptable, indicating a relatively wide propagation path. The threshold of -0.78 is an empirical cutoff value in the approximate engineering model of edge diffraction, used to determine whether additional diffraction losses can be ignored; when... At that time, it was assumed that the additional loss caused by edge diffraction at the edge of the building to the satellite observation channel was small and could be approximated as 0.

[0021] ; in The clearance height of the building edge relative to the line connecting the satellite launcher and the drone receiver. and These are the distances from the building's edge to the transmitter and receiver, respectively. It is the wavelength of the L1 band (GNSS L1) signal of the Global Navigation Satellite System.

[0022] ; in This refers to the L1 band signal frequency of the Global Navigation Satellite System. The speed of light; in this embodiment , .

[0023] Based on the nominal user equivalent ranging error constant of the GNSS receiver, the first step is to calculate the error caused by elevation angle error amplification and channel linear signal quality factor modulation. Comprehensive observation variance of visible satellite observation channels for: ; in The basic system error constant, These are the elevation angle-related error amplification parameters; 'a' and 'b' are preset based on the receiver model, GNSS frequency, simulation calibration results, and historical observation data. The specific values ​​of 'a' and 'b' depend on the hardware performance of the GNSS receiver on the UAV and the ambient noise level; in typical urban low-altitude flight scenarios, their typical range is usually... , m.

[0024] Furthermore, constructing a system based on The observation weight matrix for diagonal elements : ;in This represents a diagonal matrix.

[0025] Based on the weighted least squares estimation theory, the covariance matrix of the UAV's positioning error is calculated. for: .

[0026] Extracting the matrix The first three rows and first three columns yield the three-dimensional position error covariance submatrix. This matrix accurately describes, in mathematical analytical form, the anisotropic expansion characteristic of positioning errors caused by unilateral occlusion in the city, which stretches in a specific direction.

[0027] Existing UAV path planning methods typically employ the isotropic error assumption, which assumes that the positioning error spreads uniformly in all directions. This fails to accurately reflect the amplification of directional errors caused by building obstruction in complex urban canyon environments. By constructing a positioning error covariance matrix and extracting its first three rows and three columns to obtain a three-dimensional position error covariance submatrix, and then using a quadratic projection of the unit threat direction vector, the edge positioning error variance of the collision direction (i.e., the direction facing the nearest obstacle surface point) is established, thus realizing the coupled modeling between positioning error and obstacle spatial direction.

[0028] Step 3. Based on the spatial relationship between the current position of the UAV and the obstacle, perform directional projection on the positioning error covariance matrix to obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian over-boundary envelope model to calculate the anisotropic collision risk.

[0029] In the 3D environment map of the urban canyon, surface sampling points are extracted from building facades, roof edges, airspace boundaries, and other solid obstacles such as poles, bridges, and fences to form an obstacle surface point set. and based on Establish a KD-Tree spatial index structure; using the current centroid coordinates of the UAV As the query point, a nearest neighbor search is performed in the KD-Tree to obtain the obstacle surface point closest to the current centroid coordinates of the UAV. The calculation formula is as follows: ; in express The coordinates of any candidate obstacle surface point in the data. Indicates the current centroid coordinates of the drone. This represents the coordinates of the obstacle surface point closest to the drone's current centroid coordinates. This represents the value of the variable that minimizes the objective function. This represents the Euclidean distance in three-dimensional space.

[0030] according to and Construct a unit threat direction vector from the UAV to the nearest obstacle surface point. The calculation formula is as follows: .

[0031] The three-dimensional position error covariance submatrix Vector along the unit threat direction Perform a quadratic projection to obtain the variance of the edge localization error in the direction facing the nearest obstacle surface point. The calculation formula is as follows: .

[0032] Calculate the Euclidean distance from the current UAV's center of mass to a point on the obstacle surface, and subtract the UAV's physical equivalent safety radius to obtain the environment-driven alarm limit. The calculation formula is as follows: ; in This represents the physical equivalent safety radius of the drone. For small multi-rotor drones, The preferred value range is 0.3–1.5 m; for UAVs with large loads or large rotor deployment dimensions, The preferred value range is 2 to 5 m.

[0033] To convert the positioning error in the unit threat direction into probabilistic risk, a conservative Gaussian over-boundary envelope model is introduced, and the anisotropic collision risk is calculated using the standard complementary error function. The calculation formula is as follows: ; in The anisotropic collision risk is used to characterize the upper bound of the probability that the actual position of the UAV will exceed the environmentally driven alarm limit along the unit threat direction. The anisotropic collision risk value obtained from this is the collision risk component in the comprehensive anisotropic integrity risk field. This represents the standard complementary error function.

[0034] The expression for the standard complementary error function is: .

[0035] The formula for calculating anisotropic collision risk transforms the originally highly complex three-dimensional numerical calculus into a constant-level table lookup operation, significantly improving the real-time performance of airborne computing.

[0036] Step 4. Calculate visibility degradation risk and multipath interference risk, and fuse multi-source risk information including visibility degradation risk, multipath interference risk and anisotropic collision risk to generate a comprehensive anisotropic integrity risk field.

[0037] First, quantify the risk of visibility degradation; combine an ideal set of visible satellites in open airspace with the first ray tracing obtained from three-dimensional ray tracing. The linear signal quality factor of a visible satellite observation channel will reduce the risk of visibility degradation. Defined as the weighted missing rate of effective signal energy within a locally confined spatial domain, its calculation formula is as follows: ; in For an ideal collection of visible satellites in open airspace, To prevent extremely small positive numbers with a denominator of zero; For the first Elevation angle weights for each visible satellite observation channel.

[0038] ;in This indicates taking the maximum value, when When less than 0, The value should be 0, and negative numbers should be avoided.

[0039] When a visible satellite observation channel is completely unlocked or completely blocked, the unlocked or blocked visible satellite observation channel will not participate in the calculation of the positioning error covariance matrix, or its linear signal quality factor will be set to a preset minimum positive number. To avoid singularities in the observation weight matrix; when the visible satellite channel does not experience propagation degradation, take Close to 1.

[0040] For the propagation path between the current position of the UAV and the visible satellite, a reflection ray tracing is performed based on the 3D environment map of the urban canyon. When the visible satellite signal can reach the UAV receiver after being reflected by the building facade, glass curtain wall or roof edge, and the length of the reflection path is greater than the length of the direct path or the direct path is blocked by the building, the reflection path is marked as an effective non-line-of-sight (NLOS) ray.

[0041] Count the number of effective non-line-of-sight rays at the current location of the drone. The Poisson survival model is used to map the effective number of non-line-of-sight rays into the risk of multipath interference. The calculation formula is as follows: ; in The preset risk penalty coefficient, .

[0042] Based on the Fault Tree Analysis (FTA) model, anisotropic collision risk, visibility degradation risk, and multipath interference risk are defined as independent failure events. The conservative envelope total risk is calculated under the conservative condition that each risk component is independent. The calculation formula is as follows: .

[0043] To highlight the boundaries of high-risk areas, A nonlinear power-order shaping operator is applied to generate the final three-dimensional synthesized anisotropic integrity risk field. The calculation formula is as follows: ; in These are nonlinear power-law shaping coefficients used to suppress low-risk background noise and enhance the boundaries of high-risk regions. The value range is 1 to 5; when As the risk increases, the boundaries of high-risk areas become more prominent, but if If the difference is too large, it may lead to an over-compression of differences in low-risk areas.

[0044] Step 5. Construct a three-dimensional search graph of risk constraints based on the integrated anisotropic integrity risk field. In the process of expanding adjacent nodes, introduce a vertical relaxation coefficient to scale the vertical displacement cost, and combine it with the multiplicative transfer cost function to search for the safe path of the UAV in three-dimensional space.

[0045] The integrated anisotropic integrity risk field generated in step 4 As a priori environmental layer, the passable voxel nodes in the 3D environment map of the urban canyon are used to construct a risk-constrained 3D search graph. ; in This represents the set of accessible voxel nodes. It represents the expandable connection relationship between adjacent nodes (i.e. adjacent passable voxel nodes).

[0046] For any current node The set of its adjacent candidate nodes is denoted as The obstacle's element node or overall anisotropy integrity risk value exceeds a preset risk threshold. The nodes do not participate in the expansion.

[0047] In the A* algorithm from the current node Extend to adjacent nodes At that time, a vertical relaxation coefficient is introduced. Calculate the relaxation distance for reducing the vertical compensation weights. The calculation formula is as follows: ; in , , Representing adjacent nodes respectively With the current node exist , , Coordinate differences in three directions, , , ; , , Each is the current node exist , , Coordinates in three directions, , , Adjacent nodes exist , , Coordinates in three directions; Let represent the vertical relaxation coefficient, and satisfy: .

[0048] To overcome the problem of dimensional conflicts in multi-objective programming, this paper abandons the traditional linear weighting method and constructs a multiplicative transfer cost function based on equivalent resistance amplification. The calculation formula is as follows: ; in As a risk-sensitive weighting factor, Indicates adjacent nodes The overall anisotropy integrity risk value at the location. When adjacent nodes... When in a high-risk area, As the risk increases, the cost of relocation is also amplified, causing path searching to tend to avoid high-risk areas. This forces drones to break free from the "ground-seeking" trap and actively perform "vertical escape" maneuvers to safe high-altitude areas. This determines the tendency of heuristic search to avoid high-risk airspace; the larger the factor, the more conservative the drone's risk avoidance behavior, and its engineering value range is typically [value missing]. .

[0049] During the path search process, the cumulative cost function is updated as follows: ; in , From the starting node to , The cumulative cost.

[0050] Introducing small dynamic weight coefficients into the evaluation function To break the path symmetry in massive 3D meshes and accelerate the generation of globally optimal discrete safe waypoints, the overall evaluation function of the algorithm is obtained. for: ;

[0051] in This represents the distance from the starting node to the node to be evaluated. The cumulative cost; Indicates the node to be evaluated To the target node The estimated remaining cost, In this embodiment .

[0052] The specific search process is as follows: add the starting node to the open list and initialize its cumulative cost; Each time, the node with the smallest overall evaluation function is selected from the open list for expansion; if the current node reaches the target node, the discrete waypoint sequence of the discrete safe path, i.e. the UAV safe path, is generated by backtracking in reverse according to the parent node pointer. Otherwise, traverse its adjacent nodes, remove obstacle object nodes and nodes that exceed the risk threshold, and update the cumulative cost of adjacent nodes and parent node pointers according to the multiplicative transfer cost function described above. Repeat the above process until the target node is found or the open list is empty, thus obtaining a safe path for the drone.

[0053] Step 6. Perform trajectory smoothing on the UAV's safe path to generate a continuous three-dimensional trajectory that satisfies kinematic and dynamic constraints.

[0054] Because the discrete waypoint sequence output by the A* algorithm has defects such as right-angle turns and curvature discontinuities, directly performing low-level flight control can easily lead to trajectory overshoot in multi-rotor UAVs.

[0055] First, the discrete waypoints output by the A* algorithm are extracted as control points. Then, the control point sequence is curve-fitted using cubic B-spline basis functions to eliminate curvature abrupt changes and generate a smooth, continuous three-dimensional trajectory that satisfies the kinematic and dynamic constraints of the UAV, such as maximum speed and acceleration. Finally, the trajectory is output to the UAV flight control system for execution.

[0056] Example 2 This embodiment 2 describes an anisotropic integrity risk modeling and path planning system for urban UAVs, which is based on the same inventive concept as the anisotropic integrity risk modeling and path planning method for urban UAVs in embodiment 1 above.

[0057] A system for modeling the anisotropic integrity risk and planning the path of urban unmanned aerial vehicles (UAVs) includes the following modules: The spatial observation geometric model building module is used to construct a three-dimensional environmental map of the urban canyon in the target flight area and to establish a spatial observation geometric matrix between the UAV and visible satellites. The positioning error establishment module is used to calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and to establish the position error covariance sub-matrix based on the spatial observation geometry matrix and the linear signal quality factor. The anisotropic collision risk model building module is used to perform directional projection on the position error covariance submatrix based on the spatial relationship between the current position of the UAV and the obstacle, obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian overbound envelope model to calculate the anisotropic collision risk. The integrated anisotropic integrity risk field generation module is used to calculate visibility degradation risk and multipath interference risk, and integrates visibility degradation risk, multipath interference risk and anisotropic collision risk to generate an integrated anisotropic integrity risk field. The UAV safe path generation module is used to construct a risk constraint three-dimensional search graph based on the comprehensive anisotropic integrity risk field. During the expansion of adjacent nodes, a vertical relaxation coefficient is introduced to scale the vertical displacement cost, and the UAV safe path is obtained by searching in three-dimensional space in combination with the multiplicative transfer cost function. The trajectory smoothing module is used to smooth the trajectory of the UAV for safe travel, generating a continuous three-dimensional trajectory that meets kinematic and dynamic constraints.

[0058] It should be noted that any content not mentioned in the above-described functional modules of the system described in Embodiment 2 can be referred to the step description of the corresponding method in Embodiment 1 above, and will not be repeated in detail here.

[0059] Example 3 This embodiment 3 describes a computer device including a memory and one or more processors. Executable code is stored in the memory. When the processor executes the executable code, it implements the steps of the urban UAV anisotropic integrity risk modeling and path planning method in embodiment 1 above.

[0060] Example 4 This embodiment 4 describes a computer-readable storage medium storing a program that, when executed by a processor, is used to implement the steps of the urban UAV anisotropic integrity risk modeling and path planning method in embodiment 1 above.

[0061] The computer-readable storage medium can be an internal storage unit of any device or apparatus with data processing capabilities, such as a hard disk or memory, or an external storage device of any device with data processing capabilities, such as a plug-in hard disk, smart media card (SMC), SD card, flash card, etc. Of course, the above description is merely a preferred embodiment of the present invention, and the present invention is not limited to the embodiments listed above. It should be noted that any equivalent substitutions or obvious modifications made by those skilled in the art under the guidance of this specification fall within the scope of this specification and should be protected by the present invention.

Claims

1. A method for anisotropic integrity risk modeling and path planning of urban unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: Step 1. Construct a 3D environmental map of the urban canyon in the target flight area and establish a spatial observation geometry matrix between the UAV and visible satellites; Step 2. Calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and establish the position error covariance submatrix based on the spatial observation geometry matrix and the linear signal quality factor; Step 3. Based on the spatial relationship between the current position of the UAV and the obstacle, perform directional projection on the position error covariance submatrix to obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian overbound envelope model to calculate the anisotropic collision risk. Step 4. Calculate the visibility degradation risk and multipath interference risk, and fuse the visibility degradation risk, multipath interference risk and anisotropic collision risk to generate a comprehensive anisotropic integrity risk field; Step 5. Construct a three-dimensional search graph of risk constraints based on the comprehensive anisotropic integrity risk field. In the process of expanding adjacent nodes, introduce a vertical relaxation coefficient to scale the vertical displacement cost, and combine it with the multiplicative transfer cost function to search for the safe path of the UAV in three-dimensional space. Step 6. Perform trajectory smoothing on the UAV's safe path to generate a continuous three-dimensional trajectory that satisfies kinematic and dynamic constraints.

2. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 1, characterized in that, Step 1 specifically involves: First, acquire the three-dimensional elevation data and building outline data of the target flight area, and convert the building outlines, height information and flight airspace boundaries to the same geographic coordinate system. Then, perform voxelization on the continuous physical space according to the preset spatial resolution to obtain a three-dimensional urban canyon environment map composed of passable voxels, obstacle voxels and boundary voxels. Simultaneously, GNSS satellite ephemeris parameters were obtained, and the positions of each visible satellite in the geocentric-ground-fixed coordinate system were calculated based on Kepler orbital parameters and the WGS84 ellipsoid model. Subsequently, based on the aforementioned GNSS satellite ephemeris parameters and the UAV's current position, the UAV's position was calculated up to the [number missing]. elevation angle of visible satellites With azimuth The calculation formula is as follows: ； ； in Indicates the first The line-of-sight vector of the visible satellite relative to the current position of the UAV in the northeast-northeast celestial coordinate system of the station center. , , , They represent the first The visible satellites are located relative to the drone's current position in the east, north, and sky directions. express The Euclidean norm; Based on this, construct a system containing Space observation geometry matrix of visible satellites Its specific form is: ； Among them, the spatial observation geometric matrix The first three columns of components constitute the unit direction vector pointing to each visible satellite in the station-centered coordinate system, and the fourth column constant 1 corresponds to the clock error state of the receiver.

3. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 2, characterized in that, Step 2 specifically involves: For each satellite line-of-sight ray, the spatial relationship between the ray and building edges, building facades, and obstacle elements in the 3D urban canyon environment map is detected using a 3D ray tracing method. The ray is then used to calculate the... Linear signal quality factor of visible satellite observation channels The calculation formula is as follows: ； in Indicates the first The linear signal quality factor for each visible satellite observation channel is used to characterize the impact of urban buildings on the observation quality of a single channel. ; This indicates the additional loss from Fresnel-Kirchhoff edge diffraction; ； in These are Fresnel-Kirchhoff diffraction parameters. ; ； in The clearance height of the building edge relative to the line connecting the satellite launcher and the drone receiver. and These are the distances from the building's edge to the transmitter and receiver, respectively. This refers to the wavelength of the L1 band signal for the Global Navigation Satellite System. ； in This is the frequency point of the L1 band signal of the Global Navigation Satellite System. The speed of light; Calculate the first [value] affected by elevation angle error amplification and channel linear signal quality factor modulation. Comprehensive observation variance of visible satellite observation channels for: ； in The basic system error constant, These are parameters for amplifying errors related to elevation angle; Furthermore, constructing a system based on The observation weight matrix for diagonal elements The formula is as follows: ； in Represents a diagonal matrix; Based on the weighted least squares estimation theory, the covariance matrix of the UAV's positioning error is calculated. for: ； extract The first three rows and first three columns yield the three-dimensional position error covariance submatrix. .

4. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 3, characterized in that, Step 3 specifically involves: In the 3D environment map of the urban canyon, surface sampling points of solid obstacle boundaries are extracted to form a set of obstacle surface points. and based on Establish a KD-Tree spatial index structure; Based on the current centroid coordinates of the UAV As the query point, a nearest neighbor search is performed in the KD-Tree to obtain the obstacle surface point closest to the current centroid coordinates of the UAV. The calculation formula is as follows: ； in express The coordinates of any candidate obstacle surface point in the data. This represents the coordinates of the obstacle surface point closest to the drone's current centroid coordinates. This represents the value of the variable that minimizes the objective function. Represents the Euclidean distance in three-dimensional space; according to and Construct a unit threat direction vector from the UAV to the nearest obstacle surface point. The calculation formula is as follows: ； The three-dimensional position error covariance submatrix Vector along the unit threat direction Perform a quadratic projection to obtain the variance of the edge localization error in the direction facing the nearest obstacle surface point. The calculation formula is as follows: ； Calculate the Euclidean distance from the current UAV's center of mass to a point on the obstacle surface, and subtract the UAV's physical equivalent safety radius to obtain the environment-driven alarm limit. The calculation formula is as follows: ； in Indicates the physical equivalent safety radius of the drone; A conservative Gaussian over-boundary envelope model is introduced, and the anisotropic collision risk is calculated using the standard complementary error function. The calculation formula is as follows: ； in This represents the anisotropic collision risk and is used to characterize the upper bound of the probability that the actual position of the drone will exceed the environmentally driven alarm limit along the unit threat direction. This represents the standard complementary error function.

5. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 4, characterized in that, Step 4 specifically involves: Risk of visibility degradation Defined as the weighted missing rate of effective signal energy within a locally confined spatial domain, its calculation formula is as follows: ； in This represents an ideal set of visible satellites in open airspace. To prevent extremely small positive numbers with a denominator of zero; For the first Elevation angle weights for each visible satellite observation channel; ;in This indicates taking the maximum value; Count the number of effective non-line-of-sight rays at the current location of the drone. The Poisson survival model is used to map the effective number of non-line-of-sight rays into the risk of multipath interference. The calculation formula is as follows: ； in The preset risk penalty coefficient; Based on the fault tree analysis model, anisotropic collision risk, visibility degradation risk, and multipath interference risk are defined as independent failure events, and the conservative envelope total risk is calculated. The calculation formula is as follows: ； right A nonlinear power-order shaping operator is applied to generate the final three-dimensional synthesized anisotropic integrity risk field. The calculation formula is as follows: ； in These are nonlinear power-law shaping coefficients used to suppress low-risk background noise and enhance the boundaries of high-risk regions.

6. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 5, characterized in that, Step 5 specifically involves: The integrated anisotropic integrity risk field generated in step 4 As a priori environmental layer, the passable voxel nodes in the 3D environment map of the urban canyon are used to construct a risk-constrained 3D search graph. ; in This represents the set of accessible voxel nodes. This represents the expandable connection relationship between adjacent nodes; For any current node The set of its adjacent candidate nodes is denoted as The obstacle's element node or overall anisotropy integrity risk value exceeds a preset risk threshold. The nodes do not participate in the expansion; In the A* algorithm from the current node Extend to adjacent nodes At that time, a vertical relaxation coefficient is introduced. Calculate the relaxation distance for reducing the vertical compensation weights. The calculation formula is as follows: ； in , , Representing adjacent nodes respectively With the current node exist , , The coordinate differences in the three directions; Let represent the vertical relaxation coefficient, and satisfy: ; Constructing a multiplicative transfer cost function based on equivalent resistance amplification The calculation formula is as follows: ； in As a risk-sensitive weighting factor, Indicates adjacent nodes The overall anisotropy integrity risk value at the location; during the path search process, the cumulative cost function is updated as follows: ； in , From the starting node to , The cumulative cost; Introducing small dynamic weight coefficients into the evaluation function The overall evaluation function of the algorithm is obtained. for: ； in This represents the distance from the starting node to the node to be evaluated. The cumulative cost; Indicates the node to be evaluated To the target node The estimated remaining cost, ; The specific search process is as follows: Add the starting node to the open list and initialize its cumulative cost; Each time, the overall evaluation function is selected from the open list. Expand the smallest node; If the current node reaches the target node, then the discrete waypoint sequence of the UAV's safe path is generated by backtracking in reverse according to the parent node pointer; Otherwise, traverse its adjacent nodes, remove obstacle object nodes and nodes that exceed the risk threshold, and update the cumulative cost of adjacent nodes and parent node pointers according to the multiplicative transfer cost function described above. Repeat the above process until the target node is found or the open list is empty, thus obtaining a safe path for the drone.

7. The method for anisotropic integrity risk modeling and path planning of urban UAVs according to claim 6, characterized in that, Step 6 specifically involves: First, the discrete waypoints output by the A* algorithm are extracted as control points. Then, the control point sequence is curve-fitted using cubic B-spline basis functions to generate a smooth, continuous three-dimensional trajectory that satisfies kinematic and dynamic constraints.

8. A system for anisotropic integrity risk modeling and path planning for urban unmanned aerial vehicles (UAVs), characterized in that, Includes the following modules: The spatial observation geometric model building module is used to construct a three-dimensional environmental map of the urban canyon in the target flight area and to establish a spatial observation geometric matrix between the UAV and visible satellites. The positioning error establishment module is used to calculate the linear signal quality factor using the Fresnel-Kirchhoff edge diffraction model, and to establish the position error covariance sub-matrix based on the spatial observation geometry matrix and the linear signal quality factor. The anisotropic collision risk model building module is used to perform directional projection on the position error covariance submatrix based on the spatial relationship between the current position of the UAV and the obstacle, obtain the edge positioning error variance facing the obstacle, and introduce a conservative Gaussian overbound envelope model to calculate the anisotropic collision risk. The integrated anisotropic integrity risk field generation module is used to calculate visibility degradation risk and multipath interference risk, and integrates visibility degradation risk, multipath interference risk and anisotropic collision risk to generate an integrated anisotropic integrity risk field. The UAV safe path generation module is used to construct a risk constraint three-dimensional search graph based on the comprehensive anisotropic integrity risk field. During the expansion of adjacent nodes, a vertical relaxation coefficient is introduced to scale the vertical displacement cost, and the UAV safe path is obtained by searching in three-dimensional space in combination with the multiplicative transfer cost function. The trajectory smoothing module is used to smooth the trajectory of the UAV for safe travel, generating a continuous three-dimensional trajectory that meets kinematic and dynamic constraints.

9. A computer device, comprising a memory and one or more processors; characterized in that, The memory stores executable code, which, when executed by the processor, implements the steps of the urban unmanned aerial vehicle anisotropic integrity risk modeling and path planning method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a program stored thereon; characterized in that, When executed by the processor, the program is used to implement the steps of the urban unmanned aerial vehicle anisotropic integrity risk modeling and path planning method according to any one of claims 1 to 7.

Images