An intelligent composite auxiliary navigation method and device for a full-range unmanned aerial vehicle in a denial environment

Abstract

The application discloses an intelligent composite auxiliary navigation method and device for a full-range unmanned aerial vehicle in a denial environment, and particularly relates to the field of unmanned aerial vehicle navigation. The method comprises the following steps: selecting a plurality of control points in a flight task area; dividing the flight task area and setting a flight mode and a navigation mode for each area; after taking off, acquiring a current flight mode when a satellite signal is abnormal, and determining a navigation mode according to the flight mode; for a scene matching position point obtained through scene matching, correcting the scene matching position point by using the control points to obtain a corrected scene matching position point; determining a position of the unmanned aerial vehicle according to a terrain matching position point and the corrected scene matching position point, and combining navigation by using the position of the unmanned aerial vehicle to assist inertial navigation. Based on the above method, intelligent auxiliary navigation for a full-range unmanned aerial vehicle system can be realized, and the matching rate of the auxiliary navigation means and the geographical environment is improved.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) navigation, and in particular to an intelligent composite assisted navigation method and device for the entire flight of a UAV in a denied environment. Background Technology

[0002] Inertial navigation is commonly used in UAV navigation systems, but it suffers from drift error accumulation. Therefore, it is necessary to use BeiDou satellite positioning data to correct the inertial navigation and improve navigation accuracy. In both civilian and military applications, UAVs face diverse and complex environments, such as hilly terrain, mountainous areas, and satellite-denied environments on battlefields, which may cause BeiDou positioning to fail or become unstable.

[0003] To meet the autonomous flight requirements of UAVs in this special environment, existing methods employ terrain matching or scene matching navigation. However, for different regions, using only one matching navigation mode (terrain matching or scene matching) is not suitable for matching navigation in complex geographical environments, resulting in low matching rates or failure to match successfully. Summary of the Invention

[0004] The main objective of this application is to provide an intelligent composite assisted navigation method and device for the entire flight of an unmanned aerial vehicle (UAV) in a denied environment, aiming to solve the problem of low matching rate between navigation modes and geographical environment in the existing technology.

[0005] To achieve the above objectives, this application provides an intelligent composite assisted navigation method for the entire flight path of an unmanned aerial vehicle (UAV) in a denied environment, comprising: acquiring an electronic map of the UAV's flight mission area, including an elevation map and a visible light map; selecting multiple control points in the flight mission area and establishing a control point calibration data table based on the position and altitude of all control points; dividing the flight mission area into mountainous areas, canyon areas, and urban areas based on the terrain features of the flight mission area; setting the UAV's flight mode in the mountainous and canyon areas to low-altitude penetration mode and the navigation mode to terrain matching mode; setting the UAV's flight mode in the urban area to high-altitude cruise mode and the navigation mode to scene matching mode; and when it is determined that the UAV is in flight and the satellite signal is abnormal, acquiring the current flight mode... The method involves the following steps: When the flight mode is determined to be high-altitude cruise mode, the current flight scene image of the UAV is acquired; the area to which the current flight scene belongs is determined based on the flight scene image; if the flight scene belongs to a mountainous or canyon area, terrain matching is performed using the elevation information of the current area and an elevation map to obtain terrain matching location points; if the current flight scene is determined to be an urban area, scene matching is performed using a visible light map and the flight scene image to obtain scene matching location points; the scene matching location points are used to search for associated control points in the control point calibration database, and the scene matching location points are corrected using the control points to obtain corrected scene matching location points; based on the terrain matching location points and the corrected scene matching location points, the position of the UAV is determined, and the position of the UAV is used to assist in inertial navigation for integrated navigation.

[0006] Optionally, after obtaining the current flight mode, the method further includes: when the flight mode is determined to be a low-altitude penetration mode, using the elevation information of the current area to perform terrain matching with the elevation map to obtain the terrain matching location point.

[0007] Optionally, multiple control points are selected in the flight mission area, and a control point calibration data table is established based on the position and altitude of all control points. This includes: selecting multiple types of control points in the flight mission area and calibrating the position and altitude of each control point; numbering each control point and establishing a control point calibration data table based on the control point number, type, position, and altitude.

[0008] Optionally, the scene matching location point is corrected using control points, including: using the scene matching location point in combination with the positioning error to search for associated control points in the control point calibration database; and using the control points to replace the corresponding scene matching location point to obtain the corrected scene matching location point.

[0009] Optionally, control points include typical buildings, vehicles, and crosshairs.

[0010] Optionally, scene matching is performed on the visible light map and the flight scene image to obtain scene matching location points, including: determining the search area of ​​the visible light map based on the inertial navigation error radius, and using the electronic map of the search area as a reference image; performing orthophoto correction on the current flight scene image to obtain a corrected image; extracting features from the reference image and the corrected image, and performing feature matching to obtain matching features, and determining the mapping relationship between the reference image and the corrected image based on the matching features; and determining the matching location points between the reference image and the corrected image based on the mapping relationship, which are then used as scene matching location points.

[0011] Optionally, terrain matching can be performed using the elevation information of the current area and the elevation map to obtain terrain matching location points, including: determining the search area of ​​the elevation map based on the inertial navigation error radius; obtaining altitude information for a preset time period during the cruise through the barometric altimeter or lidar equipment of the UAV, and determining the terrain outline based on the altitude information; and matching the search area of ​​the elevation map and the terrain outline to obtain terrain matching location points.

[0012] Optionally, the search area of ​​the elevation map and the terrain contour are matched to obtain the terrain matching location points; including: using the cross-correlation matching method to find the matching points between the search area of ​​the elevation map and the terrain contour; and using the RANSAC method to detect and remove mismatches of the matching point pairs to obtain the terrain matching location points.

[0013] To achieve the above objectives, this application also provides an intelligent composite auxiliary navigation device for the entire flight path of an unmanned aerial vehicle (UAV) in a denied environment, comprising: a map acquisition module for acquiring an electronic map of the flight mission area, including an elevation map and a visible light map; a data table creation module for selecting multiple control points on the electronic map and creating a control point calibration data table based on the position and altitude of all control points; a region division module for dividing the flight mission area into mountainous areas, canyon areas, and urban areas based on the terrain features of the flight mission area; a mode preset module for setting the UAV's flight mode to low-altitude penetration mode and navigation mode to terrain matching mode in mountainous and canyon areas; and setting the UAV's flight mode to high-altitude cruise mode and navigation mode to scene matching mode in urban areas; and an image acquisition module for acquiring images when it is determined that the UAV is in flight and satellite signals are abnormal. The current flight mode, when determined to be high-altitude cruise, acquires the current flight scene image of the UAV; the terrain matching module determines the area to which the current flight scene belongs based on the flight scene image; when the flight scene belongs to a mountainous or canyon area, it uses the elevation information of the current area to perform terrain matching with an elevation map to obtain a terrain matching location point; the scene matching module, when determined to be an urban area, performs scene matching between a visible light map and the flight scene image to obtain a scene matching location point; the control point correction module, using the scene matching location point, searches for associated control points in the control point calibration database and uses the control points to correct the scene matching location point to obtain a corrected scene matching location point; the navigation module, based on the terrain matching location point and the corrected scene matching location point, determines the UAV's position and uses the UAV's position to assist inertial navigation for integrated navigation.

[0014] Compared with the prior art, the beneficial effects of this application are as follows:

[0015] The present invention provides an intelligent composite assisted navigation method for unmanned aerial vehicles (UAVs) throughout their entire flight in a satellite-denied environment. Under satellite-denied conditions, this method utilizes information such as flight altitude, scene recognition, and control point recognition to automatically select assisted navigation methods such as terrain matching, scene matching, and high-precision control point positioning during the UAV's mission execution. This achieves intelligent assisted navigation for the entire flight of the UAV system, improving the matching rate between assisted navigation methods and the geographical environment. The multiple assisted navigation methods overcome the limitations of traditional single assisted navigation methods, enhancing the flight safety of the UAV system under satellite-denied conditions throughout its entire flight.

[0016] Control point identification is performed on the scene matching location points, and the scene matching results are corrected using the control points. The control point positions are used as an auxiliary means to improve navigation accuracy, avoiding image matching errors caused by artificially increasing errors from the electronic map acquisition channels. This ensures scene matching accuracy and greatly improves navigation and positioning accuracy.

[0017] When UAVs use terrain matching for assisted navigation, they collect aircraft altitude information through the aircraft's barometric altimeter and radio altimeter to measure terrain contours, eliminating the need for additional elevation measurement equipment. When using scene matching for assisted navigation, there is no need for an additional imaging detection system. Instead, the onboard optoelectronic payload is used to acquire reconnaissance images simultaneously during mission execution for scene matching. Auxiliary navigation is achieved without the need for additional onboard equipment, significantly reducing system integration costs.

[0018] When terrain matching is used, features such as terrain outlines are relatively stable and not easily affected by factors such as seasons and climate. The pre-stored terrain data has good time-phase properties, which improves the reliability of the UAV navigation system. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating an intelligent composite assisted navigation method for the entire flight path of an unmanned aerial vehicle (UAV) under denied conditions, as proposed in this application.

[0020] Figure 2 for Figure 1 A detailed flowchart illustrating the terrain matching process;

[0021] Figure 3 for Figure 1 A detailed flowchart illustrating the scene matching process;

[0022] Figure 4 This is a location point map of the airport area obtained in the example.

[0023] Figure 5 The image shows the location points of the urban scene obtained in the example.

[0024] Figure 6 The terrain matching error map is obtained from the example.

[0025] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0027] The first embodiment of the present invention provides an intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle (UAV) in a denied environment, such as... Figure 1 As shown, the specific steps include:

[0028] Step S1: Obtain an electronic map of the flight mission area, including an elevation map and a visible light map;

[0029] Specifically, before flight, an electronic map of the corresponding location is cropped based on the drone's flight path. After the drone is powered on the ground, the cropped electronic map is loaded via the network port. For example, a satellite map of the flight mission area is downloaded from a professional GIS website. Due to the limited storage capacity of the drone's onboard equipment, only the map size of the flight mission area is cropped. The resolution of the satellite map determines the accuracy of scene matching and positioning; the higher the map level, the smaller the spatial resolution and the higher the accuracy. The specific download method is as follows:

[0030] Select Google Maps as the electronic map. Based on the center point coordinates and area size of the flight mission area, determine the coordinates of the four corners of the selection box, namely the coordinates of the northwest corner, northeast corner, southwest corner, and southeast corner. After determining the four corner coordinates, click the download menu and select the area to download. The calculation method for the four corner coordinates of the selection box is as follows.

[0031] Given: coordinates of the center point ( λ c φ c ),in λ c Longitude φ c Latitude; side length L of the rectangular task area (unit: meters);

[0032] Latitude (north-south), each degree is approximately equal to 111,000 meters; Precision (east-west), each degree is approximately equal to 111,000 × cos( φ c )rice;

[0033] Latitude difference Δλ = L / 111000; Accuracy difference Δφ = L / 111000 × cos( φ c );but:

[0034] Northwest corner coordinates:

[0035]

[0036] Northeast corner coordinates:

[0037]

[0038] Southwest corner coordinates:

[0039]

[0040] Southeast corner coordinates:

[0041]

[0042] Export the selected area map as a map file: Set the specific format for the exported map, set the map level, generally choose level 17 (scale 1:9.03k, spatial resolution 2.4m), and also select TIFF format for saving the file and WGS84 (World Geodetic System 1984) for the coordinate projection; obtain the TIFF format map file of the task area by downloading it; after powering on the drone, load the map file to the drone's onboard equipment via the network port.

[0043] Step S2: Select multiple control points in the flight mission area and establish a control point calibration data table based on the location and altitude of all control points; among them, control points may include target features, such as typical buildings, vehicles, and crosshairs.

[0044] Specifically, multiple types of control points are selected within the flight mission area, and the position and altitude of each control point are calibrated. For example, RTK (Real-Time Kinematic) standard equipment can be used to calibrate the position and altitude of control points. Each control point is numbered, and a control point calibration data table is established based on its number, type, position, and altitude to create a one-to-one correspondence between the control point number / type / position / altitude. This table is used for control point identification. The control point calibration data table is bound to the onboard device's memory via a network cable, and a lightweight SQLite database is used on the embedded onboard device for CRUD operations. Specific processing methods are as follows:

[0045] Create a table of control points, containing control point ID, control point type, control point longitude, control point latitude, and control point altitude:

[0046] CREATE TABLE C-POINT (

[0047] TargetID INTEGER PRIMARY KEY

[0048] TargetTypeTEXT NOT NULL

[0049] Target Longitude Double

[0050] TargetLatitudeDouble

[0051] TargetHighint)

[0052] Inserting data makes it easier to add new control points:

[0053] INSERT INTO C-POINT (TargetType, targetLongitude, TargetLatitude, TargetHigh)

[0054] To query data, find the longitude and latitude of a control point by its control point number:

[0055] SELECT TargetLongitude FROMC-POINT WHERE TargetID=condition.

[0056] Update the data after the control point's ID, type, latitude, longitude, or altitude changes:

[0057] UPDATE C-POINT SET column1=newvalue1, column2=newvalue2, column3=newvalue3, column4=newvalue4 WHERE condition;

[0058] Deleting data: After control points are calibrated, some control point data needs to be deleted.

[0059] DELETEFROMC-POINT WHERE condition.

[0060] Step S3: Based on the topographic features of the flight mission area, the flight mission area is divided into mountainous areas, canyon areas, and urban areas.

[0061] Step S4: Set the drone's flight mode to low-altitude penetration mode and navigation mode to terrain matching mode in mountainous and canyon areas; set the drone's flight mode to high-altitude cruise mode and navigation mode to scene matching mode in urban areas.

[0062] Specifically, in mountainous or canyon areas lacking typical buildings, the flight mode is set to low-altitude penetration mode, and the navigation mode is terrain-matching mode. If the area is an urban area with typical buildings, overpasses, and highways, the flight mode is set to high-altitude cruise mode, and the navigation mode is scene-matching mode.

[0063] Step S5: When it is determined that the UAV is in flight and the satellite signal is abnormal, the current flight mode is obtained. When the flight mode is determined to be low-altitude penetration mode, terrain matching is performed using the current area's elevation information and elevation map to obtain the terrain matching location point. When the flight mode is determined to be high-altitude cruise, the current flight scene image of the UAV is obtained.

[0064] Specifically, the UAV's optoelectronic payload operates under a wide field of view to obtain reconnaissance image information with a larger width, making it easier to obtain more terrain features; therefore, this embodiment utilizes the UAV's optoelectronic payload to acquire images of the current flight scene.

[0065] Step S6: Determine the area to which the current flight scene belongs based on the flight scene image; when the flight scene belongs to a mountainous area or a canyon area, use the elevation information of the current area to perform terrain matching with the elevation map to obtain the terrain matching location point;

[0066] Specifically, the image processing component onboard the UAV system can perform image processing. For example, the image processing component can be the YOLO (You Only Look Once) v5 single-stage target detection algorithm. This algorithm performs target recognition on flight scene images, omitting the candidate region generation stage, and directly obtaining target classification and location information (region). For example, the YOLOv5 network structure consists of four parts:

[0067] The image input terminal is used to preprocess the input image, including operations such as scaling and normalization. Assuming Iraw is the original image and X is the preprocessed input tensor, then... The Preprocess function includes operations such as resizing, color channel conversion, and normalization.

[0068] The backbone network is used to extract feature maps F from the preprocessed image X. The forward propagation process of the backbone network can be represented as follows: The Backbone function represents the operations performed on the backbone network, including convolution, batch normalization, and activation functions.

[0069] The neck network is used to fuse multi-scale features from the backbone network output, enhancing the network's ability to detect targets of different sizes. Assuming the input feature map to the neck network is Fi (i=3, 4, 5) and the output feature map is Pi, the forward propagation process of the neck network is as follows: The Neck function represents the operation of the neck network, which includes feature fusion of FPN and PANet. Finally, the neck network outputs feature maps at three scales, which are used to detect targets in the head.

[0070] The head network is used to predict the bounding box, class, and confidence score of the target based on the fused multi-scale feature map. Specifically, let the input feature map of the head network be Fi (i=3, 4, 5), and the output target detection result be Y. Then the forward propagation process of the head network can be represented as: The DetectionHead function represents the operations of the detection head, including feature fusion and prediction.

[0071] The image processing component onboard the UAV system identifies and classifies the terrain features of the flight scene images to determine whether the current flight scene belongs to a mountainous area, canyon area, or urban area. Further, the terrain matching method is as follows: the search area of ​​the elevation map is determined based on the inertial navigation error radius; altitude information during a preset time period is obtained through the UAV's barometric altimeter or lidar equipment, and the terrain contour is determined based on the altitude information; a cross-correlation matching method is used to find matching points between the search area of ​​the elevation map and the terrain contour; and the RANSAC method is used to detect and remove mismatches in the matching point pairs to obtain the terrain matching location points.

[0072] In this embodiment, when the UAV uses terrain matching assisted navigation, it can use the aircraft's barometric altimeter and radio altimeter without the need for additional elevation measurement equipment. During the UAV's flight, the terrain contour is measured using the aircraft's altitude information acquired during the flight. This achieves assisted navigation without the need for additional airborne equipment, greatly saving system integration costs.

[0073] The cross-correlation matching method in this embodiment calculates the cross-correlation value with real-time terrain data (terrain contour) at various locations by sliding the reference terrain window (search area of ​​the elevation map), and finds the location with the largest cross-correlation value, which is the optimal matching location. To improve the robustness of the matching, a normalized cross-correlation function is used.

[0074]

[0075] in, NCC (m, n) represents the cross-correlation value at location (m, n); f(i, j) represents the elevation value of the reference terrain window at location (i, j); The reference terrain window is the true value; g(i+m, j+n) is the elevation value of the real-time terrain data at the location (i+m, j+n); is the mean value of the region corresponding to the real-time terrain data; M and N are the width and height of the window, respectively.

[0076] RANSAC is a robust model parameter estimation algorithm used to estimate mathematical models from data containing a large amount of noise and outliers. In terrain matching, RANSAC is used to find the optimal transformation model. The general process is as follows: randomly select n pairs of corresponding points from the dataset to form an initial sample subset; use the selected sample subset to fit the initial transformation model (usually a rigid body transformation matrix); calculate the distance from all data points to the fitted model, and count the points whose distance is less than a threshold, these points are considered inliers; if the current number of inliers is greater than the historical best value, update the optimal model and the inlier set; obtain the preset maximum number of iterations; refit the model using all the generated inliers to improve matching accuracy.

[0077] It is understood that the inertial navigation error radius involved in this embodiment is determined based on the data of the UAV's inertial navigation system. The data includes information such as the UAV's position, heading, speed, and acceleration. The specific method for determining the inertial navigation error radius adopts conventional methods in the art and is not an improvement of this invention, so it will not be described in detail here.

[0078] The specific method for determining the search area of ​​the elevation map based on the inertial navigation error radius is as follows: taking the position calculated by the inertial navigation system at time T0 as the center, and combining it with the error radius R, a range (i.e., the search area) is delineated on the elevation map.

[0079] Step S7: When the current flight scene is determined to be an urban area, scene matching is performed on the visible light map and the flight scene image to obtain the scene matching location point;

[0080] Specifically, the scene matching method is as follows: The search area of ​​the visible light map is determined based on the inertial navigation error radius, and the electronic map of the search area is used as the reference image; orthophoto correction is performed on the current flight scene image to obtain a corrected image; features are extracted from the reference image and the corrected image, and feature matching is performed to obtain feature matching points; the RANSAC (RANdom Sampling Consensus) method is used to detect and remove mismatches in the matching feature pairs; the mapping relationship between the reference image and the corrected image is determined based on the removed matching features; and the matching position points between the reference image and the corrected image are determined based on the mapping relationship, serving as the scene matching position points.

[0081] The method for determining the mapping relationship is as follows:

[0082] The pixel coordinates of the feature matching points in the corrected image are transformed into pixel coordinates → image coordinates → camera coordinates → body coordinates → geographic coordinates to obtain the geographic coordinates of the point, i.e., the longitude and latitude of the point, denoted as (x, y); the coordinates of the corresponding feature matching points in the reference image are obtained from the longitude and latitude of the point through an electronic map, denoted as (x', y'). Then:

[0083]

[0084] Wherein, the homography matrix H is a 3×3 matrix, represented as:

[0085]

[0086] Based on the matching point pairs, construct a linear equation:

[0087]

[0088]

[0089] During the matching process, at least four pairs of feature matching points are selected to solve the homography matrix. Based on the homography matrix, the scene matching position points in the reference image and the corrected image can be obtained.

[0090] For example, commonly used feature matching algorithms include SIFT (Scale-Invariant Feature Transform), SURF (Speeded-Up Robust Features), ORB (Oriented Fast and Rotated BRIEF), and artificial intelligence algorithms based on neural networks. Based on the computing power and real-time requirements of UAV-borne equipment, this embodiment selects ORB to achieve fast and efficient algorithm implementation, and feature extraction and matching with rotation invariance.

[0091] Understandably, the specific method for determining the search area of ​​the visible light map based on the inertial navigation error radius is to take the position calculated by the inertial navigation system at time T0 as the center and, in combination with the error radius R, delineate a range (i.e., the search area) in the visible light map.

[0092] Step S8: Search for associated control points in the control point calibration database using the scene matching location points, and use the control points to correct the scene matching location points to obtain corrected scene matching location points;

[0093] Specifically, by combining the scene matching location points with the positioning error, a search is conducted in the control point calibration database for associated control points. These control points are then used to replace the corresponding scene matching location points, resulting in corrected scene matching location points. The positioning error is typically 30 mCEP, meaning that for each scene matching location point, after adding the positioning error, a search is conducted in the control point calibration database for the corresponding control point, and this control point is used to replace the corresponding scene matching location point, thus obtaining the corrected scene matching location point.

[0094] In this embodiment, control point identification is performed on the scene matching location points, and the scene matching results are corrected using the control points. The control point positions are used as a means to improve navigation accuracy, avoiding image matching errors caused by artificially amplified errors from referencing electronic maps. This ensures scene matching accuracy and significantly improves navigation and positioning accuracy. When using scene matching for assisted navigation, no additional imaging detection system is needed. The airborne optoelectronic payload is used, which simultaneously acquires reconnaissance images during its mission for scene matching.

[0095] Step S9: Based on the terrain matching location points and the scene matching location points, determine the position of the UAV, and use the position of the UAV to assist the inertial navigation system for integrated navigation.

[0096] Specifically, in the corresponding region or scenario, a set of matching location points is obtained. For example, in high-altitude cruise mode within an urban area, if control points are present, a set of control point calibration positions can be obtained. Using trigonometric relationships, the coordinates of the point to be determined are calculated by the angles or distances between the UAV position and multiple control point calibration positions: Take three control point calibration positions, such as points A (x1, y1), B (x2, y2), and C (x3, y3), and the UAV coordinates P (x, y). Angles α (field of view between PA and PB), β (field of view between PB and PC), and γ (field of view between PC and PA) are observed at point P. These angles are calculated using the load and the UAV's attitude angles. Based on the trigonometric relationships, a system of equations is established:

[0097] , ,

[0098] Since α, β, and γ are known, the positioning coordinates of point P of the UAV can be obtained by solving the above system of equations.

[0099] In this embodiment, under satellite denial conditions, information such as flight altitude, scene recognition, and control point recognition is used to automatically select auxiliary navigation methods such as terrain matching, scene matching, and high-precision control point positioning during UAV operations, achieving intelligent auxiliary navigation for the entire flight of the UAV system. Multiple auxiliary navigation methods overcome the limitations of traditional single auxiliary navigation methods and solve the flight safety issues of the UAV system under satellite denial conditions throughout the entire flight. Considering the limited resources of UAV-borne equipment, lightweight and real-time considerations are fully taken into account in the implementation of algorithms such as target recognition and scene matching, allowing for direct implementation as software modules on the airborne embedded platform, making it convenient and practical.

[0100] The navigation method of this invention was used for flight verification, as detailed below.

[0101] Example 1

[0102] In flight path airport areas and urban scenarios, the scene matching location accuracy of this embodiment is shown in Table 1:

[0103] Table 1 Scene Matching and Location Data

[0104]

[0105] As shown in Table 1, the error CEP between the ground matching point calibration value (true value) and the map matching point corresponding value (test value) is 21.9m, which is the scene matching accuracy.

[0106] See the matching point pairs between the scene images and electronic maps of the airport area. Figure 4The matching point pairs between the scene image and the electronic map are shown in Figure 5. In the figure, the left side shows the map image and the right side shows the scene image. The matching point pairs are connected by lines of different colors. As can be seen from the figure, the number of matching point pairs in the two scenes is much greater than 4, indicating that they can match well.

[0107] Based on scene matching, control points were used to correct the scene, and the correction results are shown in Table 2.

[0108] Table 2 Comparison of positioning errors before and after control point correction

[0109]

[0110] Table 2 shows the following for each scene matching location point pair: 1) the error between the ground matching point calibration value (true value) and the corresponding value (test value) of the scene matching location point is calculated; 2) the error between the ground matching point calibration value and the positioning value after control point correction is calculated. By comparing the two error values, it can be seen that the positioning accuracy after control point correction is improved.

[0111] The terrain matching performance of this embodiment in mountainous and hilly areas (the error between the matching result and the inertial navigation result) is shown in [link to documentation]. Figure 6 As can be seen from the figure, terrain matching is performed once per second, and the error of most matching points is controlled within 40 meters. This shows that the present invention can reduce positioning errors and improve navigation accuracy when performing terrain matching in mountainous and hilly areas.

[0112] The second embodiment of the present invention provides an intelligent composite auxiliary navigation device for a UAV's entire flight range in a denied environment, comprising: a map acquisition module for acquiring an electronic map of the flight mission area, including an elevation map and a visible light map; a data table establishment module for selecting multiple control points on the electronic map and establishing a control point calibration data table based on the position and altitude of all control points; a region division module for dividing the flight mission area into mountainous areas, canyon areas, and urban areas based on the terrain features of the flight mission area; a mode preset module for setting the UAV's flight mode to low-altitude penetration mode and navigation mode to terrain matching mode in mountainous and canyon areas; and setting the UAV's flight mode to high-altitude cruise mode and navigation mode to scene matching mode in urban areas; and an image acquisition module for acquiring images when it is determined that the UAV is in flight and the satellite signal is abnormal. The system comprises the following modules: First, a high-altitude cruise flight mode is selected. The first module acquires an image of the current flight scene. A terrain matching module determines the area of ​​the current flight scene based on the image. If the flight scene is in a mountainous or canyon area, it performs terrain matching between the elevation information of the current area and an elevation map to obtain a terrain-matched location point. A scene matching module performs scene matching between a visible light map and the flight scene image when the current flight scene is determined to be an urban area, obtaining a scene-matched location point. A control point correction module searches for associated control points in a control point calibration database using the scene-matched location point and corrects the scene-matched location point to obtain a corrected scene-matched location point. Finally, a navigation module determines the drone's position based on the terrain-matched location point and the corrected scene-matched location point, and uses the drone's position to assist inertial navigation for integrated navigation.

[0113] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A smart composite assisted navigation method for the entire flight path of an unmanned aerial vehicle (UAV) in a denied environment, characterized in that, include: Obtain electronic maps of the UAV's flight mission area, including elevation maps and visible light maps; Multiple control points are selected in the flight mission area, and a control point calibration data table is established based on the position and altitude of all control points. Based on the topographic features of the flight mission area, the flight mission area is divided into mountainous areas, canyon areas, and urban areas. Set the drone's flight mode to low-altitude penetration mode and navigation mode to terrain matching mode when flying in mountainous and canyon areas; set the drone's flight mode to high-altitude cruise mode and navigation mode to scene matching mode when flying in urban areas. When it is determined that the drone is in flight and the satellite signal is abnormal, the current flight mode is obtained. When the flight mode is determined to be high-altitude cruise mode, the current flight scene image of the drone is obtained. The current flight scene area is determined based on the flight scene image; when the flight scene belongs to a mountainous area or a canyon area, the terrain is matched with the elevation map using the elevation information of the current area to obtain the terrain matching location point; When the current flight scenario is determined to be an urban area, scene matching is performed on the visible light map and the flight scenario image to obtain the scene matching location point; The scene matching location point is used to search for associated control points in the control point calibration database, and the scene matching location point is corrected using the control points to obtain the corrected scene matching location point; Based on the terrain matching location points and the corrected scene matching location points, the position of the UAV is determined, and the position of the UAV is used to assist the inertial navigation system for combined navigation. The process involves selecting multiple control points within the flight mission area and establishing a control point calibration data table based on the location and altitude of all control points, including: Select multiple types of control points in the flight mission area and calibrate the position and altitude of each control point; Each control point is numbered, and a control point calibration data table is established based on the control point's number, type, location, and altitude. The step of correcting the scene matching location point using control points includes: Using the scene matching location points combined with the positioning error, search for associated control points in the control point calibration database; By replacing the corresponding scene matching location points with control points, the corrected scene matching location points are obtained.

2. The intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle in a denied environment according to claim 1, characterized in that, After obtaining the current flight mode, the method further includes: when the flight mode is determined to be a low-altitude penetration mode, using the elevation information of the current area to perform terrain matching with the elevation map to obtain the terrain matching location point.

3. The intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle in a denied environment according to claim 1, characterized in that, The control points include typical buildings, vehicles, and crosshairs.

4. The intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle in a denied environment according to claim 1, characterized in that, The step of performing scene matching between the visible light map and the flight scene image to obtain scene matching location points includes: The search area of ​​the visible light map is determined based on the inertial navigation error radius, and the electronic map of the search area is used as a reference image. Perform orthophoto correction on the current flight scene image to obtain the corrected image; Feature extraction and feature matching are performed on the reference image and the calibration image to obtain matching features, and the mapping relationship between the reference image and the calibration image is determined based on the matching features. Based on the mapping relationship, the matching position points of the reference image and the corrected image are determined as scene matching position points.

5. The intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle in a denied environment according to claim 1, characterized in that, The step of matching the terrain with the elevation information of the current area and the elevation map to obtain the terrain matching location point includes: The search area of ​​the elevation map is determined based on the inertial navigation error radius; the altitude information during the cruise is obtained through the barometric altimeter or lidar equipment of the UAV for a preset time period, and the terrain outline is determined based on the altitude information; The search area and terrain contour of the elevation map are matched to obtain the terrain matching location points.

6. The intelligent composite assisted navigation method for the entire flight of an unmanned aerial vehicle in a denied environment according to claim 1, characterized in that, The process of matching the search area and terrain contour of the elevation map to obtain terrain matching location points includes: The cross-correlation matching method is used to find the matching points between the search area of ​​the elevation map and the terrain contour; and the RANSAC method is used to detect and remove mismatches of the matching point pairs to obtain the terrain matching location points.

7. An intelligent composite auxiliary navigation device for a UAV in a denied environment, employing the intelligent composite auxiliary navigation method for the entire flight of a UAV in a denied environment as described in claim 1, characterized in that, include: The map acquisition module is used to acquire electronic maps of the flight mission area, including elevation maps and visible light maps; The data table creation module is used to select multiple control points on the electronic map and create a control point calibration data table based on the position and altitude of all control points. The region division module is used to divide the flight mission area into mountainous areas, canyon areas, and urban areas based on the terrain features of the flight mission area. The mode preset module is used to set the drone's flight mode to low-altitude penetration mode and navigation mode to terrain matching mode in mountainous and canyon areas; and to set the drone's flight mode to high-altitude cruise mode and navigation mode to scene matching mode in urban areas. The image acquisition module is used to acquire the current flight mode when it is determined that the UAV is in flight and the satellite signal is abnormal; and to acquire the current flight scene image of the UAV when the flight mode is determined to be high-altitude cruise. The terrain matching module determines the region to which the current flight scene belongs based on the flight scene image; when the flight scene belongs to a mountainous region or a canyon region, it uses the elevation information of the current region to perform terrain matching with the elevation map to obtain the terrain matching location point. The scene matching module is used to perform scene matching between the visible light map and the flight scene image when the current flight scene is determined to be an urban area, so as to obtain the scene matching location point; The control point correction module is used to search for associated control points in the control point calibration database using the scene matching location points, and to correct the scene matching location points using the control points to obtain corrected scene matching location points. The navigation module is used to match location points based on the terrain, correct the scene matching location points, determine the position of the UAV, and use the position of the UAV to assist the inertial navigation system for combined navigation.

Images