[0037] The method of the present invention first performs oblique photographic modeling on the path planning area of the UAV, and then based on the 2D and 3D integration technology, such as figure 1 As shown, the initial path of the UAV that satisfies the kinematic constraints is planned in the two-dimensional map window, and at the same time, the oblique photographic model is realized in the three-dimensional full-element real-scene map window with the visibility algorithm and the limited surface analysis algorithm in the GIS aerial analysis function The three-dimensional path of the UAV in satisfies the security constraints, and the specific implementation steps are as follows:
[0038] (1) Oblique photographic modeling of the UAV path planning area
[0039] With the development of photogrammetry and computer graphics, "digital earth" has become a trend and gradually formed. The current oblique photography technology realizes rapid real-world modeling of the survey area through multi-view overlapping photos collected by drones, and software such as Context Capture (Smart3D) even realizes automatic and rapid full-element real-world modeling without control points. mold. Its full-element reality modeling process mainly includes four processes: feature point recognition, dense point cloud generation, irregular TIN grid construction, and texture automatic mapping, such as figure 2 As shown, the specific process is as follows:
[0040] (1.a) Data preparation: use drones to collect multi-view aerial images and POS data in the planning area;
[0041] (1.b) Dense point cloud generation: According to the principles of photogrammetry and computer graphics, the feature points of multi-view aerial images collected by drones are identified;
[0042] (1.c) Irregular TIN grid construction: construct irregular TIN grid through aerial triangulation calculation;
[0043] (1.d) Full-element reality model generation: Obtain an oblique photographic model, that is, a full-element reality model, through automatic texture mapping.
[0044] (2) Plan the initial UAV path and height attribute value in the two-dimensional window
[0045] (2.a) Data preparation: Manually plan waypoints in the 2D window and obtain initial 2D coordinates a i (x i ,y i ) and the height attribute value is hi;
[0046] (2.b) Generate the initial two-dimensional path of the UAV: plan the initial UAV path that satisfies kinematics (curvature) constraints in the two-dimensional map window (LaValle S M. Planning Algorithms. Cambridge University Press; 2006.p.5 -17.), the horizontal direction can meet the turn rate constraint by adjusting the position of the waypoint on the two-dimensional map, and the vertical direction can satisfy the pitch rate constraint by changing the height value of one of the waypoints, such as image 3 Shown, where the waypoint a i (x i ,y i ), the height attribute value is hi.
[0047] (3) Use the full-element real-scene model to generate a three-dimensional full-element real-scene map window, and judge whether the waypoint in the three-dimensional window is outside the building and other ground objects based on the GIS limit surface analysis algorithm
[0048] (3.a) Limiting surface analysis: through the four directions of the UAV's path direction in the three-dimensional full-element real-scene map window, the upper, lower, front and rear directions are analyzed;
[0049] (3.b) Determine whether the waypoint is outside the building: the waypoint a in the two-dimensional map window i (x i ,y i ), height value h i Perform coordinate system conversion and projection transformation into three-dimensional coordinates A i (X i ,Y i ,H i ), and judge the height H of the waypoint in the three-dimensional full-element real scene map window i Is it greater than the height H of the building at this location i限 ,like Figure 4 shown. If H iH i限 , then the path requirements of the initial UAV are satisfied, and the step (4) is continued; i such that H in the 3D window i The value is greater than the limit face height.
[0050] (4) Calculate the visibility between waypoints in the three-dimensional full-element real-scene map window based on the GIS visibility algorithm
[0051] (4.a) Calculation of line-of-sight distance: Calculate the line-of-sight distance between two waypoints according to the line-of-sight algorithm, such as Figure 5 As shown, the formula is as follows:
[0052]
[0053] In the formula, D 2 is the obstacle to the waypoint P s and waypoint P t The line-of-sight distance between, h 1 、h 2 P respectively s and P t The height of the point from the ground.
[0054] (4.b) Judgment of visibility between waypoints: Judgment A i with A i+1 Visibility between waypoints, if there is visibility, then a in the two-dimensional map i →a i+1 The path meets the security constraints, that is, the position and height attribute values in the two-dimensional window meet the requirements, and continue to calculate A i+1 with A i+2 Visibility between waypoints; otherwise, maintain A i The height value remains unchanged, and continue to make the following judgments:
[0055] 1) If waypoint A i the height H i < Waypoint A i+1 the height H i+1 , ① and A i with A i+1 There is no ratio between θ 1 Larger elevation angle θ i , then A i+1 Waypoint altitude increase value △H i+1 for Y Ai+1 -Y Ai; ② If there is A i →A i+1 Elevation angle θ 1 bigger θ i , then take A i → New A i+1 ’Elevation angle θ i Path Calculation △H Determined by Maximum Value i+1 , the formula is as follows:
[0056]
[0057]
[0058] In the formula, X Ai→Ai+1 , Y Ai→Ai+1 for A i →A i+1 The X coordinate value and Y coordinate value of the obstacles between them in the three-dimensional full-element real scene map window; △H i+1 for A i+1 Altitude value added to the waypoint.
[0059] i.e. with A i with A i+1 The maximum visibility elevation angle of the projected area between to determine the safe path of the UAV, such as Image 6 as shown in (a);
[0060] 2) If H iH i+1 , similarly determine A with the minimum depression angle i+1 Increased height values, such as Image 6 In (b) shown.
[0061] (5) Judging whether the new waypoint is outside the building and other ground objects in the three-dimensional full-element real-scene map window
[0062] For new waypoint A i+1 (X i+1 ,Y i+1 ,H i+1 +△H i+1 ) Use the method of step (3) to judge whether it is outside the building and other features, if it is outside, then A i+1 The height of the waypoint is fixed; if it is inside, adjust the A i+1 Waypoint (X i+1 +△X,Y i+1 +△Y) until the horizontal and vertical curvatures in step (2) are met, such as Figure 7 As shown, then perform steps (4) and (5).
[0063] (6) Obtain the UAV path waypoint position and height attribute value in the two-dimensional map window
[0064] Traversing all waypoints, get the three-dimensional coordinates of the UAV waypoint as A i (X i +△X i ,Y i +△Y i ,H i +△H i ), then the new coordinate a of the waypoint in the two-dimensional map window is obtained through coordinate system conversion and projection transformation i (x i ,y i ) and the height attribute value h i.
[0065] In the following, a certain area in Suzhou City is selected as a sample area to further describe this embodiment.
[0066] 1. Overview of the test area
[0067] In order to verify the feasibility of this UAV dynamic path planning method based on 2D and 3D integration and the feasibility of real-time early warning and obstacle avoidance, firstly realize the 2D and 3D integrated system platform based on B/S architecture, and then use single camera and multi-rotor The UAV collects oblique image data in a certain test area and produces a three-dimensional model of the full-element real scene; then realizes real-time early warning and obstacle avoidance of the two-dimensional and three-dimensional paths. The overall framework of the system is as follows Figure 8 As shown, the system interface is as follows Figure 9 shown.
[0068] 2. Aerial photography parameters and data preparation
[0069] The aerial photography test area is about 10km 2, using a self-developed multi-rotor UAV equipped with a Sony DSC-RX1RM2 camera (focal length 35mm, pixel 7952 5304) for route planning, flying once in the direction and in the opposite direction to achieve the effect of five cameras, and the course and side overlaps are 75% respectively , 80%, using Context Capture 4.4.11 to process the production of full-element reality models.
[0070] 3. Dynamic planning effect of UAV path planning
[0071] Use the above data and early warning and obstacle avoidance methods to plan the UAV path, and provide early warning and obstacle avoidance prompts for the real-time UAV three-dimensional path, such as Figure 10 shown.
