A method for controlling unmanned aerial vehicles (UAVs) used for the removal of power transmission towers.

Abstract

This invention relates to the technical field, specifically disclosing a drone control method for dismantling power transmission towers, comprising the following steps: dividing the network into sub-grids; acquiring the target shape based on the sub-grids; controlling the drone to acquire target images; if the area of ​​the minimum bounding rectangle does not belong to an area range, acquiring a comparison shape and a standard shape; when the similarity between the two is greater than a preset value, adjusting the drone's flight altitude to a second altitude; when the similarity between the two is less than or equal to a preset value, replacing the corresponding comparison shape with the standard shape, and acquiring a new minimum bounding rectangle area S1; if S1 does not belong to an area range, adjusting the drone's flight altitude to a second altitude; if S1 belongs to an area range, controlling the drone to survey the next grid area; and constructing a three-dimensional curve of the cable in the real scene based on the survey information for use in dismantling power transmission lines. This invention can reduce the energy consumption of drones.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) control technology, and more specifically to a UAV control method for dismantling power transmission towers. Background Art

[0002] With continuous breakthroughs in drone technology and a continuous decrease in costs, its application in various fields such as surveying, inspection, and logistics is accelerating. By using drones to recover cables, high-altitude operations can be automated and intelligentized, which can significantly improve operational efficiency and reduce the risks of manual operations.

[0003] Before officially dismantling the cable, the drone first collects high-resolution and depth images of the target area. After the collection is completed, the images are classified pixel by pixel using a pre-trained convolutional neural network (such as a semantic segmentation model based on U-Net or DeepLab) to accurately label the spatial location and direction of the cable. The depth image is then fused with the segmentation results to reconstruct the three-dimensional curve of the cable in the real scene.

[0004] A Chinese patent application with publication number CN119295678B discloses a method for constructing a three-dimensional spatial model for power transmission line dismantling. The method mainly includes correcting the three-dimensional model using a second aerial photograph to obtain a three-dimensional real-world model, and then obtaining the final three-dimensional model based on the real-world model and power line modeling. However, when determining whether to acquire the second aerial photograph, errors can occur because some tower sections may be obscured by vegetation, leading to errors in the determination. This results in the drone being controlled to a second altitude to capture the second aerial photograph, increasing the drone's energy consumption. Summary of the Invention

[0005] The purpose of this invention is to provide a drone control method for dismantling power transmission towers, thereby solving the aforementioned technical problems.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A method for controlling unmanned aerial vehicles (UAVs) used for dismantling power transmission towers includes the following steps:

[0008] Set a grid area and divide it into subgrids; move the grid area so that the projection of the center point of the power tower on the ground is at the center point of any subgrid; control the flight altitude of the UAV to the first altitude; acquire images within the grid area; and obtain the target shape corresponding to the subgrid based on the images.

[0009] During the surveying process, the drone is controlled to move along the planned route, and the aerial images collected at the center point of the wide side of the grid area are recorded as the target images;

[0010] If the minimum bounding rectangle area covering all power tower areas in the target image is... Then, the smallest bounding rectangle of a single power tower in the target image is taken and recorded as the comparison shape; and the projection point of the center point of the power tower in the target image on the ground is obtained and recorded as the standard point. The target shape corresponding to the sub-grid where the standard point is located is recorded as the standard shape.

[0011] When the similarity between the comparison shape and the corresponding standard shape is greater than a preset value, the drone's flight altitude is adjusted to a second altitude to acquire a second aerial image;

[0012] When the similarity between the standard shape and the comparison shape is less than or equal to a preset value, the standard shape replaces the corresponding comparison shape, and a new minimum bounding rectangle area S1 is obtained.

[0013] like Then adjust the drone's flight altitude to the second altitude to obtain the second aerial image;

[0014] If S1∈[f, g], then control the UAV to survey the next grid area along the current surveying route;

[0015] After the survey is completed, a three-dimensional curve of the cable in the real scene is constructed based on the survey information, which is used for the dismantling of the transmission line.

[0016] As a further aspect of the present invention: obtaining the target shape includes:

[0017] Set a grid area and divide it into several sub-grids, with the length and width of each sub-grid being a preset first length. Determine the projection point A of the center point of the power tower on the ground. Move the grid area so that the projection point A is located at the center point of sub-grid i. Keep the UAV's flight altitude at a first altitude. Collect an image of the grid area at the center point of the wide side of the grid area. Mark the minimum bounding rectangle of the power tower in the image and denote it as the target shape.

[0018] As a further aspect of the present invention, obtaining the target shape also includes:

[0019] Starting from the beginning of the preset second altitude range, select several altitudes at preset altitude intervals and record them as target altitudes to obtain the target shape when the drone's flight altitude is any target altitude.

[0020] As a further aspect of the present invention: obtaining the similarity includes:

[0021] Obtain the area difference C1 between the comparison shape and the corresponding standard shape, and obtain the area C2 of the overlapping part of the comparison shape and the corresponding standard shape, and calculate the similarity P=C2 / C1.

[0022] As a further aspect of the present invention: if the area difference C1 is less than a preset area difference threshold, then the similarity degree P is equal to the value of C2.

[0023] As a further aspect of the present invention: after acquiring the second aerial image, the drone's flight altitude is maintained at the second altitude, and after acquiring a new aerial image, the drone's flight altitude is determined based on the target shape at the same target altitude as the second altitude.

[0024] As a further aspect of the present invention: during the surveying process, after the target area is divided into grids, the aspect ratio of the grid area is 2, and the width is a preset second length.

[0025] The beneficial effects of this invention compared to the prior art are as follows:

[0026] 1) This invention establishes a baseline morphology database of power towers at different heights through sub-mesh partitioning and standard shape pre-acquisition mechanisms. When determining whether to trigger secondary mapping, a similarity analysis (area difference and overlap) between the compared shape and the standard shape is introduced. When the tower body is obscured by vegetation, causing an anomaly in the measured area, if the similarity is lower than a preset value, the system can identify it as "occlusion interference" and recalculate the area using the unobstructed standard shape, avoiding the misjudgment of "insufficient resolution" due to local occlusion and the initiation of secondary mapping. This mechanism effectively filters out false anomalies caused by occlusion, reduces additional flight missions for UAVs due to misjudgment, and improves the robustness of the judgment logic.

[0027] 2) This invention employs a pre-survey calibration mechanism to make minor translational adjustments to the grid area, ensuring that the projection points of the same power tower fall sequentially at the centers of different sub-grids. It also extracts the minimum bounding rectangle at each location, constructing multiple sets of "target shape" sample libraries. During actual measurement, the system matches the real-time acquired comparison shapes with the sample libraries, effectively compensating for contour changes caused by factors such as actual tower position offset and shooting angle distortion. Even if the tower body is partially obscured by vegetation or affected by light and shadow, resulting in incomplete contours, the system can accurately assess whether the current imaging quality meets the requirements through feature matching from the sample library. This allows for correct judgment on whether the flight altitude needs to be adjusted for reshooting, significantly reducing the probability of misjudgments due to obstruction or viewing angle issues, avoiding repeated flights of the drone due to invalid commands, and reducing unnecessary energy consumption.

[0028] 3) This invention uses the linear correction formula from the background technology to adjust the UAV's flight altitude, setting the switched second altitude as the new reference altitude. During subsequent mapping, altitude adjustment is only triggered when significant deviations occur in multiple consecutive grid areas. This mechanism eliminates the traditional frequent adjustment mode of "independent ascent and descent per grid," avoiding the push-brake cycle caused by frequent altitude switching and reducing wear and tear on the battery and motors. By maintaining a relatively stable flight state, the UAV can cover most of the mapping area with a more efficient path, effectively extending flight time while ensuring modeling accuracy, and improving the overall smoothness and economy of the operation. Attached Figure Description

[0029] The invention will now be further described with reference to the accompanying drawings.

[0030] Figure 1 This is a flowchart illustrating a drone control method for dismantling power transmission towers according to the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figure 1 As shown, this invention provides a drone control method for dismantling power transmission towers, comprising the following steps:

[0033] Set a grid area and divide it into subgrids; move the grid area so that the projection of the center point of the power tower on the ground is at the center point of any subgrid; control the flight altitude of the UAV to the first altitude; acquire images within the grid area; and obtain the target shape corresponding to the subgrid based on the images.

[0034] In a preferred embodiment of the present invention, obtaining the target shape includes:

[0035] Set a grid area and divide the grid area into several sub-grids. The length and width of the sub-grids are both preset first lengths. Determine the projection point A of the center point of the power tower on the ground. Move the grid area so that the projection point A is at the center point of the sub-grid i. Keep the flight altitude of the UAV at the first altitude. Collect the image of the grid area at the center point of the wide side of the grid area. Mark the minimum bounding rectangle of the power tower in the image and record it as the target shape.

[0036] It should be noted that, firstly, a grid area is set and divided into several sub-grids. For example, the length and width of each sub-grid are both set to 5 meters (the preset first length), forming a regular grid matrix. Next, the projection point A of the center point of the power tower on the ground is determined. The grid area is translated manually or by program control so that the projection point A is precisely aligned with the center point of sub-grid i. If sub-grid i is the first sub-grid in the upper left corner, the grid position is adjusted so that point A coincides with the cross center point of that sub-grid. Then, the drone is controlled to fly at a first altitude (e.g., 50 meters, the flight altitude at the lowest target accuracy) to the center point of the wide side of the grid area, fly stably along the planned route, and collect aerial images of the grid area. In the acquired images, the outline of the power tower is automatically detected using an image recognition algorithm, and the smallest bounding rectangle that wraps around the main body of the tower is marked. For example, for a tower with a rectangular cross-section, the algorithm will identify its four edge vertices and generate a rectangle that fits the outline. This rectangle is the target shape of the corresponding sub-grid i.

[0037] By aligning the tower projection point with the center point of the subgrid, the tower is ensured to be in a fixed relative position in each acquired image, facilitating the establishment of a unified shape benchmark. Regular subgrid division (e.g., 5-meter side length) subdivides the grid region into quantifiable units, establishing a one-to-one correspondence between the tower position and the subgrid, avoiding shape analysis errors caused by positional deviations. Acquiring images at a fixed height (e.g., 50 meters) ensures consistent resolution during initial mapping, facilitating feature comparison at different heights. The process of annotating the minimum bounding rectangle (e.g., identifying tower edge vertices) transforms the tower's visual features into calculable geometric parameters, providing foundational data for subsequent similarity assessment and area calculation. This series of operations collectively constructs a standardized "position-shape" mapping system, enabling the system to establish a reliable target shape sample library based on precise alignment and feature extraction at the subgrid level. In actual measurements, by comparing the shape's matching degree with the sample library, interference factors such as occlusion and viewpoint distortion can be effectively identified, providing a basis for accurately determining whether secondary mapping is necessary, ultimately improving the energy efficiency of UAV operations.

[0038] In a preferred embodiment, obtaining the target shape further includes:

[0039] Starting from the beginning of the preset second altitude range, select several altitudes at preset altitude intervals and record them as target altitudes to obtain the target shape when the drone's flight altitude is any target altitude;

[0040] It is worth noting that after acquiring the second aerial image, the drone's flight altitude is maintained at the second altitude, and after acquiring a new aerial image, the drone's flight altitude is adjusted based on the target shape at the same target altitude as the second altitude.

[0041] In the specific implementation process, starting from the preset second height range (e.g., 30 meters), several heights are selected sequentially as target heights at preset height intervals (e.g., 5 meters) (e.g., 30 meters, 35 meters, 40 meters, etc.). At each target height, the process of "moving the grid area to align the tower projection point with the center of the sub-grid → acquiring the image → marking the minimum bounding rectangle" is repeated to obtain the target shape at the corresponding height. This approach aims to establish multiple sets of "height-shape" feature pairs at different flight altitudes, forming a target shape sample library covering multiple resolutions. When adjustments to a second altitude are needed during actual measurements (e.g., altitude correction triggered by area anomalies), the system can directly call the pre-collected target shape at that altitude as the new benchmark, eliminating the need for temporary re-collection and avoiding secondary judgment delays caused by a lack of benchmark data after altitude switching. Simultaneously, with the new altitude set as the benchmark, subsequent continuous grid mapping can be directly compared to the target shape at that altitude, only requiring further adjustments when significant deviations occur continuously, reducing ineffective "trial ascent and descent" operations. By pre-constructing a multi-altitude feature library, an instantly available comparison benchmark is provided for dynamic altitude adjustments, enabling the UAV to quickly enter a stable mapping state after switching to a second altitude, avoiding frequent starts and stops and altitude oscillations, thus improving the continuity and energy efficiency of UAV operations.

[0042] During the surveying process, the drone is controlled to move along the planned route, and the aerial images collected at the center point of the wide side of the grid area are recorded as the target images;

[0043] It is worth noting that during the surveying process, after the target area is divided into grids, the aspect ratio of the grid area is 2, and the width is the preset second length;

[0044] During the surveying process, the UAV moves at a constant speed along a pre-planned route (such as using the line connecting the center points of the two wide sides of the grid area as the flight path). When it flies to the center point of the wide side of the grid area (such as the coordinates of the midpoint of the left wide side being (X1,Y1)), it triggers an image acquisition command, takes an aerial image of that location, and records it as the target image. For example, in hilly terrain, the UAV takes a panoramic image of a grid area containing multiple power towers in a stable attitude.

[0045] If the minimum bounding rectangle area covering all power tower areas in the target image is... Then, the smallest bounding rectangle of a single power tower in the target image is taken and recorded as the comparison shape; and the projection point of the center point of the power tower in the target image on the ground is obtained and recorded as the standard point. The target shape corresponding to the sub-grid where the standard point is located is recorded as the standard shape.

[0046] If the minimum bounding rectangle area S covering all power tower areas in the target image, calculated by the image analysis algorithm, exceeds the preset interval [f, g] (e.g., S is less than f or greater than g) (the interval [f, g] is the preset area interval), then the image is further subdivided: the outline of each independent power tower is identified, and the minimum bounding rectangle of a single tower is extracted (e.g., for the first tower from the left in the image, its independent rectangle is generated), which is recorded as the comparison shape; at the same time, through geolocation data and image coordinate system transformation, the actual projection point coordinates of the tower center point on the ground (e.g., latitude and longitude coordinates (Lon, Lat)) are determined, and it is defined as the standard point. The coordinates of the standard point are matched to the corresponding subgrid (e.g., the coordinates fall within the range of subgrid i), and the target shape marked by the subgrid in the pre-collection stage (i.e., the minimum bounding rectangle of the tower corresponding to the subgrid at the first height) is retrieved as the standard shape;

[0047] When the similarity between the comparison shape and the corresponding standard shape is greater than a preset value, the drone's flight altitude is adjusted to a second altitude to acquire a second aerial image;

[0048] In another preferred embodiment of the present invention, obtaining the degree of similarity includes:

[0049] Obtain the area difference C1 between the comparison shape and the corresponding standard shape, and obtain the area C2 of the overlapping part of the comparison shape and the corresponding standard shape, and calculate the similarity P=C2 / C1;

[0050] It should be noted that if the area difference C1 is less than the preset area difference threshold, then the similarity P is equal to the value of C2.

[0051] By calculating the area difference C1 (i.e., |area of ​​the comparison shape - area of ​​the standard shape|) and the overlapping area C2 (i.e., the area of ​​the overlapping region), and using P=C2 / C1 as a quantitative indicator of similarity (if C1=0, then P is directly taken as C2, in which case the areas are the same, and the degree of overlap is the similarity), the shape difference can be transformed into a calculable numerical relationship. The advantage of this approach is that the area difference C1 reflects the absolute magnitude of the difference between the two, while the overlapping area C2 reflects the degree of contour matching, and the ratio P of the two can comprehensively evaluate the similarity between the comparison shape and the standard shape. For example, if a comparative shape is reduced in area due to occlusion but the main outline has a high degree of overlap with the standard shape (C2 larger, C1 smaller, P value higher), it may be a problem with the actual size or resolution, requiring adjustment of the altitude and reshooting. If the area difference is large and the overlap is low (C2 smaller, C1 larger, P value lower), it is likely due to occlusion interference, which can be compensated for by the standard shape. This quantitative judgment mechanism provides the system with an objective basis for decision-making, avoiding the ambiguity of subjective threshold judgment, enabling the drone to accurately identify "real anomalies" and "occlusion misjudgments" based on data, so that secondary mapping is only initiated when necessary, reducing unnecessary altitude adjustments and flight energy consumption.

[0052] When the similarity between the standard shape and the comparison shape is less than or equal to a preset value, the standard shape replaces the corresponding comparison shape, and a new minimum bounding rectangle area S1 is obtained.

[0053] like Then adjust the drone's flight altitude to the second altitude to obtain the second aerial image;

[0054] If S1∈[f, g], then control the UAV to survey the next grid area along the current surveying route;

[0055] Understandably, when the similarity P (calculated by P=C2 / C1, e.g., P=0.8) between the comparison shape (such as the smallest circumscribed rectangle of a power tower due to partial obstruction) and the corresponding standard shape (the rectangle in the pre-collected unobstructed state) is greater than the preset value (e.g., 0.7), it indicates that the difference in their outlines is small, and the area anomaly is more likely caused by insufficient resolution at the current flight altitude. At this time, the system triggers a command to adjust the drone's flight altitude to a second altitude (e.g., from 50 meters to 30 meters, the specific value adopts the calculation method of the document cited in the background technology), and flies along the original route at the new altitude to collect a second aerial image. For example, in sparsely vegetated areas, a clearer image of the entire tower can be obtained after the altitude is reduced. When the similarity P is less than or equal to the preset value (e.g., P=0.4), it indicates that the comparison shape differs significantly from the standard shape due to factors such as occlusion. The system automatically replaces the measured comparison shape with the standard shape (e.g., the pre-collected complete rectangle) and recalculates the area S1 of the new minimum bounding rectangle covering all power tower areas. For example, after correcting the area of ​​the occluded tower to the area of ​​the standard shape, it is recombined with the measured shapes of other towers to calculate the global area. If the corrected S1 still exceeds the preset interval [f,g] (e.g., S1 is greater than g), it is determined that the actual area is abnormal, and the height is adjusted to the second height for reshooting; if S1 falls within the interval (e.g., f≤S1≤g), it is considered that the image at the current height meets the accuracy requirements, and the drone is controlled to continue mapping the next grid.

[0056] The mechanism distinguishes the root cause of area anomalies by using similarity analysis: for cases with high similarity (P > preset value), the altitude is directly adjusted to match the resolution requirements; for cases with low similarity (P ≤ preset value), occlusion interference is first compensated with a standard shape before the global area is evaluated, avoiding misjudgments of local occlusion as resolution issues. For example, when a tower is obscured by trees, resulting in a smaller measured area and low overlap with the standard shape, the calculation deviation caused by the occlusion can be eliminated by replacing it with the standard shape, avoiding unnecessary descents of the drone for reshoots. This layered processing mechanism ensures an effective response to insufficient true resolution while filtering out false anomalies caused by occlusion, allowing the drone to adjust its altitude only when necessary, reducing unnecessary flights and energy consumption.

[0057] After the survey is completed, a three-dimensional curve of the cable in the real scene is constructed based on the survey information, which is used for the dismantling of the transmission line.

[0058] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for controlling unmanned aerial vehicles (UAVs) used for dismantling power transmission towers, characterized in that, Includes the following steps: Set a grid area and divide it into subgrids; move the grid area so that the projection of the center point of the power tower on the ground is at the center point of any subgrid; control the flight altitude of the UAV to the first altitude; acquire images within the grid area; and obtain the target shape corresponding to the subgrid based on the images. During the surveying process, the drone is controlled to move along the planned route, and the aerial images collected at the center point of the wide side of the grid area are recorded as the target images; If the minimum bounding rectangle area covering all power tower areas in the target image is... Then, the smallest bounding rectangle of a single power tower in the target image is taken and recorded as the comparison shape; and the projection point of the center point of the power tower in the target image on the ground is obtained and recorded as the standard point. The target shape corresponding to the sub-grid where the standard point is located is recorded as the standard shape; [f, g] represents the preset area interval; When the similarity between the comparison shape and the corresponding standard shape is greater than a preset value, the drone's flight altitude is adjusted to a second altitude to acquire a second aerial image, where the second altitude is lower than the first altitude; When the similarity between the standard shape and the comparison shape is less than or equal to a preset value, the standard shape replaces the corresponding comparison shape, and a new minimum bounding rectangle area S1 is obtained. like Then adjust the drone's flight altitude to the second altitude to obtain the second aerial image; If S1∈[f, g], then control the UAV to survey the next grid area along the current surveying route; After the survey is completed, a three-dimensional curve of the cable in the real scene is constructed based on the survey information, which is used for the dismantling of the transmission line.

2. The UAV control method for dismantling power transmission towers according to claim 1, characterized in that, Obtaining the target shape includes: Set a grid area and divide it into several sub-grids. The length and width of each sub-grid are preset first lengths. Determine the projection point A of the center point of the power tower on the ground. Move the grid area so that the projection point A is at the center point of sub-grid i. Keep the UAV's flight altitude at the first altitude. Collect images of the grid area at the center point of the wide side of the grid area. Mark the minimum bounding rectangle of the power tower in the image and record it as the target shape.

3. The UAV control method for dismantling power transmission towers according to claim 2, characterized in that, Obtaining the target shape also includes: Starting from the beginning of the preset second altitude range, select several altitudes at preset altitude intervals and record them as target altitudes to obtain the target shape when the drone's flight altitude is any target altitude.

4. The UAV control method for dismantling power transmission towers according to claim 1, characterized in that, Obtaining the similarity includes: Obtain the area difference C1 between the comparison shape and the corresponding standard shape, and obtain the area C2 of the overlapping part of the comparison shape and the corresponding standard shape, and calculate the similarity P=C2 / C1.

5. The UAV control method for dismantling power transmission towers according to claim 4, characterized in that, If the area difference C1 is less than the preset area difference threshold, then the similarity P is equal to the value of C2.

6. The UAV control method for dismantling power transmission towers according to claim 3, characterized in that, After acquiring the second aerial image, maintain the drone's flight altitude at the second altitude. After acquiring a new aerial image, determine whether to adjust the drone's flight altitude based on the target shape at the same target altitude as the second altitude.

7. The UAV control method for dismantling power transmission towers according to claim 1, characterized in that, During the surveying process, after the target area is divided into grids, the aspect ratio of the grid area is 2, and the width is the preset second length.

Images