Unmanned aerial vehicle equipment for building outer wall damage inspection and application method of unmanned aerial vehicle equipment

Abstract

The invention discloses unmanned aerial vehicle equipment for building outer wall damage inspection and an application method thereof, and belongs to the technical field of unmanned aerial vehicle application, and the unmanned aerial vehicle equipment comprises a power system, a control system and a sensor system. The power system comprises a motor, blades and a battery, the control system comprises an airborne computer, an electronic speed controller and a flight controller, and the sensor system mainly comprises a depth camera. According to the method, a plane fitting algorithm and a multi-level traveling salesman problem algorithm are applied in the damage inspection process, in addition, different from a technical method for performing damage detection on the building by using an infrared thermal imager, the detection range and efficiency can be greatly improved, the detection precision is improved, and the detection cost is reduced. In addition, potential dangers caused by traditional manual high-altitude operation and part of detection means can be avoided, and a reference method is provided for the practicability of building outer wall damage detection.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicles, and in particular to unmanned aerial vehicle equipment for detecting damage to exterior walls of urban buildings and an application method thereof. Background Art

[0002] Although there are various technologies for detecting damage to building exterior walls, each has its own shortcomings. Infrared thermal imaging is easily affected by ambient temperature, wind speed, etc., and is insensitive to minor damage. Laser scanning is difficult to process data in complex building structures and the equipment is expensive. Computer vision relies on a large number of image samples for training, and cannot accurately identify special lighting and occlusion conditions. Drone remote sensing technology is limited by endurance and load, and image accuracy is limited. Knock detection technology is highly subjective, inefficient, and difficult to locate deep damage. Spectral analysis has high requirements for operators and is not suitable for noisy environments.

[0003] Traditional ground surveys are often limited by low mobility and flexibility, limiting their effectiveness in complex terrain and wide areas. In addition, the monitoring range and accuracy of these systems may not be sufficient for areas that require fine manipulation and wide coverage, requiring operation within a limited field of view, where traditional equipment has difficulty in achieving accurate and efficient data collection.

[0004] Therefore, there is a need to develop a new type of UAV equipment that can overcome the mobility and accuracy limitations associated with ground-based monitoring. Summary of the invention

[0005] The present invention overcomes the shortcomings of the prior art and provides a drone device for inspecting damage to building exterior walls and an application method thereof, which can greatly increase the detection range and efficiency, improve the detection accuracy, and avoid the potential dangers of traditional manual high-altitude operations and some detection methods, providing a reference method for the practicality of building exterior wall damage detection.

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

[0007] A drone device used for detecting damage to building exterior walls comprises a power system, a control system and a sensor system.

[0008] The power system comprises a motor, a blade and a battery; the motor is arranged at the four ends of the wing plate, the blade is connected to the motor, and the battery is arranged on the battery plate.

[0009] The control system includes an onboard computer, an electric regulator and a flight controller. The onboard computer is placed above the wing plate, the electric regulator is fixed above the fixed damping plate, and the flight controller is arranged below the fixed damping plate.

[0010] The sensor system is mainly composed of a depth camera. The depth camera is bonded to the top of the onboard computer.

[0011] The depth camera consists of a pair of stereo infrared sensors, an infrared laser transmitter and an RGB color camera. The depth camera is based on the principle of stereo vision. With the help of the infrared laser transmitter to project a pattern, the left and right infrared cameras capture the image and then calculate the parallax through an algorithm to obtain the depth information and merge it with the color image.

[0012] A method for using a drone for inspecting the exterior wall of a building uses the above-mentioned device and includes the following steps:

[0013] (1) Point cloud data acquisition and preprocessing: Before conducting building damage inspections, it is necessary to use advanced 3D laser scanning technology to pre-acquire high-resolution point cloud data of the building. After completing data acquisition, the plane fitting algorithm is used to determine the parameters of the acquired point cloud data. For buildings that conform to the standard geometric model under ideal conditions, its four main planes can be accurately fitted. Subsequently, the centroids, i.e., center points, of each of the four planes are determined through precise geometric calculations. These center points will serve as key nodes for subsequent path planning;

[0014] (2) Path planning between walls: Based on the known initial take-off position of the drone and the precisely measured coordinate information of the four wall center points, a weighted undirected graph is constructed between these key nodes. The nearest neighbor algorithm in TSP is used to comprehensively consider multi-dimensional factors such as the Euclidean distance between nodes and the connectivity of the path. The optimal flight path for the drone to perform damage inspection tasks between the walls of the building is planned to ensure that the flight trajectory of the drone between different walls is the shortest and most efficient, thereby maximizing the savings in inspection time and energy consumption.

[0015] (3) Path planning inside a single wall: First, the wall is divided into several regions of equal area using a uniform grid division strategy, and the center point coordinates of each region are determined by calculating the centroid of each region. The TSP algorithm is then used to solve the path planning of these center points, taking into account constraints such as the flight attitude of the drone, the field of view of the sensor, and the scanning resolution requirements, so as to obtain a refined scanning path for the drone on a single wall, ensuring that the drone can perform a full-scale scanning operation on the wall and accurately capture any signs of damage that may exist on the wall.

[0016] In step (1), the plane fitting algorithm gradually finds the optimal model parameters through multiple iterations of random sampling and model fitting. Each iteration calculates a model based on the current sampled subset and evaluates its fit with the entire data set, continuously optimizing the model. For models with a high proportion of outliers, the plane fitting algorithm can also find the optimal model parameters through multiple iterations, thereby obtaining accurate and reliable model estimates.

[0017] To ensure the accuracy of the plane model, the plane fitting algorithm needs to be iterated a certain number of times. The iteration formula is as follows:

[0018]

[0019] Where: N is the number of iterations, p is the expected probability of obtaining the correct model (usually 0.99 or higher), ω is the proportion of internal points in the data set, and k is the number of samples.

[0020] In steps (2) and (3), the nearest neighbor algorithm in the TSP constructs a path in sequence based on a greedy strategy, starting from the start node, selecting the nearest unvisited node at each step until a loop is formed, and finally obtaining a result that is relatively close to the optimal solution.

[0021] In the process of finding the shortest path, point P i and point P j The distance between them is calculated as follows:

[0022]

[0023] Where: d(P i ,P j ) represents the Euclidean distance between points, P i The i-th point in the table point set, (x i ,y i ,z i ) is the coordinate of the point in three-dimensional space, x i ,y i 、z i Respectively represent point P i The coordinate values ​​on the x-axis, y-axis, and z-axis.

[0024] The mathematical model for finding the nearest point is as follows:

[0025]

[0026] Among them, P near Indicates the found target point P t The nearest point. P-{P t} Except for the target point P t Point sets outside .

[0027] Compared with the prior art, the present invention has the following advantages and effects:

[0028] The drone equipment of the present invention can quickly approach and cover various areas of the building's outer wall, overcoming the problems of poor mobility and flexibility of traditional ground detection methods, and can obtain a large amount of data in a short time, greatly improving the detection range and efficiency.

[0029] During the point cloud data collection process of the drone of the present invention, the laser radar can quickly scan the exterior walls of buildings to obtain comprehensive spatial information, which significantly improves efficiency compared to traditional manual detection methods.

[0030] The present invention uses a plane fitting algorithm to extract building exterior wall features, can accurately identify the building exterior wall contour feature point set, effectively eliminate noise and abnormal point interference, thereby improving detection accuracy.

[0031] The drone application method of the present invention avoids the potential dangers of traditional manual high-altitude operations. The operator can complete the inspection task by remotely controlling the drone on the ground, thereby reducing the risk of casualties. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 is a front view of the drone device of the present invention;

[0033] Figure 2 It is a left view of the drone device of the present invention;

[0034] Figure 3 is a side view of the drone device of the present invention;

[0035] Figure 4 is a front view of the depth camera of the present invention;

[0036] Figure 5 is a side view of the depth camera of the present invention;

[0037] Figure 6 It is the technical roadmap of the drone of the present invention;

[0038] Figure 7 It is a flow chart of area division of the present invention. DETAILED DESCRIPTION

[0039] In order to facilitate the understanding of the present invention, the present invention will be described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be pointed out that for those skilled in the art, the present invention can also be modified and improved without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

[0040] Embodiment 1:

[0041] like Figure 1 , Figure 2 , Figure 3As shown, this embodiment provides a drone device for inspecting damage to the exterior wall of a building, including a wing panel 3, an onboard computer 2 is provided above the wing panel 3, and a depth camera 1 is bonded to the top of the onboard computer 2. A motor 4 is provided at the end of the wing panel 3, and the bottom of the motor 4 is connected to the blade 6. The electric adjustment board 11 is connected to the wing panel 3 through an aluminum column, and the electric adjustment 5 is bonded to the top of the electric adjustment board 11. The flight controller 7 is bonded to the bottom of the electric adjustment board 4. The battery 9 is fixed to the battery panel 10 through a strap, and the battery panel is connected to the fixed shock-absorbing plate 8 through an aluminum column.

[0042] Specifically, the onboard computer is connected to the flight controller via a USB data transmission cable, the flight controller is connected to the ESC via a signal cable, the ESC is connected to the 5V interface of the flight controller via a power cable, and the motor is connected to the motor via a power cable. The model of the onboard computer is INTEL NUC12-i5, the model of the motor is SmoothX-kv1750, the model of the ESC is XRotor Micro 65A4in1, and the model of the flight controller is Pixhawk6C mini.

[0043] Embodiment 2:

[0044] Specifically, taking a rectangular building as the inspection object, the working steps of the drone application device of this embodiment are as follows:

[0045] 1) Point cloud data acquisition and preprocessing: Before conducting building damage inspections, it is necessary to use advanced 3D laser scanning technology to pre-acquire high-resolution point cloud data of the building. After collecting a certain amount of point cloud data, it is pre-processed using statistical filtering methods. By analyzing the spatial distribution characteristics of each point in the point cloud data, noise points and abnormal points that are obviously deviated from the normal data range are identified and removed, and the judgment thresholds for noise points and abnormal points are set. According to the distribution of point cloud data, points that are more than 2 times the standard deviation of the average distance from adjacent points are judged as noise points or abnormal points and removed.

[0046] According to the actual situation of the building's exterior wall and the requirements for detection accuracy, the number of iterations of the plane fitting algorithm is set to 1000 times, the distance threshold is 0.05 meters, and the minimum number of sample points is 3. Randomly select 3 non-collinear points from the collected point cloud data, use these 3 points to establish the fitting plane equation, and solve the parameters in the equation. Calculate the distance from the remaining points to the fitting plane, determine the points with a distance less than the set threshold as inliers, and count the number of inliers. Repeat the above process of randomly selecting points, establishing plane equations, calculating distances, and counting inliers for 1000 iterations. Finally, select the model with the most inliers and the corresponding point set, which is the final feature point set of the building's exterior wall contour. These feature points can accurately describe the shape and contour of the building's exterior wall.

[0047] Subsequently, the centroids, or center points, of each of the four planes are determined through precise geometric calculations. These center points will serve as key nodes for subsequent path planning.

[0048] To ensure the accuracy of the plane model, the plane fitting algorithm needs to be iterated a certain number of times. The iteration formula is as follows:

[0049]

[0050] Where: N is the number of iterations, p is the expected probability of getting the correct model (usually 0.99 or higher), ω is the proportion of internal points in the data set, and k is the number of samples;

[0051] 2) Path planning between walls: Based on the known initial take-off position of the drone and the precisely measured coordinate information of the four wall center points, a weighted undirected graph is constructed between these key nodes. The nearest neighbor algorithm in TSP is used to comprehensively consider multi-dimensional factors such as the Euclidean distance between nodes and the connectivity of the path. The optimal flight path for the drone to perform damage inspection tasks between the walls of the building is planned to ensure that the flight trajectory of the drone between different walls is the shortest and most efficient, thereby maximizing the savings in inspection time and energy consumption. The specific process is as follows:

[0052] The starting point is the take-off position of the drone. c =P s , start by searching for the distance P among the points that have not been visited c The nearest point P n , P n Add to the path and update the current point P c =P n Repeat the above steps of searching for the nearest point and adding it to the path until all target points are visited. Finally, add the starting point to the end of the path to form a closed loop path, which is the inspection path between the walls.

[0053] In the process of finding the shortest path, point P i and point P j The distance between them is calculated as follows:

[0054]

[0055] Where: d(P i ,P j ) represents the Euclidean distance between points, P i The i-th point in the table point set, (x i ,y i ,z i ) is the coordinate of the point in three-dimensional space, x i ,y i 、zi Respectively represent point P i The coordinate values ​​on the x-axis, y-axis, and z-axis.

[0056] The mathematical model for finding the nearest point is as follows:

[0057]

[0058] Among them, P near Indicates the found target point P t The nearest point. P-{P t}Except the target point P t Point sets other than ;

[0059] 3) Path planning inside a single wall: First, a uniform grid division strategy is used to divide the wall into several areas of equal area, and the center point coordinates are determined by calculating the centroid of each area. The TSP algorithm is then used to solve the path planning of these center points, taking into account constraints such as the flight attitude of the drone, the field of view of the sensor, and the scanning resolution requirements, so as to obtain a refined scanning path for the drone on a single wall, ensuring that the drone can perform a full range of scanning operations on the wall and accurately capture possible signs of damage to the wall. The specific process is as follows:

[0060] The starting point is the center of the wall. From the current point P c =P s , start by searching for the distance P among the points that have not been visited c The nearest point P n , P n Add to the path and update the current point P c =P n Repeat the above steps of searching for the nearest point and adding it to the path until all target points are visited. Finally, add the starting point to the end of the path to form a closed loop path, which is the inspection path of a single wall.

[0061] It can be understood that the above specific description of the present invention is only used to illustrate the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that the present invention can still be partially modified or replaced by equivalents to achieve the same technical effects: as long as the use requirements are met, they are within the protection scope of the present invention.

Claims

1. An unmanned aerial vehicle device for detecting damage to building exterior walls, characterized in that: The UAV equipment includes a power system, a control system and a sensor system; The power system comprises a motor (4), a propeller blade (6) and a battery (9); the motor (4) is arranged at the four ends of the wing panel (3), the propeller blade (6) is connected to the motor (4), and the battery (9) is used to supply power to the motor; the power system receives a flight control instruction sent by the control system; The control system comprises an onboard computer (2), an electric regulator (5) and a flight controller (9); the onboard computer (2) is arranged above the wing plate (3), the electric regulator (5) is fixed above the fixed shock absorbing plate (11), and the flight controller (7) is arranged below the fixed shock absorbing plate (11).

2. The drone device according to claim 1, characterized in that: The sensor system is composed of a depth camera (1), which is bonded to the top of an onboard computer (2), and the depth camera (1) includes a stereo infrared sensor, an infrared laser transmitter, and an RGB color camera; The depth camera is based on the principle of stereoscopic vision. The stereo infrared sensor captures images and then calculates parallax to obtain depth information, which is then fused with the color image.

3. A method for using a drone to inspect building exterior walls, characterized in that: The method uses the drone equipment according to claim 1 or 2, and comprises the following steps: (1) Point cloud data acquisition and preprocessing: Based on 3D laser scanning technology, high-resolution point cloud data of the building is acquired in advance, and the plane fitting RANSAC algorithm is used to determine the parameters of the acquired point cloud data to fit the main planes of the building's exterior wall; then the centroid of each plane, i.e., the center point, is determined through precise geometric calculation, and the center point will be used as the key node for subsequent path planning; (2) Path planning between walls: Based on the known initial take-off position of the drone and the coordinate information of the key nodes, a weighted undirected graph is constructed between the key nodes. The nearest neighbor algorithm in the traveling salesman problem model TSP is used to plan the optimal flight path for the drone to perform damage inspection tasks between the walls of the building's exterior walls; (3) Path planning inside a single wall: First, a uniform grid division strategy is used to divide each of the building's exterior walls into a number of regions of equal area, and the center point coordinates of each region are determined by calculating the centroid of each region; Then, the TSP algorithm is used to solve the path planning of the center point coordinates to obtain a refined scanning path of the UAV on a single wall.

4. The method for using a drone for building exterior wall inspection according to claim 3 is characterized in that: In step (1), the plane fitting RANSAC algorithm, through the iterative process of multiple random sampling and plane equation model fitting, calculates a model based on the current sampled subset in each iteration, and evaluates its degree of fit with the entire data set, thereby continuously optimizing the model.

5. The method for using a drone for building exterior wall inspection according to claim 4, characterized in that: The plane fitting RANSAC algorithm is iterated a set number of times, and the iteration formula is as follows: Where: N is the number of iterations, p is the expected model probability, ω is the proportion of internal points in the data set, and k is the number of samples.

6. The method for using a drone for building exterior wall inspection according to claim 3, characterized in that: In step (2), the optimal flight path for performing damage inspection tasks between the walls of the building's exterior walls is obtained by: Point P i and point P j The distance between them is calculated as follows: Where: d(P i ,P j ) represents the Euclidean distance between points, P i represents the i-th point in the key node point set of path planning, (x i ,y i ,z i ) is the coordinate of the point in three-dimensional space, x i ,y i 、z i Respectively represent point P i Coordinate values ​​on the x-axis, y-axis, and z-axis; The mathematical model for finding the nearest point is as follows: Among them, P near Indicates the found target point P t The nearest point, P-{P t }Except the target point P t Points outside .

7. The method for using a drone for building exterior wall inspection according to claim 3, characterized in that: In steps (2) and (3), the nearest neighbor algorithm is based on a greedy strategy and constructs a path in sequence, starting from the starting node, selecting the nearest unvisited node at each step until a loop is formed, and finally obtaining the optimal solution.

Images