An unmanned intelligent bridge inspection system

Abstract

The application belongs to the technical field of bridge inspection management, in particular to a bridge inspection system based on unmanned intelligence, which comprises an image acquisition module, an analysis module and a navigation module.The image acquisition module comprises a UAV, which acquires images of the surface of a bridge.The analysis module comprises an identification unit and a grading unit.The identification unit identifies the acquired images of the surface of the bridge according to a disease identification model, and the grading unit grades the identified bridge diseases.The navigation module generates an inspection route according to a three-dimensional model of the bridge to be inspected.The UAV in the image acquisition module comprises a long-range inspection UAV and a short-range inspection UAV.The short-range inspection UAV is located between the long-range inspection UAV and the bridge.The short-range inspection UAV reacquires images of suspicious areas on the bridge.The application improves the quality of acquired images and the accuracy of identification through the mutual cooperation of UAVs at different distances, and improves the inspection effect and efficiency through multistage analysis.

Description

Technical Field

[0001] This invention belongs to the field of bridge inspection and management technology, specifically a bridge inspection system based on unmanned intelligence. Background Technology

[0002] Bridge inspection work is usually based on visual inspection, supplemented by measuring tools when necessary. Professional inspectors check the bridges and fill in inspection record forms so that bridge management and maintenance units can grasp the technical condition and the development of defects of each active bridge in a timely and accurate manner. Traditional bridge inspection is mainly based on visual inspection, which requires inspectors to fill in forms and collect photos of bridge defects on site. The quality of inspection data is affected by many factors such as personnel quality and environmental conditions, and the accuracy is difficult to guarantee. Thanks to innovative breakthroughs in artificial intelligence, image processing, and drone technology, drone inspection technology combined with image recognition has been gradually applied to bridge inspection. However, in complex environments, drone inspection combined with image recognition is easily affected by factors such as light and weather, resulting in insufficient data reliability, difficulty in fully covering key parts of the bridge, and significant limitations in the accuracy and stability of the inspection results. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes a bridge inspection system based on unmanned intelligence, which improves the quality of acquired images by cooperating with multiple drones and performing multi-level analysis on the images, thereby enhancing the accuracy, efficiency, and effectiveness of inspection results.

[0004] The technical solution adopted by this invention to solve its technical problem is: a bridge inspection system based on unmanned intelligence, comprising: An image acquisition module, comprising a drone equipped with a camera to acquire images of the bridge surface; The analysis module includes an identification unit and a grading unit. The identification unit identifies the collected bridge surface images according to the defect identification model. The identification unit divides the bridge surface images into three categories: intact, questionable, and defective according to the identification results. The grading unit grades the identified bridge defects. The navigation module generates an inspection route based on the 3D model of the bridge to be inspected, and the UAV inspects the bridge according to the inspection route. The drones in the image acquisition module include remote detection drones and near detection drones. The near detection drones are located between the remote detection drones and the bridge surface, and the number of near detection drones is greater than the number of remote detection drones. The near-inspection drone re-captures images of the areas on the bridge surface that are classified as questionable. The near-inspection drone is within the shooting range of the far-inspection drone. The identification unit detects the distance between the drone and the bridge surface, and between drones, in the acquired images.

[0005] Preferably, the identification unit includes: A model recognition component, wherein the model recognition component is a pre-trained disease recognition model, and the disease recognition model is pre-trained using collected bridge disease images; An anomaly monitoring component identifies the acquired bridge images, and based on the fact that the image occupies most of the normal area, it identifies and marks the small areas in the image that significantly deviate from the expected distribution as defects. The anomaly monitoring component then re-identifies images that the model recognition component has determined to be defect-free.

[0006] Preferably, the anomaly monitoring component includes: In the normal mode, the collected single image is analyzed and identified to find anomalies in the image, namely bridge defects. The re-inspection mode combines multiple collected bridge images to form a re-inspection image. The re-inspection mode analyzes and identifies the re-inspection image to find any abnormalities, i.e., bridge defects. The re-inspection mode also re-identifies images that were deemed intact in the conventional mode.

[0007] Preferably, the navigation module includes a segmentation unit and a planning unit. The segmentation unit segments the three-dimensional model of the bridge into multiple independent work areas, and the planning unit plans inspection routes for the remote inspection UAV and the near inspection UAV respectively. The remote inspection drone and the near inspection drone are scheduled in a time-sharing manner so that the inspection routes do not overlap, and the image acquisition module acquires images of each work area in sequence.

[0008] Preferably, there are overlapping areas between the working areas divided by the slicing unit, and the overlapping area between the working areas is between 2 / 5 and 3 / 5.

[0009] Preferably, the analysis module further includes an analysis unit, which analyzes the acquired bridge surface images to identify the parts with poor image quality, and provides improvement conditions for the shooting based on the poor quality images. The navigation module also includes an adjustment unit, which adjusts the inspection route of the UAV based on the improvement conditions given by the preprocessing unit. The adjustment unit adjusts the inspection route based on the inspection route given by the planning unit. After the UAV completes the inspection of its work area, it flies and collects images according to the inspection route adjusted by the adjustment unit.

[0010] Preferably, the near-inspection UAV is equipped with a flaw detection device, which includes at least one of laser detection equipment, ultrasonic detection equipment, and radar detection equipment.

[0011] Preferably, the near-inspection drone is equipped with a marking device, which is not installed on the same drone as the flaw detection equipment. The marking device marks the identified defects at the corresponding positions on the bridge surface. The marking device marks by launching paint eggs, which are filled with fluorescent marking pigments.

[0012] The beneficial effects of this invention are as follows: 1. The bridge inspection system based on unmanned intelligence described in this invention, by setting up an analysis module, performs multi-level hierarchical recognition and detection on the acquired images, avoiding omissions in the inspection and improving the accuracy and efficiency of the inspection results. At the same time, the remote inspection drone and the near inspection drone work together to quickly identify areas with potential defects during the inspection and re-inspect those areas, thereby acquiring clear and high-resolution images for recognition, improving the efficiency and accuracy of the inspection. Furthermore, the remote inspection drone detects the distance between the near inspection drone and the bridge, avoiding blind spots and collisions between the near inspection drone and the bridge when the near inspection drone is close to the image acquisition area, which has a relatively small obstacle avoidance detection range.

[0013] 2. The bridge inspection system based on unmanned intelligence described in this invention marks the locations of defects on the bridge surface by setting a marking device, making it easier for subsequent staff to find the locations of defects and repair them. Attached Figure Description

[0014] The invention will now be further described with reference to the accompanying drawings.

[0015] Figure 1 This is a system flowchart of the present invention. Detailed Implementation

[0016] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0017] like Figure 1 As shown, the bridge inspection system based on unmanned intelligence according to the present invention includes: An image acquisition module, comprising a drone equipped with a camera to acquire images of the bridge surface; The analysis module includes an identification unit and a grading unit. The identification unit identifies the collected bridge surface images according to the defect identification model. The identification unit divides the bridge surface images into three categories: intact, questionable, and defective according to the identification results. The grading unit grades the identified bridge defects. The navigation module generates an inspection route based on the 3D model of the bridge to be inspected, and the UAV inspects the bridge according to the inspection route. The drones in the image acquisition module include remote detection drones and near detection drones. The near detection drones are located between the remote detection drones and the bridge surface, and the number of near detection drones is greater than the number of remote detection drones. The near-inspection drone re-captures images of the areas on the bridge surface that are classified as questionable. The near-inspection drone is within the shooting range of the far-inspection drone. The recognition unit detects the distance between the drone and the bridge surface, and between drones, in the acquired images. Before the inspection begins, the navigation module generates an inspection route using a known 3D model of the bridge. Then, the drone in the image acquisition module automatically inspects the bridge and collects images of various surfaces on the bridge according to the inspection route. The collected bridge surface images are then transmitted to the analysis module, where the recognition unit identifies the bridge surface images. Based on the recognition results, the bridge surface images are divided into three categories: intact, questionable, and damaged. Intact means that there are no defects on the bridge surface corresponding to the image; questionable means that it is uncertain whether there are defects on the bridge surface corresponding to the image; and damaged means that there are defects on the bridge surface corresponding to the image. Then, the grading unit classifies the identified defects so that timely reminders can be issued, enabling staff to repair the bridge defects found during the inspection and ensure bridge safety. Meanwhile, since the near-inspection drone is located between the far-inspection drone and the bridge surface, the area of ​​the bridge surface included in the image captured by the far-inspection drone is larger than that included in the image captured by the near-inspection drone. Therefore, the recognition unit identifies and analyzes the bridge surface images captured by the far-inspection drone, finds obvious defects, and classifies the corresponding image types as defects. Similarly, images that the recognition module fails to identify or is uncertain about are classified as questionable. Subsequently, the navigation module plans the inspection route of the near-inspection drone, so that the near-inspection drone gets close to the bridge surface and takes a second, closer-range picture of the area corresponding to the image classified as questionable in the image captured by the far-inspection drone. This allows the recognition unit to re-analyze the relatively high-resolution and clear images to determine whether there are defects on the bridge surface at that location. This avoids misjudgments by the recognition unit due to low resolution or clarity of the images captured by the far-inspection drone, thereby improving the bridge inspection effect and efficiency. Meanwhile, because the proximity inspection drone is relatively close to the bridge surface, the detection range of its sensors is relatively small when it is close to the bridge surface to collect images, making it prone to blind spots. For example, it may fail to detect protrusions or depressions in corners or other nearby proximity inspection drones, causing the obstacle avoidance algorithm to malfunction and leading to contact or collision between the drone and the bridge, affecting its safety. In this situation, since the proximity inspection drone is within the shooting range of the distance inspection drone, its image may appear in the image captured by the distance inspection drone, thus affecting the identification unit's recognition of the distance inspection drone. When analyzing and recognizing images collected by the drone, the system simultaneously identifies and judges the distance between the drone and the bridge surface, as well as the surrounding conditions of the drone. Since there are a relatively large number of drones, the recognition unit also detects the distance between multiple drones in the image and sends this distance or status data to the navigation module. This allows the navigation module to plan the drone's inspection route in a timely and comprehensive manner, improving its obstacle avoidance performance and ensuring its safety and stability when the shooting distance is close and the detection range is limited, thus preventing collisions that could lead to drone crashes.

[0018] In one embodiment of the present invention, the identification unit includes: A model recognition component, wherein the model recognition component is a pre-trained disease recognition model, and the disease recognition model is pre-trained using collected bridge disease images; An anomaly monitoring component identifies the acquired bridge images, and based on the fact that the image occupies most of the normal area, it finds small areas in the image that significantly deviate from the expected distribution and marks them as defects. The anomaly monitoring component also re-identifies images that the model recognition component has determined to be without defects. When identifying the collected bridge surface images, the pre-trained defect identification model in the model identification component is used to identify and analyze the defects, quickly finding the more obvious defects in the images. That is, the images are divided into two types: doubtful and defective, in order to improve inspection efficiency. Then, for images that the defect identification model determines have no abnormalities, they are classified as intact images and input into the anomaly supervision component, so as to re-examine these images and avoid defects that exist in these images but are not in the collected bridge defect images, which would cause the pre-trained defect identification model to fail to identify the existing defects and cause omissions. Meanwhile, when the anomaly monitoring component analyzes and identifies images, since the area occupied by the defects is relatively small, most areas in the image are normal bridge surfaces. That is, most areas in the image are normal areas, and a small part are defect areas. Moreover, there are differences between the defect areas and the normal areas. Therefore, based on the normal areas that occupy most of the image, a dynamic background model is constructed through local statistics, such as local mean, variance, and gray-level distribution. Then, each pixel or local area is compared with the background model, and its deviation degree is calculated, such as residuals and Z-scores. In this way, areas with small areas and significant statistical deviations are judged as anomalies, so as to identify defects that are not in the collected defect feature images, avoid misjudgments and omissions in the analysis module, improve the effectiveness and efficiency of inspection, and ensure bridge safety.

[0019] As one embodiment of the present invention, the anomaly monitoring component includes: In the normal mode, the collected single image is analyzed and identified to find anomalies in the image, namely bridge defects. The re-inspection mode combines multiple collected bridge images to form a re-inspection image. The re-inspection mode analyzes and identifies the re-inspection image to find any abnormalities in the image, i.e., bridge defects. The re-inspection mode also re-identifies images that were deemed intact in the conventional mode. When analyzing the collected images, the system first enters the normal mode to analyze and identify individual images, thus improving identification efficiency, inspection efficiency, and effectiveness. Simultaneously, for images identified as normal (i.e., intact) in the normal mode, a re-inspection mode is initiated. This mode combines images of adjacent areas on the bridge with the original image to form a re-inspection image. Essentially, images that did not identify defects in the normal mode are stitched together to create a re-inspection image with a relatively large area. This re-inspection image is then analyzed and identified to prevent defects from being overlooked in the normal mode due to large areas of defects or the collected image being only a part of a defect, leading to misjudgments. This process improves the efficiency and effectiveness of bridge inspections.

[0020] In one embodiment of the present invention, the navigation module includes a segmentation unit and a planning unit. The segmentation unit segments and divides the three-dimensional model of the bridge into multiple independent working areas, and the planning unit plans inspection routes for the remote inspection UAV and the near inspection UAV respectively. The remote inspection drone and the near inspection drone are scheduled in a time-sharing manner to ensure that the inspection routes do not overlap, and the image acquisition module acquires images of each work area in sequence. The bridge surface is divided into multiple independent working areas by segmentation units. Then, the drone sequentially inspects and collects images of each working area, refining the bridge surface inspection work so that all positions on the bridge surface can be fully inspected. This avoids corners, angles, gaps, and voids on the bridge surface becoming blind spots or dead zones on the inspection route, which would lead to omissions in the bridge inspection and affect the efficiency and effectiveness of the inspection.

[0021] In one embodiment of the present invention, there are overlapping areas between the working areas divided by the slicing unit, and the overlapping area between the working areas is between 2 / 5 and 3 / 5. Because there are overlapping areas between the working areas divided by the slicing unit, when the drone in the image acquisition module is inspecting, the drone will fully acquire images within the working area. This avoids the situation where the drone does not approach the edge of the working area during inspection, and the edge of the working area is also located at the edge of the image in the acquired image. This would cause the recognition unit to easily misjudge or miss the defects located at the edge of the working area when recognizing the acquired image, thus affecting the efficiency and effectiveness of the inspection. Meanwhile, since there are overlapping areas between the work areas, when the drone is collecting images of one work area, it will also capture the edge positions of the adjacent work areas. This avoids the situation where the edge positions of the work areas are also located at the edge of the image, resulting in unclear images that would affect the recognition unit's ability to identify bridge defects and thus impact the effectiveness and efficiency of the inspection.

[0022] In one embodiment of the present invention, the analysis module further includes an analysis unit, which analyzes the acquired bridge surface images to identify the parts with poor image quality, and provides improvement conditions for the shooting based on the poor quality images. The navigation module also includes an adjustment unit, which adjusts the inspection route of the UAV based on the improvement conditions given by the preprocessing unit. The adjustment unit makes adjustments based on the inspection route given by the planning unit. After the UAV completes the inspection of its work area, it flies and collects images based on the inspection route adjusted by the adjustment unit. The analysis unit analyzes the acquired images to identify parts with poor image quality, such as out-of-focus, wide-angle distortion, blur, and low resolution. Then, it further analyzes the images with poor quality and provides improvement conditions for shooting, such as increasing or decreasing the shooting distance, in order to improve the quality of the acquired images, thereby facilitating the identification of changes and improving the efficiency and effectiveness of inspections. Meanwhile, the adjustment unit adjusts the inspection route of the drone in real time according to the improvement conditions given by the analysis unit, so that the drone is always in the best shooting position, improving the quality of the acquired images, making it easier to identify the defects in the images, and improving the efficiency and effectiveness of the inspection.

[0023] As one embodiment of the present invention, the near-inspection UAV is equipped with a flaw detection device, which includes at least one of a laser detection device, an ultrasonic detection device, and a radar detection device. Because bridge surface defects are diverse and vary in severity—such as rust, water stains, and coating peeling—their texture and color characteristics are extremely similar to real cracks, making them difficult to distinguish when analyzing and identifying the collected images. This can easily lead to misjudgments, affecting the effectiveness and efficiency of inspections. In this context, when re-collecting images of areas on the bridge surface identified as suspicious by the identification unit, a flaw detection device mounted on a close-range inspection drone can simultaneously detect the suspicious areas while re-collecting images at close range. This, combined with the identification unit's analysis of the re-collected images, provides a basis for judgment, thereby improving the detection effectiveness and efficiency of potential bridge defects and preventing omissions during inspections. Furthermore, the flaw detection device accurately measures existing defects, such as crack width, length, and peeling area, to determine the degree of crack development, facilitating subsequent analysis unit-level classification and alerts for defects.

[0024] As one embodiment of the present invention, the near-inspection drone is equipped with a marking device. The marking device is not installed on the same drone as the flaw detection equipment. The marking device marks the identified defects at the corresponding positions on the bridge surface. The marking device marks by launching paint eggs, which are filled with fluorescent marking pigments. Because bridge defects vary in size, shape, and location, smaller defects in remote locations can be difficult for workers to locate, hindering repair efforts. During inspections, the grading unit classifies bridge defects and issues alerts, transmitting this information via an image acquisition module to a marking device on a drone. This device then launches a marker containing a fluorescent ink pellet, propelled by compressed air stored in a small cylinder. The pellet impacts the defect, breaking it and releasing the fluorescent pigment inside, creating a clear mark on the bridge surface. This facilitates easy and quick location of defects during subsequent repairs. Furthermore, since multiple drones are deployed for close-in inspections, some can be equipped with marking devices. This allows for easy marking of defects discovered during inspections, avoiding the overload and control issues that would result from simultaneously installing marking devices and flaw detection equipment.

[0025] The specific workflow is as follows: Before the inspection begins, the navigation module generates an inspection route using a known 3D model of the bridge. Then, the drone in the image acquisition module automatically inspects the bridge and collects images according to the inspection route. After that, the recognition unit identifies the bridge surface images and classifies them into three categories: intact, questionable, and damaged, based on the recognition results. Then, the grading unit classifies the identified damages to issue timely reminders. Meanwhile, the identification unit identifies and analyzes the bridge surface images collected by the remote inspection drone, finds obvious defects and classifies the corresponding image types as defects. Similarly, images that the identification module fails to identify or is uncertain about are classified as questionable. Then, the navigation module plans the inspection route of the near inspection drone, so that the near inspection drone gets close to the bridge surface and collects images again for analysis and identification of the areas corresponding to the images classified as questionable in the images taken by the remote inspection drone. Meanwhile, when the identification unit analyzes and identifies the images collected by the remote inspection drone, it will simultaneously identify and judge the distance between the near inspection drone and the bridge surface and the state around the near inspection drone in the image, and send the distance or state data to the navigation module so that the navigation module can plan the inspection route of the near inspection drone in a timely and sufficient manner to avoid collisions. When identifying the collected bridge surface images, the pre-trained defect identification model in the model identification component is used to identify and analyze the defects to quickly find the more obvious defects in the images. Then, for images that the defect identification model determines are not abnormal, they are input into the anomaly supervision component for re-examination. Meanwhile, when the anomaly monitoring component analyzes and identifies the image, it identifies the abnormal parts that occupy a small area based on the normal area that occupies most of the image, that is, the diseases that are not in the collected disease feature image. When analyzing the collected images, we first enter the normal mode to analyze and identify the individual images and find out the defects in the images. At the same time, for images that are identified as having no abnormalities or are intact in the normal mode, we enter the re-inspection mode to combine the images of the adjacent areas of the corresponding area on the bridge with the image to form a re-inspection image. Then, we analyze and identify the re-inspection image. The bridge surface is divided into multiple independent working areas by the segmentation unit. Then, the UAV inspects and collects images of each working area in turn, which refines the bridge surface inspection work so as to fully inspect all positions on the bridge surface. Since there are overlapping areas between the working areas divided by the slicing unit, when the drone in the image acquisition module is performing inspection, when the drone is acquiring images of one of the working areas, it will also capture the edge positions of the adjacent working areas to avoid the edge positions of the working areas being located at the edge of the image in the acquired image, and to avoid the image being unclear when the edge positions of the working areas are also located at the edge of the image. The analysis unit analyzes the acquired images to identify the parts with poor image quality. Then, it further analyzes the poor-quality images and provides improvement conditions for shooting. Meanwhile, the adjustment unit adjusts the inspection route of the drone based on the improvement conditions given by the analysis unit, so that the drone is in the best shooting position and the quality of the acquired images is improved. When re-collecting images of areas on the bridge surface that have been classified as questionable by the identification unit, the flaw detection equipment installed on the close-in inspection drone can simultaneously detect the questionable areas while re-collecting images at close range. This, in conjunction with the identification unit's recognition of the re-collected images, provides a basis for judgment. At the same time, the flaw detection equipment can accurately measure existing defects, such as the width, length, and peeling area of ​​cracks, to determine the degree of crack development. Meanwhile, during the inspection, after the grading unit classifies the bridge defects, it will also issue a reminder and transmit the reminder to the marking device on the drone through the image acquisition module. This will cause the marking device to activate and launch the paint egg inside the marking device using compressed air stored in a small air cylinder. The paint egg will then impact the location of the defect on the bridge surface. After the paint egg is impacted and breaks, the fluorescent marker pigment inside will be released, forming a clear mark on the bridge surface.

[0026] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A bridge inspection system based on unmanned intelligence, characterized in that: include: An image acquisition module, comprising a drone equipped with a camera to acquire images of the bridge surface; The analysis module includes an identification unit and a grading unit. The identification unit identifies the collected bridge surface images according to the defect identification model. The identification unit divides the bridge surface images into three categories: intact, questionable, and defective according to the identification results. The grading unit grades the identified bridge defects. The navigation module generates an inspection route based on the 3D model of the bridge to be inspected, and the UAV inspects the bridge according to the inspection route. The drones in the image acquisition module include remote detection drones and near detection drones. The near detection drones are located between the remote detection drones and the bridge surface, and the number of near detection drones is greater than the number of remote detection drones. The near-inspection drone re-captures images of the areas on the bridge surface that are classified as questionable. The near-inspection drone is within the shooting range of the far-inspection drone. The identification unit detects the distance between the drone and the bridge surface, and between drones, in the acquired images.

2. The bridge inspection system based on unmanned intelligence according to claim 1, characterized in that: The identification unit includes: A model recognition component, wherein the model recognition component is a pre-trained disease recognition model, and the disease recognition model is pre-trained using collected bridge disease images; An anomaly monitoring component identifies the acquired bridge images, and based on the fact that the image occupies most of the normal area, it identifies and marks the small areas in the image that significantly deviate from the expected distribution as defects. The anomaly monitoring component then re-identifies images that the model recognition component has determined to be defect-free.

3. The bridge inspection system based on unmanned intelligence according to claim 2, characterized in that: The anomaly monitoring component includes: In the normal mode, the collected single image is analyzed and identified to find anomalies in the image, namely bridge defects. The re-inspection mode combines multiple collected bridge images to form a re-inspection image. The re-inspection mode analyzes and identifies the re-inspection image to find any abnormalities, i.e., bridge defects. The re-inspection mode also re-identifies images that were deemed intact in the conventional mode.

4. The bridge inspection system based on unmanned intelligence according to claim 1, characterized in that: The navigation module includes a segmentation unit and a planning unit. The segmentation unit segments and divides the three-dimensional model of the bridge into multiple independent work areas. The planning unit plans inspection routes for remote inspection drones and near inspection drones respectively. The remote inspection drone and the near inspection drone are scheduled in a time-sharing manner so that the inspection routes do not overlap, and the image acquisition module acquires images of each work area in sequence.

5. The bridge inspection system based on unmanned intelligence according to claim 4, characterized in that: The work areas divided by the slicing unit have overlapping regions, and the overlapping regions between the work areas are between 2 / 5 and 3 / 5.

6. The bridge inspection system based on unmanned intelligence according to claim 4, characterized in that: The analysis module also includes an analysis unit, which analyzes the acquired bridge surface images to identify the parts with poor image quality, and provides improvement conditions for the shooting process based on the analysis of the poor quality images. The navigation module also includes an adjustment unit, which adjusts the inspection route of the UAV based on the improvement conditions given by the preprocessing unit. The adjustment unit adjusts the inspection route based on the inspection route given by the planning unit. After the UAV completes the inspection of its work area, it flies and collects images according to the inspection route adjusted by the adjustment unit.

7. The bridge inspection system based on unmanned intelligence according to claim 1, characterized in that: The near-inspection drone is equipped with flaw detection equipment, which includes at least one of laser detection equipment, ultrasonic detection equipment, and radar detection equipment.

8. The bridge inspection system based on unmanned intelligence according to claim 7, characterized in that: The close-in inspection drone is equipped with a marking device. The marking device is not installed on the same drone as the flaw detection equipment. The marking device marks the identified defects at the corresponding positions on the bridge surface. The marking device marks by launching paint eggs, which are filled with fluorescent marking pigments.

Images