Drone-based Automatic Bridge Crack Detection System

Abstract

The present invention relates to the technical field of crack measurement. The present invention discloses an automatic bridge crack detection system based on an unmanned aerial vehicle (UAV), which includes a UAV and a collector disposed at the UAV. The system further includes a trajectory planning module for planning the flight trajectory of the UAV to circle around the bridge piles. The trajectory planned by the trajectory planning module includes a spiral orbit and an extension orbit. A starting target platform is provided at the bottom starting point of each spiral orbit, so that the acquisition trajectory planes formed by the UAV along the spiral orbits complement each other to form a complete or overlapping crack acquisition area. In order to reduce safety risks and improve the detection frequency, by utilizing the extension characteristics of cracks, a form of large pitch, multi-track route, and single-track detection is adopted to increase the detection frequency and reduce the duration of a single detection process. The advantage of high-frequency detection is that it can shorten the detection blank period and reduce safety risks, and it can easily extend to a single spiral orbit, so as to timely detect gap problems.

Description

Technical Field

[0001] The present invention relates to the technical field of crack measurement, and specifically to an automatic bridge crack detection system based on an unmanned aerial vehicle (UAV). Background Art

[0002] UAV wall-attached detection is the main means for bridge crack detection. After wall-attaching, since the distance between the camera and the target object is short, the acquisition area is small but the image clarity is high. At this time, the number and size of cracks are obtained through image processing calculations, thereby improving the safety of the bridge.

[0003] Since the more times and the shorter the frequency of bridge detection are, the easier it is to discover potential safety hazards. However, due to the small acquisition area during wall-attached flight, if the UAV conducts a complete and comprehensive acquisition once, the flight distance is too long and high-frequency detection cannot be achieved.

[0004] Since the detection trajectory of the bridge deck is fixed and single, the road surface can be collected by a collection vehicle in a wall-attached manner, and multiple cameras are collected at one time. Under the bridge, since the detection trajectory is fixed and single, a track can be set on the side or under the bridge to control the movement of a linear collector for collection. However, it is very difficult to use a linear collector to collect a cylindrical bridge pile. When linear collection is adopted, a closed loop is formed, making it difficult to separate from the bridge pile. Since the number of bridge piles is large, the form of linear collection is not applicable to bridge piles. The bridge pile still chooses single-image track collection for crack detection. Therefore, during comprehensive detection, the flight distance is still very long and high-frequency detection cannot be achieved. Summary of the Invention

[0005] The purpose of the present invention is to provide an automatic bridge crack detection system based on an unmanned aerial vehicle to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solution:

[0007] An automatic bridge crack detection system based on an unmanned aerial vehicle includes the unmanned aerial vehicle and a collector arranged at the unmanned aerial vehicle, and further includes a trajectory planning module for planning the flight trajectory of the unmanned aerial vehicle around the bridge pile;

[0008] The trajectory set by the trajectory planning module includes a spiral track and an extension track. Among them, the number of the spiral tracks is at least two. A starting target platform is arranged at the bottom starting point of each spiral track. A plurality of the starting target platforms are annularly and equally angularly distributed on the bottom surface of the bridge pile. A plurality of the spiral tracks are mutually staggered, so that the acquisition trajectory surfaces formed by the unmanned aerial vehicle along the spiral tracks complement each other to form a complete or overlapping crack acquisition area;

[0009] After the data collected by the collector is processed by the target orientation processing module for image processing, when there is a single endpoint in any crack in the image, the orientation of the crack in the image is obtained, and the trajectory planning module is used to plan the UAV to move along the crack orientation using the extension track until a complete crack image is obtained and the crack length is measured.

[0010] As a further solution of the present invention: the flight coordinates of the UAV are recorded by the coordinate positioning module. The coordinate positioning module horizontally and circularly divides the bridge pile into multiple horizontal positioning points through the circular coordinate calculation unit. The height and distance measuring unit is installed at the bottom of the UAV to monitor the real-time height of the UAV from the horizontal plane of the starting target platform. The coordinate positioning module can know the current orientation of the UAV based on the position of the starting target platform where the UAV takes off and the actual height of the UAV monitored by the height and distance measuring unit.

[0011] As a further solution of the present invention: under normal conditions, each time the UAV flies, each pile body sequentially selects the starting target platform as the take-off point in order. Under abnormal conditions, the initial take-off point of the abnormal pile body selects the starting target platform corresponding to the abnormal spiral crack collection area as the starting take-off point.

[0012] As a further solution of the present invention: the collection punctuation marks of the starting target platform of the same pile body are different from each other. The marked positions of the starting target platform are recorded by the trajectory planning module, and the flight track to the starting target platform is planned.

[0013] As a further solution of the present invention: the flight track of the extension track includes a circular flight track around the pile and an axial flight track along the pile body.

[0014] As a further solution of the present invention: the target orientation processing module divides the image into four recognition areas, and obtains the single-end gap position and the gap extension azimuth angle in the four recognition areas. The timing inspection unit changes the take-off order of the starting target platform according to the image recognition information of the coordinate positioning module, and the trajectory planning module controls and selects the spiral track or the extension track as the flight track of the UAV according to the image recognition information of the coordinate positioning module.

[0015] As a further solution of the present invention: the bottom surface of the pile body is fixedly connected with a horizontal plane, the starting target platform is embedded inside the horizontal plane, and the top surface of the starting target platform is flush with the horizontal plane.

[0016] Compared with the prior art, the beneficial effects of the present invention are:

[0017] In order to reduce safety risks and increase the detection frequency, taking advantage of the extension characteristics of cracks, a large pitch, multi-track route, and single-track detection form are adopted to increase the detection frequency and reduce the duration of a single detection process. The benefit of high-frequency detection is that it can shorten the detection blank period and reduce safety risks. At the same time, the spiral ascending method has two-way movement in the horizontal and vertical directions, while the gaps are actually irregular, and most gaps will have horizontal or vertical extensions. Therefore, it is easy to extend to a single spiral track, thus timely detecting gap problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 Sweeping trajectory schematic diagram of the bridge crack automatic detection system based on an unmanned aerial vehicle;

[0020] Figure 2 Three-track extension schematic diagram of the bridge crack automatic detection system based on an unmanned aerial vehicle;

[0021] Figure 3 Dual-track extension schematic diagram of the bridge crack automatic detection system based on an unmanned aerial vehicle;

[0022] Figure 4 Bridge crack automatic detection system based on an unmanned aerial vehicle;

[0023] Figure 5 Graphic partition schematic diagram of the bridge crack automatic detection system based on an unmanned aerial vehicle;

[0024] In the figure: 1, start the target platform; 100, unmanned aerial vehicle; 101, collector; 2, trajectory planning module; 21, spiral track; 22, extension track; 23, regular inspection unit; 3, target orientation processing module; 4, coordinate positioning module; 41, height and distance measurement unit; 42, circumferential coordinate calculation unit. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] Please refer to Figures 1 - 5 , in this embodiment, it includes an unmanned aerial vehicle 100 and a collector 101 disposed at the unmanned aerial vehicle 100, and also includes a trajectory planning module 2 for planning the flight trajectory of the unmanned aerial vehicle 100 around the bridge pile.

[0026] In this embodiment, the current drone has a function of ascending in a spiral flight. When the drone ascends in a spiral and flies along the wall of the bridge pile, the collected image band is a spiral band. Please refer to Figures 1 - 3 , and the width of the spiral band is obtained by obtaining the image acquisition width according to the wall-attached spacing. Through Figure 1 It can be seen that if the drone comprehensively collects the pile body, in the case of a large pitch, more than two spiral bands are required. If the pitch is reduced, one spiral band can complete the comprehensive collection, but the flight speed is too slow and the detection period is long. In order to reduce the safety risk and improve the detection frequency, taking advantage of the extension characteristics of cracks, a form of large pitch, multi-track route, and single-track detection is adopted to improve the detection frequency and reduce the duration of a single detection process. The advantage of high-frequency detection is that it can shorten the detection blank period and reduce the safety risk. At the same time, the spiral ascending method has two-way movements in the horizontal and vertical directions, while the actual gaps are irregular, and most gaps will have horizontal or vertical extensions. Therefore, it is easy to extend to a single spiral track, so as to detect gap problems in time.

[0027] In this embodiment, the trajectory planning module 2 sets the trajectory including a spiral track 21 and an extension track 22. Among them, the trajectory of the spiral track 21 has at least two. A starting target platform 1 is set at the bottom starting point of each spiral track 21. A number of starting target platforms 1 are annularly and equally angularly distributed on the bottom surface of the bridge pile. A number of spiral tracks 21 are staggered from each other, so that the collection trajectory surface formed by the drone 100 along the spiral track 21 complements each other to form a complete or overlapping crack collection area.

[0028] In this embodiment, in order to accurately distinguish and plan the track of a single detection, a starting target platform 1 is set at the bottom of the bridge pile. The number of starting target platforms 1 is the same as the number of spiral tracks 21. The existing drones have a function of target fixed-point hovering. At present, the target object is comprehensively recognized for hovering through image recognition combined with image recognition algorithms and flight control algorithms. Hovering also includes fixed-point hovering and fixed-height hovering. The purpose of setting the starting target platform 1 is to facilitate the trajectory planning module 2 to sort the trajectories of multiple piles. Using the starting target platform 1 as the starting point for multiple piles makes the trajectory more accurate. And the trajectory planning module 2 only needs to set the trajectory of flying from the starting target platform 1 of one pile to the starting target platform 1 of another pile. The trajectory flown through the starting target platform 1 is a fixed spiral trajectory.

[0029] In this embodiment, in addition to the spiral track 21, an extension track 22 is also set. Since during the flight, it is easy to find complete gaps and single-end gaps in the image acquisition interface. Please refer to Figure 5, the complete trajectory has two endpoints, while other fissures will present different single-end gaps, and the ambiguous orientation of the gap relative to the UAV can be judged by the extension direction of the single-end gap. Therefore, it is necessary to use the extension track 22 to present the fissure completely.

[0030] In this embodiment, Figures 1 - 3 The image drawn shows double-helix and triple-helix spliced to cover the pile body. To further improve the comprehensiveness of a single acquisition and reduce the process duration, a double-helix track is adopted. The pitch of the single helix is reduced, and the pitch width is made smaller than the width of the helical acquisition surface, so that the width of the remaining blank area is smaller than the width of the subsequent double-helix track. The simplest way to adjust the width of the helical track is to adjust the wall-attached spacing. By increasing the size of the acquired image, the width of the helical surface can be adjusted. At this time, during a single acquisition, the width of the blank helical surface is narrow, and it is easier to find the endpoints of the fissures. When extending the fissure endpoints, a complete fissure can be obtained, thereby improving the comprehensiveness of the acquisition while achieving non-complete acquisition. During actual operation, the pitch can be adjusted according to the maximum size of the allowed fissures.

[0031] In this embodiment, after the data collected by the collector 101 is processed by the target orientation processing module 3 for image processing, when any fissure in the image has a single endpoint, the orientation of the fissure in the image is obtained, and the trajectory planning module 2 plans for the UAV 100 to move along the fissure orientation using the extension track 22 until a complete fissure image is obtained and the length of the fissure is measured. The target orientation processing module 3 divides the image into four recognition areas, and obtains the position of the single-end gap and the gap extension azimuth angle in the four recognition areas. The regular inspection unit 23 changes the takeoff order of the starting target platform 1 according to the image recognition information of the coordinate positioning module 4. The trajectory planning module 2 controls the selection of the helical track 21 or the extension track 22 as the flight track of the UAV 100 according to the image recognition information of the coordinate positioning module 4. Under normal conditions, each time the UAV 100 flies, each pile body sequentially selects the starting target platform 1 as the takeoff point in order. Under abnormal conditions, the initial takeoff point of the abnormal pile body selects the starting target platform 1 corresponding to the abnormal helical crack acquisition area as the starting takeoff point.

[0032] In this embodiment, for the convenience of understanding, the track selection priorities for the double helix and triple helix are explained as follows:

[0033] Double helix: The double helix is more convenient. When Figure 5 a problem occurs, assuming Figure 5 it moves in the upper right direction. The fissure areas in the upper left of area A, the lower right of area C, and the lower left of area D cannot be collected. Therefore, during the first inspection, the central fissure information and the fissure information in the upper right of area B can be collected. Figure 5 The priority for the second acquisition then selects the second helical track to collect the remaining image information of A, C, and D.

[0034] Triple helix: In the triple helix, assume that Figure 5 is located on track b, then the image collected in area A is located on track a, and the images collected in areas D and C are located on track c. Therefore, when the number of track c is greater than that of track a, the priority of the next collection track is track c, and the fissure image is quickly supplemented through track c.

[0035] In this embodiment, an explanation is given for image extension:

[0036] Still with the help of Figure 5 , the flight track of the extension track 22 includes a circumferential flight trajectory around the pile and an axial flight trajectory along the pile body. When there is a problem with the image in area A, the extension track 22 can be used for extension and supplementation. By spiraling horizontally to the left and then moving vertically upward, the residual information of image A can be collected. For the gap located on the vertical center line, the complete fissure image can be directly collected by moving vertically upward. This method mainly improves the comprehensiveness of fissure detection during the single helix detection process. When the extension track 22 is flying, it is necessary to completely record the flight coordinates, or prompt manual operation, or use the method of identifying the flight with markers for extension. After finding the other end point of the gap, reset. During the extension flight process, the image can be measured to obtain the accurate gap length.

[0037] In this embodiment, an explanation is given for image extension: The flight coordinates of the unmanned aerial vehicle 100 are recorded by the coordinate positioning module 4. The coordinate positioning module 4 horizontally divides the bridge pile into multiple horizontal positioning points through the circumferential coordinate calculation unit 42. The height measurement and ranging unit 41 is installed at the bottom of the unmanned aerial vehicle 100 to monitor the real-time height of the unmanned aerial vehicle 100 from the horizontal plane of the starting target platform 1. The coordinate positioning module 4 can know the current orientation of the unmanned aerial vehicle 100 based on the position of the starting target platform 1 where the unmanned aerial vehicle 100 takes off and the actual height of the unmanned aerial vehicle 100 monitored by the height measurement and ranging unit 41.

[0038] In this embodiment, after all tracks have flown once, the areas where cracks are collected during the flight process can be marked. Therefore, at the bridge pile, it is divided into multiple positioning points by the circumferential coordinate calculation unit 42. According to the flight track and combined with the flight height position, in the case of a fixed spiral line, the current circular coordinate position of the unmanned aerial vehicle 100 can be obtained. The circumferential coordinate calculation unit 42 is a computer, which mainly obtains the current coordinates through the height measurement and ranging unit 41 combined with the spiral track. After integrating the coordinates, the coordinates of multiple problem points of the entire bridge body can be marked, so as to perform inspection of the marked points.

[0039] In this embodiment, the acquisition punctuation points of the starting target platform 1 of the same pile body are different from each other. The marked points of the starting target platform 1 are recorded by the trajectory planning module 2, and the flight trajectory to the starting target platform 1 is planned; it is more convenient to control through different marked points.

[0040] In this embodiment, a horizontal plane is fixedly connected to the bottom surface of the pile body, and the starting target platform 1 is embedded inside the horizontal plane, and the top surface of the starting target platform 1 is flush with the horizontal plane. In order to avoid errors in the acquisition of the height and distance measuring unit 41 caused by the starting target platform 1 being lower or higher than the horizontal plane.

[0041] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and should be covered within the protection scope of the present invention.

Claims

1. An automatic bridge crack detection system based on an unmanned aerial vehicle, comprising an unmanned aerial vehicle (100) and a collector (101) arranged at the unmanned aerial vehicle (100), characterized in that: It also includes a trajectory planning module (2) for planning a flight trajectory of the drone (100) around a bridge pile; The trajectory planning module (2) sets a trajectory including a spiral track (21) and an extended track (22), wherein the spiral track (21) has at least two trajectories, a starting target platform (1) is provided at the bottom starting point of each spiral track (21), a plurality of the starting target platforms (1) are distributed in a circular manner at equal angles on the bottom surface of the bridge pile, and a plurality of the spiral tracks (21) are mutually staggered, so that the collection trajectory surfaces formed by the drone (100) along the spiral tracks (21) complement each other to form a complete crack collection area or a crack collection area with overlapping areas; After the data collected by the collector (101) is processed by the target orientation processing module (3), when any crack in the image has a single endpoint, the orientation of the crack in the image is obtained, and the trajectory planning module (2) is used to plan the drone (100) to move along the crack orientation using the extended track (22) until a complete crack image is obtained and the crack length is measured; The flight coordinates of the unmanned aerial vehicle (100) are recorded by a coordinate positioning module (4), wherein the coordinate positioning module (4) comprises a height-fixing distance-measuring unit (41) and a circumferential coordinate calculating unit (42); The target orientation processing module (3) divides the image into four identification areas, and obtains the single-end gap position and the gap extension azimuth in the four identification areas. The timing inspection unit (23) changes the take-off order of the start target platform (1) according to the image recognition information of the coordinate positioning module (4). The trajectory planning module (2) controls the selection of the spiral track (21) or the extension track (22) as the flight track of the UAV (100) according to the image recognition information of the coordinate positioning module (4).

2. The automatic bridge crack detection system based on drone according to claim 1 is characterized by: The coordinate positioning module (4) divides the horizontal ring of the bridge pile into a plurality of horizontal positioning points through a circumferential coordinate calculation unit (42), and the height determination and ranging unit (41) is installed at the bottom of the drone (100) to monitor the real-time height of the drone (100) from the horizontal plane of the starting target platform (1).

3. The automatic bridge crack detection system based on drone according to claim 2 is characterized by: The coordinate positioning module (4) obtains the current position of the drone (100) based on the position of the starting target platform (1) from which the drone (100) takes off, combined with the actual height of the drone (100) monitored by the altitude-fixing and ranging unit (41).

4. The automatic bridge crack detection system based on drone according to claim 3 is characterized by: Under normal conditions, each time the drone (100) flies, each pile body sequentially selects the starting target platform (1) as a take-off point. Under abnormal conditions, the initial take-off point of an abnormal pile body selects the starting target platform (1) corresponding to the abnormal spiral crack collection area as the starting take-off point.

5. The automatic bridge crack detection system based on drone according to claim 4 is characterized in that: The collection marks of the starting target platform (1) of the same pile body are different from each other, and the marking points of the starting target platform (1) are recorded by the trajectory planning module (2), and the flight trajectory to the starting target platform (1) is planned.

6. The automatic bridge crack detection system based on drone according to claim 1 is characterized by: The flight track of the extension track (22) comprises a circular flight track around the pile and a flight track along the axial direction of the pile body.

7. The automatic bridge crack detection system based on drone according to claim 1 is characterized by: The bottom surface of the pile body is fixedly connected to a horizontal plane, the starting target platform (1) is embedded in the horizontal plane, and the top surface of the starting target platform (1) is flush with the horizontal plane.

Images