Bridge area low-altitude complex environment combined navigation method

Abstract

The invention relates to the field of bridge area integrated navigation, and discloses a bridge area low-altitude complex environment integrated navigation method which comprises the following steps: synchronously aligning multi-source sensor data; multi-source data are fused for autonomous positioning of the inspection unmanned aerial vehicle; performing data storage based on a preset storage rule; planning the flight path of the unmanned aerial vehicle based on a full-coverage path planning algorithm; after the flight path planning of the unmanned aerial vehicle is completed, the unmanned aerial vehicle starts to move and completes autonomous navigation based on the autonomous following path model of the inspection unmanned aerial vehicle; based on the autonomous obstacle avoidance technology of the inspection unmanned aerial vehicle and the visual image, autonomous obstacle avoidance of the inspection unmanned aerial vehicle is performed by fusing the autonomous obstacle avoidance technology of the laser radar. According to the invention, the three-dimensional laser radar inspection unmanned aerial vehicle and the visual image can be fused, and accurate navigation positioning is provided for the unmanned aerial vehicle.

Description

Technical Field

[0001] The invention relates to the field of bridge area integrated navigation, and in particular to a bridge area low-altitude complex environment integrated navigation method. Background Art

[0002] With the rapid development of drone technology, drones are increasingly used in bridge inspection, patrol, maintenance and other fields. As an important transportation infrastructure, the safety and reliability of bridges are of vital importance. Traditional bridge inspection methods mainly rely on manual inspection and ground vehicle inspections, which have problems such as low efficiency, high cost, and difficulty in covering large areas. With its flexibility, high efficiency and low cost, drones have become an ideal tool for bridge inspection. However, bridge areas usually have complex low-altitude environments, such as tall piers, dense cables, and changeable weather conditions, which pose severe challenges to the navigation and positioning of drones. In order to ensure the safe flight and precise navigation of drones in bridge areas, it is particularly necessary to develop a combined navigation method suitable for low-altitude complex environments. Summary of the invention

[0003] The present invention aims to provide a combined navigation method for low-altitude complex environments in bridge areas, so as to integrate three-dimensional laser radar inspection UAVs and visual images to provide accurate navigation and positioning for the UAVs.

[0004] In order to achieve the above object, the present invention adopts the following technical scheme:

[0005] A combined navigation method for low-altitude complex environment in a bridge area, comprising:

[0006] S1, synchronize and align multi-source sensor data;

[0007] S2, integrates multi-source data for autonomous positioning of inspection drones;

[0008] S3, data storage based on preset storage rules;

[0009] S4, UAV flight path planning based on full coverage path planning algorithm;

[0010] S5, after the UAV flight path planning is completed, the UAV starts to move and completes autonomous navigation based on the inspection UAV autonomous following path model;

[0011] S6, based on the autonomous obstacle avoidance technology of the inspection drone and visual images, integrates the autonomous obstacle avoidance technology of the lidar to perform autonomous obstacle avoidance for the inspection drone.

[0012] The principles and advantages of this solution are: in actual application, the multi-source sensor data are synchronized and aligned to make the trigger time of each sensor consistent; the multi-source data are fused for autonomous positioning of the inspection UAV, which is convenient for predicting and updating the status of the inspection UAV; data is saved based on preset saving rules, so that data that meets the specifications can be saved when necessary; the bridge inspection UAV autonomous navigation completes path planning based on the full coverage path planning algorithm after obtaining the raster map, providing a path source for subsequent path tracking; after the UAV flight path planning is completed, the UAV starts moving and completes autonomous navigation based on the inspection UAV autonomous following path algorithm; there may be a risk of encountering obstacles during the autonomous navigation of the bridge inspection UAV, and the image sensor continuously acquires visual images. The inspection UAV's autonomous obstacle avoidance technology integrates the laser radar to perform autonomous obstacle avoidance for the inspection UAV.

[0013] Preferably, as an improvement, S2 includes:

[0014] S21, collect original data and conduct bridge inspection and survey to obtain inspection task planning;

[0015] S22, acquiring three-dimensional digital terrain model data, orthophoto data, and three-dimensional spatial position data of bridges through multi-source sensors, and constructing a three-dimensional scene based on lidar point cloud data;

[0016] S23, based on the coordination of laser point cloud, multi-eye visible light and deep multi-source data, locks the inspection drone path;

[0017] S24 achieves autonomous positioning using a combined navigation and positioning algorithm based on global positioning, inertial navigation, real-time dynamic differential global positioning system, lidar odometer and visual odometer.

[0018] Technical effect: Through autonomous positioning of the inspection UAV, it is convenient to estimate the attitude, position and speed of the inspection UAV.

[0019] Preferably, as an improvement, S5 includes:

[0020] S51, during the flight of the inspection UAV executing path following, the 3D laser radar point cloud data is acquired in real time and posture compensation is performed to achieve real-time update of the global point cloud;

[0021] S52, classify the point cloud data, combine the three-dimensional position relationship and flight control algorithm of the inspection drone, and control the inspection drone to follow the path line for autonomous navigation flight.

[0022] Technical effect: Compared with the traditional manual route planning technology, autonomous navigation flight can effectively make up for various deficiencies, especially solving the poor adaptability of traditional technology that relies on highly reliable mobile digital networks. In addition, by analyzing and optimizing the best path for drone inspection routes, more accurate and robust autonomous navigation is achieved.

[0023] Preferably, as an improvement, the controlling the inspection UAV to follow the path line for autonomous navigation flight also includes: inputting the relative position relationship and posture relationship between the reference objects into a PID control algorithm to obtain the control amount of each control channel of the inspection UAV.

[0024] Technical effect: It is easy to control the flight path to achieve path tracking in the horizontal and vertical directions.

[0025] Preferably, as an improvement, the autonomous following path model of the drone is:

[0026] D OA-OB =|OA-OB|#(19)

[0027] D AA′-BB′ =|AA′-BB′|#(20)

[0028]

[0029] Among them, OA is the distance from point A on the path to the drone; OB is the distance from point B on the path to the drone; AA' is the distance from point A to the left of the drone's direction of travel; BB' is the distance from point B to the right of the drone's direction of travel; D OA-OB is the distance difference between the drone and points A and B; D AA′-BB′ is the difference in distances that points A and B deviate to the left and right in the direction of the UAV’s travel; R is the distance between the wire and the UAV; ω is the angle along the direction of the UAV’s travel when the airborne laser radar scans the bridge; α is the angle perpendicular to the direction of the UAV’s travel when the airborne laser radar scans the wire; x is the distance that the point on the wire scanned by the laser radar deviates to the left and right in the direction of the UAV’s travel;

[0030] When D AA′-BB′ <-δ, the drone’s attitude is on the right side of the path and needs to fly left; when D AA′-BB′ >δ, the UAV is on the left side of the path and needs to move to the right; when -δ <D AA′-BB′ <δ, the drone’s attitude is aligned with the path above.

[0031] Technical effect: It is convenient to make timely and effective adjustments to the posture of the inspection drone.

[0032] Preferably, as an improvement, S6 includes:

[0033] S61, acquiring three-dimensional spatial position information in real time and synchronously feeding it back to the edge computing unit;

[0034] S62, the edge computing unit calculates the distance between the inspection drone and the obstacle;

[0035] S63, generating a control amount based on the distance to control the inspection drone to maintain a safe distance from the obstacle.

[0036] Technical effect: To facilitate inspection drones to avoid obstacles in time.

[0037] Preferably, as an improvement, the method further includes S7, calculating the shooting point position, obtaining a new height value through the relative height and the rotation angle, obtaining the relative rotation point according to the (x, y, z) value of the center point of the shooting target component in the bridge, rotating the camera relative to the center point, and obtaining a new camera posture parameter, wherein the calculation method is:

[0038] Get the radian r of the horizontal deviation:

[0039]

[0040] Get the distance d from the target point:

[0041] d=dcosr

[0042] Get the location of the shooting point:

[0043] Prot(x,y,z)=(x-dtanr,y+d,z)

[0044] Get the parameter value of the plane corresponding to the camera point and calculate the value of the center point of the plane:

[0045] Surface(x,y)=(x+d ​​sinr,yd cosr)

[0046] Get the side length of a plane:

[0047] Side=0.01*DIS pp

[0048] Get the (x, y, z) values ​​of the four corner points of the plane:

[0049]

[0050] Technical effect: It can accurately obtain the shooting position of the inspection drone, combine the inspection drone camera parameter values, calculate the camera position of all shooting target points, consider safety and obstacle avoidance factors, and automatically generate the best safety inspection path for the drone. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1It is a structural schematic diagram of an embodiment of the present invention;

[0052] Figure 2 A schematic diagram of multi-sensor data flow for a combined navigation method in a low-altitude complex environment in a bridge area;

[0053] Figure 3 This is the principle block diagram of the INS-GNSS deep coupling positioning system;

[0054] Figure 4 This is a schematic diagram of the principle flow of the inspection drone following technology;

[0055] Figure 5 Schematic diagram of the UAV path following model.

[0056] Figure 6 This is a flowchart of obstacle avoidance for inspection drones based on 3D LiDAR;

[0057] Figure 7 The timing process of autonomous obstacle avoidance for inspection drones that integrate visual images and lidar is shown in the figure;

[0058] Figure 8 It is a timing flow chart of visual image processing;

[0059] Fig. 9 This is the timing flow chart of visual image obstacle avoidance;

[0060] Fig.10 Provide a plan view of the location and field of view for the inspection drone;

[0061] Fig.11 Optimal safe inspection path process for inspection drones;

[0062] Fig.12 Point cloud diagram for path planning under the bridge Figure 1 ;

[0063] Fig.13 Point cloud diagram for path planning under the bridge Figure 2 . DETAILED DESCRIPTION

[0064] The following is further described in detail through specific implementation methods:

[0065] The embodiment is basically as shown in the attached Figure 1 As shown:

[0066] A combined navigation method for low-altitude complex environment in a bridge area, comprising:

[0067] S1, synchronize and align multi-source sensor data; Figure 2As shown, specifically, the antenna receives satellite signals and base station signals and provides GNSS and RTK module time, position, ephemeris and other information to the GNSS / IMU and RTK modules; the GNSS / IMU and RTK modules send the GNSS time and base station time PPS signals to the CPLD on the main control board, and the IMU synchronizes the attitude information to the CPLD, and the CPLD sends the information of the recorded time and attitude to the processing unit CPU on the main control board; the main control performs a preliminary comparison and fusion of the GNSS time and the RTK time into the system time, and then sends the PPS and the processed time to the laser radar to achieve the time synchronization of the laser radar. The multi-eye camera sends the visible light image and the depth image to the main board through USB, and the visible light sensor and the depth sensor send the exposure signal to the main board, and the main board then matches the time information triggered by the signal with the image data. Through the above process, the time of POS, three-dimensional laser radar, binocular visible light, and binocular depth is unified into the main board system time, achieving the purpose of time synchronization of multiple data sources.

[0068] S2, integrates multi-source data for autonomous positioning of inspection drones; Figure 3 As shown, it specifically includes: S21, collecting original data and conducting bridge inspection and survey, and customizing inspection task planning; S22, acquiring three-dimensional digital terrain model data, orthophoto data and three-dimensional spatial position data of bridges through multi-source sensors; S23, locking the inspection drone path based on the coordination of laser point cloud, multi-eye visible light and deep multi-source data; S24, the inspection drone uses a combined navigation and positioning algorithm based on global positioning system (GNSS), inertial navigation system (INS), real-time dynamic differential global positioning system (RTK), lidar odometer (Lidar-Odometry) and vision odometer (Vision-Odometry) to achieve autonomous positioning.

[0069] S3, data is saved based on preset saving rules so as to save data that meets the specifications when necessary.

[0070] S4, UAV flight path planning based on full coverage path planning algorithm.

[0071] S5, after the UAV flight path planning is completed, the UAV starts to move and completes autonomous navigation based on the inspection UAV autonomous following path model; specifically, S5 includes:

[0072] S51, during the flight of the inspection drone performing path following, the three-dimensional laser radar point cloud data is obtained in real time through laser radar scanning, and the posture solution of multi-source data is deeply coupled. During the flight of the inspection drone performing path following, the point cloud is collected in real time and the posture compensation is performed to realize the real-time update of the global point cloud; the posture compensation algorithm uses the attitude matrix and position offset from the carrier system to the geographic system to transform each point of the original point cloud into the geographic system, thereby forming a global point cloud. Subsequently, the point cloud is classified and processed using the embedded computing unit carried on the inspection drone. After classification, the spatial three-dimensional position relationship of the point cloud of obstacles such as the inspection drone, bridges, tree barriers, ground reference objects, and crossings is used, and the flight control algorithm and real-time feedback data of multiple sensors such as visible light are combined to realize the autonomous navigation flight of the inspection drone following the path line. Compared with the traditional manual route planning technology, autonomous navigation flight can effectively make up for various deficiencies, especially solving the problem of poor adaptability of traditional technology relying on highly reliable mobile digital networks. In addition, by analyzing and optimizing the optimal path of the drone inspection route, more accurate and robust autonomous navigation is achieved.

[0073] S52, classify the point cloud data, combine the three-dimensional position relationship of the inspection drone and the flight control algorithm, and control the inspection drone to follow the path line for autonomous navigation flight. Specifically, use the point cloud data to identify the bridge span, abutment, pier and other bridge surface objects in real time to achieve point cloud data classification; obtain the classification results and three-dimensional coordinate data of the bridge point cloud, and the inspection drone performs path tracking flight according to the planned flight path. Perform the relative posture relationship solution of the inspection drone, bridge and obstacles, and calculate the relative position relationship between the inspection drone, bridge, obstacles and ground objects in real time through the high-precision position information and classification results of the point cloud. Since the point cloud itself is high-precision three-dimensional position information, the relative position relationship directly calculates the three-dimensional spatial position information of the bridge and the distance perpendicular to the horizontal direction of the inspection drone through the coordinates of each physical point. Generate multi-channel PID attitude, speed and position control quantities, input the relative position relationship and attitude relationship between each physical object into the PID control algorithm, obtain the control quantity of each control channel of the inspection drone, control it to maintain it in the horizontal and vertical directions of the path, and realize path tracking flight. The straight-line distance between the UAV and the bridge is determined by comparing it to the safety distance threshold. If the straight-line distance is less than the safety distance threshold, it is considered that the UAV has not flown far enough and there is a risk of collision. The embedded computing unit sends instructions to the flight control system to adjust the altitude or distance to keep away from the bridge.

[0074] like Figure 5 As shown, the autonomous following path model of the inspection drone is:

[0075] D OA-OB =|OA-OB|#(19)

[0076] DAA′-BB′ =|AA′-BB′|#(20)

[0077]

[0078] Among them, OA is the distance from point A on the path to the drone; OB is the distance from point B on the path to the drone; AA' is the distance from point A to the left of the drone's direction of travel; BB' is the distance from point B to the right of the drone's direction of travel; D OA-OB is the distance difference between the drone and points A and B; D AA′-BB′ is the difference in distances that points A and B deviate to the left and right in the direction of the UAV’s travel; R is the distance between the wire and the UAV; ω is the angle along the direction of the UAV’s travel when the airborne laser radar scans the bridge; α is the angle perpendicular to the direction of the UAV’s travel when the airborne laser radar scans the wire; x is the distance that the point on the wire scanned by the laser radar deviates to the left and right in the direction of the UAV’s travel;

[0079] When D AA′-BB′ <-δ, the drone’s attitude is on the right side of the path and needs to fly left; when D AA′-BB′ >δ, the UAV is on the left side of the path and needs to move to the right; when -δ <D AA′-BB′ <δ, the drone’s attitude is aligned with the path above.

[0080] If the straight-line distance is greater than the safety distance threshold, it is judged that the drone's flight distance is too high and the data collection effect is poor. It is necessary to send instructions to the flight control unit through the embedded computing unit to adjust and lower the altitude to get closer to the path; if the straight-line distance is within the safety distance threshold, the drone's autonomous flight altitude is normal, and the drone will continue to follow the path to fly continuously, reach the target point, and proceed to the next stage of inspection.

[0081] The embedded computing unit on the inspection drone can respond to and process the data of the three-dimensional laser radar sensor in real time, control the flight of the inspection drone, and achieve path-following flight. The inspection drone can achieve autonomous flight through the flight control system, and send high-frequency data to the flight control system through the communication channel to achieve autonomous control flight of the inspection drone.

[0082] There is a possibility of encountering obstacles during the autonomous navigation of bridge inspection drones. Therefore, S6 also obtains the best safe flight path for inspection drones based on the autonomous obstacle avoidance technology of inspection drones and the coordinated autonomous obstacle avoidance technology of visual images and three-dimensional lidar, ensuring the obstacle avoidance effect of inspection drones during autonomous flight.

[0083] like Figure 6As shown, the S6 includes: S61, the inspection UAV scans the bridge environment in real time through the three-dimensional laser radar it carries, obtains three-dimensional spatial position information in real time, and synchronously feeds back to the edge computing unit; S62, the edge computing unit processes the feedback information online and calculates the distance between the inspection UAV and the obstacle; and obtains a real-time updated global point cloud through posture compensation, and uses a classifier constructed by a point cloud data classification algorithm to automatically identify line ground wires, poles and obstacles, and further calculates the relative position relationship between the inspection UAV and the inspection UAV; S63, generates a control amount based on the distance to control the inspection UAV to maintain a safe distance from the obstacle.

[0084] Visual images and radar point cloud data complement each other, and through the UAV flight control system, they assist the inspection UAV to avoid obstacles in complex inspection environments, thereby further improving its flight safety performance. Compared with the sparseness of radar scanning information and the instability of scanning effects on slender targets, images have the advantages of high information density and fine texture detail restoration. Specifically:

[0085] During the autonomous flight of the inspection drone, if there is an obstacle below the drone, such as a tree obstacle, whose vertical distance to the drone is less than the preset value, the drone needs to cross it. The drone uses the lidar and visible light sensor to collaboratively identify the obstacle, and notifies the drone to perform actions such as hovering, pulling up, horizontal crossing and descending based on the identification results. The timing process of the autonomous obstacle avoidance of the inspection drone with the fusion of visual images and lidar is as follows: Figure 7 As shown in the figure, (1) main: the main program entry of the method; (2) center: message processing object, receiving messages from different sensors and forwarding them; (3) camera: image data processing object, detecting obstacles through image segmentation technology and passing the detection results to the center; (4) lidar: lidar hidden danger point processing object, identifying obstacles from the hidden danger points returned by the lidar through the classification algorithm and passing the results to the center.

[0086] The visual image processing process includes: during the autonomous flight of the drone, the image sensor continuously collects visual images in front of it. When the lidar identifies an obstacle, it uses image recognition to confirm the type of obstacle, such as a bridge pier or a tree barrier. The drone rises to the top of the obstacle, stops rising and prepares to fly forward. The timing process is as follows: Figure 8 As shown, Camera: (1) init: initialization; (2) waiting: waiting for the center to notify to start working; (3) working: an obstacle is found, the center notifies to start image detection, the drone rises to a safe height, stops rising and prepares to fly forward, and then returns to the waiting state; (4) error: (1)(2)(3) error report.

[0087] When avoiding obstacles, the drone increases its flight altitude to avoid them, but the specific altitude required is unknown at the beginning. Therefore, the drone's embedded computing unit performs multiple steps of elevation and gradually analyzes the altitude and distance. If the distance to the obstacle is greater than the safe distance, the flight altitude needs to be lowered to ensure the inspection effect. At the same time, as the drone rises, the image of the obstacle gradually shrinks. The visual image obstacle avoidance timing process is as follows: Fig. 9 As shown in the figure, Center: (1) init: system initialization; (2) safe: the lidar did not detect any obstacles and no further processing is required; (3) confirm_20: the lidar detected obstacles such as steel beams between 30 and 20 meters away from the drone, which requires further confirmation through images; (4) confirm_15: the lidar detected obstacles such as steel beams between 20 and 15 meters away from the drone, which requires further confirmation through images; (5) confirm_10: the lidar detected obstacles such as steel beams between 15 and 10 meters away from the drone, which requires further confirmation through images; (6) ascend: the drone starts to rise, and the lidar and image are both in working condition; (7) top: the drone rises to the highest point and is ready to cross the obstacle; (8) reconfirm: before crossing the obstacle, the lidar may return to the potential danger point, and it is necessary to confirm whether there is an obstacle through images again; (9) h-fly: the drone flies horizontally; (10) descend: the drone descends; (11) error: system error.

[0088] As the drone ascends, obstacles such as bridge piers or tree barriers gradually move away from the drone, causing the size of the obstacles in the image to gradually decrease. Therefore, by adjusting the obstacle recognition algorithm to improve its recognition accuracy, this embodiment uses a target positioning algorithm that fuses visible light images and laser point cloud data to achieve autonomous obstacle avoidance for inspection drones based on the fusion of visual images and laser radars, and then generates obstacle avoidance strategies at different distances, so as to further improve the safety and reliability of autonomous flight of drones.

[0089] The method for generating the best safe flight path for the inspection drone is as follows: obtain the bridge safety point, that is, the center point (x, y) value of the pier, through the point cloud number query, load the point cloud data based on the octree structure calculation method, improve the loading rate, and determine the top of the pier, bridge deck, bridge foundation and channel parts through the point cloud elevation value and save them. Calculate the (x, y) value of the characteristic point at the bottom of the pier to obtain the horizontal distance from the target point to the pier. Obtain the slope of the bridge deck, calculate its radian value, and finally obtain the rotation angle. According to the middle point of the pier and the map characteristic point, obtain the height of the bridge, and calculate the new height value by the relative height (x, y) and the rotation angle. The expression is:

[0090]

[0091] According to the new height value, calculate the x, y values ​​of other photographed parts of the bridge, as well as the values ​​of the large and small side channels. For the side target points, determine the direction of the target points according to the crossarm direction, and sort them according to the elevation z value. Obtain the z value of the camera position above the bridge through the z value of the tower top and the height value from the bridge. And set the z value of the bridge safety point to this z value. According to the (x, y, z) value of the center point of the photographed target component in the bridge, obtain the relative rotation point, rotate the camera relative to the center point, and obtain the new camera posture parameters. The specific calculation method is:

[0092] First calculate the arc r that deviates from the horizontal direction:

[0093]

[0094] Calculate the distance d from the target point:

[0095] d=dcosr

[0096] Calculate the position of the shooting point:

[0097] Prot(x,y,z)=(x-dtanr,y+d,z)

[0098] Get as Fig.10 The inspection drone shooting location is shown.

[0099] The circle is the camera point, the rectangular area is the direction of the field of view plane corresponding to the camera, obtain the parameter value of the plane corresponding to the camera point, and calculate the value of the center point (x, y) of the plane:

[0100] Surface(x,y)=(x+d ​​sinr,yd cosr)

[0101] Compute the side length of a plane:

[0102] Side=0.01*DIS pp

[0103] Calculate the (x, y, z) values ​​of the four corner points of the plane:

[0104]

[0105]

[0106] Combined with the inspection drone camera parameter values, the camera positions of all shooting target points are calculated, and the safety and obstacle avoidance factors are considered to automatically generate the best safety inspection path for the drone, such as Fig.11 , Fig.12 , Fig.13 shown.

[0107] The above is only an embodiment of the present invention, and the common knowledge such as the known specific technical solutions and / or characteristics in the solution is not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several modifications and improvements can be made, which should also be regarded as the protection scope of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A combined navigation method for low-altitude complex environment in a bridge area, characterized in that: include: S1, synchronize and align multi-source sensor data; S2, integrates multi-source data for autonomous positioning of inspection drones; S3, data storage based on preset storage rules; S4, UAV flight path planning based on full coverage path planning algorithm; S5, after the UAV flight path planning is completed, the UAV starts to move and completes autonomous navigation based on the inspection UAV autonomous following path model; S6, based on the autonomous obstacle avoidance technology of the inspection drone and visual images, integrates the autonomous obstacle avoidance technology of the lidar to perform autonomous obstacle avoidance for the inspection drone.

2. According to claim 1, a combined navigation method for low-altitude complex environment in a bridge area is characterized in that: The S2 includes: S21, collect original data and conduct bridge inspection and survey to obtain inspection task planning; S22, acquiring three-dimensional digital terrain model data, orthophoto data, and three-dimensional spatial position data of bridges through multi-source sensors, and constructing a three-dimensional scene based on lidar point cloud data; S23, based on the coordination of laser point cloud, multi-eye visible light and deep multi-source data, locks the inspection drone path; S24 achieves autonomous positioning using a combined navigation and positioning algorithm based on global positioning, inertial navigation, real-time dynamic differential global positioning system, lidar odometer and visual odometer.

3. The method for combined navigation in a low-altitude complex environment in a bridge area according to claim 1 is characterized in that: The S5 includes: S51, during the flight of the inspection UAV executing path following, the 3D laser radar point cloud data is acquired in real time and posture compensation is performed to achieve real-time update of the global point cloud; S52, classify the point cloud data, combine the three-dimensional position relationship and flight control algorithm of the inspection drone, and control the inspection drone to follow the path line for autonomous navigation flight.

4. The method for combined navigation in a low-altitude complex environment in a bridge area according to claim 1 is characterized in that: The controlling the inspection drone to follow the path line for autonomous navigation flight also includes: inputting the relative position relationship and posture relationship between the reference objects into a PID control algorithm to obtain the control amount of each control channel of the inspection drone.

5. The method for integrated navigation in a low-altitude complex environment in a bridge area according to claim 1 is characterized in that: The autonomous following path model of the UAV is: D OA-OB =|OA-OB|#(19) D AA′-BB′ =|AA'-BB′|#(20) Among them, OA is the distance from point A on the path to the drone; OB is the distance from point B on the path to the drone; AA' is the distance from point A to the left of the drone's direction of travel; BB' is the distance from point B to the right of the drone's direction of travel; D OA-OB is the distance difference between the drone and points A and B; D AA′-BB′ is the difference in distances that points A and B deviate to the left and right in the direction of the UAV’s travel; R is the distance between the wire and the UAV; ω is the angle along the direction of the UAV’s travel when the airborne laser radar scans the bridge; α is the angle perpendicular to the direction of the UAV’s travel when the airborne laser radar scans the wire; x is the distance that the point on the wire scanned by the laser radar deviates to the left and right in the direction of the UAV’s travel; When D AA ′- BB ′<- δ When D AA ′- BB ′> δ When the drone is on the left side of the path, it needs to move to the right; when - δ <D AA ′- BB ′< δ When the drone flies above the path, its attitude is aligned.

6. The method for combined navigation in a low-altitude complex environment in a bridge area according to claim 1 is characterized in that: The S6 includes: S61, acquiring three-dimensional spatial position information in real time and synchronously feeding it back to the edge computing unit; S62, the edge computing unit calculates the distance between the inspection drone and the obstacle; S63, generating a control amount based on the distance to control the inspection drone to maintain a safe distance from the obstacle.

7. The method for integrated navigation in a low-altitude complex environment in a bridge area according to claim 1 is characterized in that: The process also includes S7, calculating the shooting point position, obtaining a new height value through the relative height and the rotation angle, obtaining a relative rotation point according to the (x, y, z) value of the center point of the shooting target component in the bridge, rotating the camera relative to the center point, and obtaining a new camera attitude parameter, the calculation method of which is: Get the radian r of the horizontal deviation: Get the distance d from the target point: d=dcosr Get the location of the shooting point: Prot(x,y,z)=(x-dtanr,y+d,z) Get the parameter value of the plane corresponding to the camera point and calculate the value of the center point of the plane: Surface(x,y)=(x+d ​​sinr,yd cosr) Get the side length of a plane: Side=0.01*DIS pp Get the (x, y, z) values ​​of the four corner points of the plane:

Images