A method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs)

Abstract

This application relates to the field of UAV inspection route generation technology, specifically to a method for automatically generating inspection routes for vehicle-mounted UAVs. The method includes: real-time acquisition of vehicle-related data before the vehicle arrives at the UAV takeoff point; calculation of real-time GPS / RTK positioning deviation; calculation of GPS / RTK positioning error variation in each acquisition cycle; calculation of GPS positioning reliability in each acquisition cycle; calculation of the center coordinates of the UAV takeoff and landing surface at the last acquisition moment of each acquisition cycle; loading digital elevation model data of the inspection area, defining a rectangular inspection area, inputting the target inspection height and performing a rationality check; and automatically generating a takeoff route, a serpentine inspection path, and a landing route based on the center coordinates of the UAV takeoff and landing surface at the last acquisition moment of the acquisition cycle in which the vehicle arrives at the UAV takeoff point. This application aims to improve the accuracy of vehicle-mounted UAV inspection route generation.

Description

Technical Field

[0001] This application relates to the field of drone inspection route generation technology, specifically to a method for automatically generating vehicle-mounted drone inspection routes. Background Technology

[0002] The vehicle-mounted drone inspection system consists of a vehicle-mounted hangar and drones. The hangar is an automated take-off, landing, and storage platform installed on the vehicle, normally housing and protecting the drones to prevent equipment damage. Inspection personnel can remotely send take-off commands via a ground control terminal, 4G / 5G network, or satellite communication link. The hangar will automatically open the hatch, release the drone, and initiate ejection upon receiving the remote command. After the mission is completed, the drone will be automatically recovered.

[0003] Currently, existing positioning methods for vehicle-mounted drone inspections mainly rely on vehicle-mounted GPS positioning combined with a fixed-altitude flight path generation algorithm. The vehicle's location is obtained via GPS, and the drone's take-off and landing points are determined based on the vehicle's GPS coordinates. A serpentine inspection route is then generated using a path planning algorithm. However, vehicle-mounted drone inspections typically take place in various environments, such as areas with multipath effects or dense vegetation, which weaken GPS signals. This results in significant positioning errors at the drone's take-off and landing points, and these positioning deviations can directly lead to flight path failures. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs) to solve the existing issues.

[0005] The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs) in this application adopts the following technical solution: One embodiment of this application provides a method for automatically generating inspection routes of vehicle-mounted unmanned aerial vehicles (UAVs), the method comprising the following steps: Step 1: Real-time data collection is performed using vehicle-mounted GPS, IMU, and RTK sensors to collect the vehicle's positioning coordinates, vehicle speed, vehicle acceleration, and hangar attitude angles for each collection cycle before the vehicle arrives at the UAV takeoff point, and the data is preprocessed. Step 2: Using real-time collected data, calculate the real-time GPS / RTK positioning deviation based on the difference between the actual displacement distance traveled by the vehicle per unit time and the GPS / RTK positioning distance; Step 3: Fit the positioning deviation within a collection cycle to a straight line in chronological order, and calculate the root mean square difference between the positioning deviation and the fitted value. Combine the slope of the fitted line with the root mean square difference to calculate the GPS / RTK positioning error variation for each collection cycle. Step 4: Based on the difference in positioning error variation between RTK and GPS in one acquisition cycle and the mean difference in positioning deviation at all acquisition times within the same acquisition cycle, calculate the GPS positioning reliability for each acquisition cycle. Step 5: Based on the GPS positioning reliability of a data acquisition cycle, perform weighted fusion of the GPS positioning coordinates and RTK positioning coordinates at the last data acquisition moment of the same data acquisition cycle to calculate the center coordinates of the UAV take-off and landing surface at the last data acquisition moment of each data acquisition cycle. Step 6: Load the digital elevation model data of the inspection area, define the rectangular inspection area, input the target inspection height, and perform a reasonableness check; Step 7: Based on the center coordinates of the UAV take-off and landing surface at the last data collection moment of the data collection cycle when the vehicle arrives at the UAV take-off point, automatically generate the take-off route, serpentine inspection path, and landing route.

[0006] Preferably, the positioning deviation is positively correlated with the difference and negatively correlated with the actual displacement distance.

[0007] Preferably, the variation in positioning error is positively correlated with the slope and root mean square error of the fitted straight line, respectively.

[0008] Preferably, the location reliability is positively correlated with the difference in the degree of variation of the location error and the difference in the mean.

[0009] Preferably, the method for calculating the location confidence is further as follows: The normalized value of the difference between the positioning error variation between RTK and GPS in the current acquisition period is used as the first confidence weight of GPS in the current acquisition period. The normalized value of the difference between the mean of the positioning deviation at all RTK acquisition times and the mean of the positioning deviation at all GPS acquisition times in the current acquisition period is used as the second confidence weight of GPS in the current acquisition period. Calculate the sum of the first confidence weight and the second confidence weight, and then calculate the product of the sum with the GPS positioning confidence of the previous acquisition period. The minimum value between the product result and the preset maximum value of GPS positioning reliability is taken as the GPS positioning reliability for the current acquisition period.

[0010] Preferably, the weights for weighting the GPS positioning coordinates and RTK positioning coordinates at the last acquisition time of the same acquisition cycle are: the GPS positioning reliability of the same acquisition cycle and 1 minus the GPS positioning reliability of the same acquisition cycle.

[0011] Preferably, the method for selecting the inspection area includes: preloading the digital elevation model data of the inspection area, clicking on the diagonal vertex of the rectangular inspection area on the vehicle map interface, and converting the screen coordinates into absolute coordinates in the WGS-84 coordinate system.

[0012] Preferably, the method for generating the takeoff route includes: taking the center coordinates of the UAV takeoff and landing surface at the last acquisition moment of the acquisition cycle when the vehicle arrives at the UAV takeoff point as the reference point, offsetting the takeoff point by a preset distance along the vehicle's heading angle, climbing in a straight line to the absolute height of the inspection, and then turning into a horizontal path pointing to the nearest vertex on the edge of the inspection area. The connection between the climbing section and the horizontal section is smoothly transitioned using the minimum turning radius of the UAV.

[0013] Preferably, the method for generating the serpentine inspection path includes: calculating the horizontal spacing between adjacent inspection paths based on the effective detection range of the UAV sensor and a preset overlap rate; generating a parallel path along the main direction starting from the vertex in the inspection area closest to the current UAV; and generating a serpentine full-coverage path by connecting the vertices with the minimum turning radius of the UAV. The main direction is the direction from the starting point to the point in the inspection area farthest from the starting point.

[0014] Preferably, the preprocessing involves filling in the missing data for each acquisition cycle using linear interpolation.

[0015] This application has at least the following beneficial effects: 1. This application calculates the difference between the GPS / RTK positioning distance and the actual vehicle displacement distance, which can accurately identify the intensity of GPS / RTK positioning drift and effectively reduce the impact of signal interference on the positioning of the take-off and landing points.

[0016] 2. This application calculates the variation of GPS / RTK positioning error based on the fusion calculation of the changing trend and fluctuation degree of positioning deviation, eliminates abnormal positioning data caused by sudden environmental changes, effectively characterizes the error degree of GPS / RTK positioning data, and improves the signal-to-noise ratio and stability of the calculation of the center coordinates of the take-off and landing surface of vehicle-mounted UAVs.

[0017] 3. This application constructs the center coordinates of the hangar UAV take-off and landing surface by adaptively and dynamically adjusting the GPS and RTK positioning coordinates through GPS positioning reliability, thereby improving the accuracy of the take-off point spatial position and thus improving the reliability and safety of the inspection route generation of vehicle-mounted UAVs in complex terrain. Attached Figure Description

[0018] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating an automatic generation method for vehicle-mounted unmanned aerial vehicle (UAV) inspection routes provided in this application. Detailed Implementation

[0020] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs) according to this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0022] The following description, in conjunction with the accompanying drawings, details a specific scheme for an automatic generation method of vehicle-mounted unmanned aerial vehicle (UAV) inspection routes provided in this application.

[0023] One embodiment of this application provides a method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs).

[0024] Specifically, the following method for automatically generating inspection routes for vehicle-mounted drones is provided; please refer to [link / reference]. Figure 1 The method includes the following steps: Step 1: Real-time data collection is performed using onboard GPS, IMU, and RTK sensors to collect the vehicle's positioning coordinates, vehicle speed, vehicle acceleration, and hangar attitude angles for each collection cycle before the vehicle arrives at the UAV takeoff point, and the data is preprocessed.

[0025] First, vehicle-mounted GPS, IMU (Inertial Measurement Unit), and RTK sensors are installed on the vehicle. During vehicle-mounted drone inspections, the vehicle transports the drone to the takeoff point. During the transport of the drone to the hangar, the vehicle-mounted GPS and RTK sensors collect the GPS and RTK positioning coordinates of the drone in real time. The collected GPS and RTK positioning coordinates are then uniformly converted to a local planar projection coordinate system (such as the Northeast ENU coordinate system) using a map projection algorithm, with the coordinate unit uniformly set to meters.

[0026] The IMU collects the vehicle's speed and acceleration during operation, and measures the attitude angle data of the hangar (the equipment that houses the drone), including pitch angle. Roll angle and along the vehicle heading angle .

[0027] In this embodiment, all data before the vehicle arrives at the drone takeoff point is collected, with a collection cycle of 1 minute and a data collection frequency of 10Hz. The data collected in each collection cycle are arranged in chronological order of collection time, and missing data is filled in by linear interpolation to obtain the GPS positioning sequence, RTK positioning sequence, vehicle running speed sequence, and vehicle running acceleration sequence for each collection cycle.

[0028] It should be noted that all elements in each sequence are three-dimensional data, corresponding to the horizontal, vertical, and height dimensions. The latitude and longitude directions are used as the horizontal and vertical axes, respectively, with the direction perpendicular to the ground upwards as the positive half-axis of the height coordinate. The linear interpolation method used is a well-known technique, and the specific calculation steps will not be elaborated here.

[0029] Step 2: Using real-time collected data, calculate the real-time GPS / RTK positioning deviation based on the difference between the actual displacement distance traveled by the vehicle per unit time and the GPS / RTK positioning distance.

[0030] During the vehicle's journey to the drone's takeoff point, its speed is not constant, causing variations in velocity and acceleration across different dimensions. This results in the drone traveling within the vehicle traveling different distances within different units of time. Because GPS and RTK positioning require the transmission of positioning signals, and these signals are susceptible to interference during transmission, errors occur between the vehicle's actual distance at each of its two endpoints within a unit of time and the distance traveled by the vehicle.

[0031] The greater the difference between the two distances mentioned above, the lower the accuracy of the positioning signal. Therefore, for a vehicle per unit time (the reciprocal of the data acquisition frequency), the accuracy is... The distance traveled within s) is calculated by taking the vehicle's speed and acceleration in different dimensions as inputs to the displacement formula, and outputting the distance traveled in each dimension. These distances are then grouped into triplets in the order of horizontal, vertical, and vertical, and arranged chronologically to obtain the actual displacement sequence of the vehicle, which characterizes the actual distance traveled by the vehicle in different units of time. The calculation of the displacement formula is a well-known technique, and the specific calculation steps will not be elaborated here.

[0032] During vehicle movement, the vehicle's displacement coordinates are constantly changing, resulting in different positioning coordinates at different data collection times and thus different positioning distances. Within a unit of time, the greater the difference between the vehicle's GPS / RTK positioning distance and the actual displacement distance, the greater the potential positioning error, and consequently, the lower the reliability of the GPS / RTK positioning.

[0033] Therefore, this application calculates the real-time GPS / RTK positioning deviation based on the difference between the actual displacement distance traveled by the vehicle per unit time and the GPS / RTK positioning distance, to characterize the degree of error in the positioning technology. The positioning deviation is positively correlated with the difference and negatively correlated with the actual displacement distance. It is understood that a positive correlation means the dependent variable increases as the independent variable increases, and decreases as the independent variable decreases; a negative correlation means the dependent variable decreases as the independent variable increases, and increases as the independent variable decreases. This is determined by the actual application, and this application does not impose any special limitations.

[0034] Specifically, in this embodiment, the positioning deviation of GPS at the t-th data collection time is used. For example: In the formula, , These represent the GPS positioning coordinates at the (t+1)th and tth data acquisition times, respectively. This represents the Euclidean distance function, used to calculate the topological distance between the vehicle's location coordinates at two acquisition times, in meters. This represents the GPS positioning distance at the t-th unit of time; This represents the actual distance traveled by the vehicle in the t-th unit of time. Indicates the absolute value sign; This represents a preset constant parameter, a distance compensation constant to prevent the denominator from being zero. The unit is meters, and the default value range is [0.001, 1]. In this embodiment, the value is 0.01.

[0035] In existing technologies, the accuracy of GPS positioning can be characterized by the difference between positioning distances from multiple satellites. In this embodiment, by combining the vehicle's actual operating speed and acceleration, the actual displacement distance of the vehicle per unit time is calculated. The difference between the GPS positioning distance and the vehicle's actual displacement distance can effectively identify positioning drift caused by satellite signal multipath effects or atmospheric delay, characterizing the positioning reliability at each time point. This is used to determine the interference intensity of GPS signals on the vehicle's position perception in complex environments. A larger positioning deviation indicates a greater intensity of interference to the GPS, and a higher likelihood that the vehicle is entering a complex environment.

[0036] Similarly, the positioning deviation of RTK at the t-th acquisition time and The calculation method is the same. By analyzing the difference between the RTK positioning distance and the vehicle's actual displacement distance, the interference intensity of the RTK signal on the vehicle's position perception in complex environments is determined. The greater the positioning deviation, the greater the interference intensity of the RTK, and the more likely the vehicle is to enter a complex environment.

[0037] Step 3: Fit the positioning deviation within a collection cycle to a straight line in chronological order, and calculate the root mean square difference between the positioning deviation and the fitted value. Combine the slope of the fitted line with the root mean square difference to calculate the GPS / RTK positioning error variation for each collection cycle.

[0038] When vehicles enter a city from open areas, the multipath effect is significantly enhanced, and building surfaces strongly refract and reflect GPS / RTK signals. Simultaneously, dense clusters of tall buildings block GPS / RTK signals, causing a continuous decrease in GPS / RTK positioning signal quality and a corresponding increase in positioning error, thus reducing signal reliability. The stronger the increasing trend of positioning error, the faster the rate of decrease in GPS / RTK positioning accuracy. When vehicles enter areas with tall buildings or fewer satellites, the reliability of GPS / RTK signals should also decrease when locating the vehicle.

[0039] Accordingly, this application arranges the GPS / RTK positioning deviations at different acquisition times within an acquisition cycle in chronological order to obtain the GPS / RTK positioning deviation sequence for each acquisition cycle.

[0040] This embodiment uses the GPS positioning deviation sequence of one acquisition period as an example. The GPS positioning deviation sequence is used as input to the linear least squares method (Note: To ensure the meaning of subsequent calculation results, the abscissa of the positioning deviation sequence is normalized before linear fitting). The algorithm outputs the slope of the fitted line of the GPS positioning deviation sequence and the fitted value of each element in the sequence. The slope of the fitted line characterizes the error trend during GPS positioning; a positive value indicates that the positioning deviation is increasing, and the vehicle is entering a high-rise area; a negative value indicates that the vehicle is entering an open area. The root mean square difference between all elements in the positioning deviation sequence and their corresponding fitted values ​​is calculated to characterize the change in GPS positioning accuracy; a larger value indicates a greater change in positioning accuracy. The linear least squares method and the calculation of the root mean square difference are both well-known techniques, and the specific calculations will not be elaborated here.

[0041] Similarly, using the methods described above for calculating the slope and root mean square error of the fitted line of the GPS positioning deviation sequence, a straight line can also be fitted to the RTK positioning deviation sequence to calculate the slope and root mean square error of the fitted line of the RTK positioning deviation sequence.

[0042] Furthermore, this application combines the slope and root mean square error of the fitted straight line to calculate the variability of GPS / RTK positioning error for each acquisition cycle, which characterizes the intensity of the change in positioning error as the vehicle travels. The variability of positioning error is positively correlated with both the slope and root mean square error of the fitted straight line.

[0043] Specifically, this embodiment uses the GPS positioning error variation rate during the current data acquisition period. For example: In the formula, This indicates the degree of variation in GPS positioning error during the current data collection period; This represents the slope of the linear fit to the GPS positioning deviation sequence during the current acquisition period. This represents the root mean square difference between all elements in the GPS positioning deviation sequence during the current acquisition period and their corresponding fitted values. This represents the sigmoid normalization function, and its purpose is to prevent the slope from being negative, which would reverse the meaning of the result.

[0044] In existing technologies, the variance and standard deviation of data can be used to characterize the fluctuation state of data. In this embodiment, the changing trend and fluctuation of the positioning deviation sequence characterize the change in the intensity of GPS positioning error. The larger the fitting slope of the positioning deviation sequence, the lower the positioning accuracy becomes when the vehicle travels to areas with tall buildings. The larger the root mean square difference between the positioning deviation sequence and the fitted value, the greater the noise interference. The product of the fitting slope and the root mean square difference characterizes the intensity of systematic positioning drift and random noise superposition caused by the vehicle in different areas. The larger the value, the less reliable the GPS positioning data.

[0045] Similarly, based on the current data collection period, the GPS positioning error variation... The calculation method can be used to calculate the positioning error variation of RTK in the current acquisition period. The specific implementation process will not be elaborated here.

[0046] Step 4: Based on the difference in positioning error variation between RTK and GPS in one acquisition cycle and the mean difference in positioning deviation at all acquisition times within the same acquisition cycle, calculate the GPS positioning reliability for each acquisition cycle.

[0047] When fusing GPS and RTK positioning to accurately determine the location of a vehicle-mounted drone, the degree of variation in GPS and RTK positioning errors determines the level of trust in the GPS and RTK coordinates. When the variation in GPS positioning error is greater than that of RTK, the RTK coordinates should be trusted more. Conversely, when the GPS positioning deviation is greater than that of RTK, the reliability of the GPS coordinates is lower.

[0048] Therefore, this application calculates the GPS positioning reliability for each acquisition cycle based on the difference in positioning error variation between RTK and GPS within an acquisition cycle and the mean difference in positioning deviation at all acquisition times within the same acquisition cycle. This reliability characterizes the credibility of GPS positioning coordinates; a larger value indicates higher credibility, while a smaller value indicates lower credibility. The positioning reliability is positively correlated with both the difference in positioning error variation and the mean difference.

[0049] Specifically, this embodiment uses the GPS positioning reliability of the current acquisition period. For example: In the formula, This represents the GPS positioning reliability during the current data acquisition period; min is a minimum value function used to truncate data and prevent data overflow. This indicates the GPS positioning reliability in the previous data collection period; This represents the maximum preset GPS positioning reliability value, with a default range of [0.8, 1]. In this embodiment, the value is 1, and because... and The values ​​of all values ​​are greater than 0, that is... The value range is (0,1). Note: In this embodiment, the GPS positioning confidence level is 0.5 in the initial acquisition period, which is used to represent the same level of confidence in GPS and RTK; in the first acquisition period after startup, the initial value of the GPS positioning confidence level in the previous acquisition period is set to 0.5.

[0050] in, This indicates the first confidence weight of GPS data for the current data collection period; The sigmoid normalization function is used to normalize data to the range (0,1). The sigmoid normalization function is a well-known technique and will not be elaborated further. and The range of values ​​for is (0,1); This indicates the degree of variation in RTK positioning error during the current acquisition period; This indicates the degree of variation in GPS positioning error during the current data collection period; This represents the second confidence weight of GPS during the current data collection period; This represents the average positioning deviation across all RTK acquisition moments in the current acquisition period. This represents the average positioning deviation of GPS at all acquisition times during the current acquisition period.

[0051] In existing multi-source fusion positioning technologies, fusion weights are typically assigned based on the static errors and variances of each sensor. In this embodiment, instead of solely relying on the absolute error magnitude at a particular acquisition moment, the "strength of the trend in positioning error" and the "historical average deviation level" are comprehensively considered. This dual-dimensional assessment of the relative reliability of GPS and RTK in complex environments allows for a comprehensive evaluation. By adjusting the sum of the first and second confidence weights to ensure the precise and adaptive dynamic adjustment of GPS positioning reliability, the vehicle-mounted UAV can seamlessly switch to a high-precision RTK-based positioning mode when GPS signals are interfered with.

[0052] Step 5: Based on the GPS positioning reliability of a data acquisition cycle, perform weighted fusion of the GPS positioning coordinates and RTK positioning coordinates at the last data acquisition moment of the same data acquisition cycle, and calculate the center coordinates of the UAV take-off and landing surface at the last data acquisition moment of each data acquisition cycle.

[0053] The weights for weighting the GPS positioning coordinates and RTK positioning coordinates at the last acquisition time of the same acquisition cycle are: the GPS positioning reliability of the same acquisition cycle and 1 minus the GPS positioning reliability of the same acquisition cycle.

[0054] Specifically, in this embodiment, the GPS positioning coordinates and RTK positioning coordinates at the last acquisition moment of the current acquisition period are weighted and fused based on the GPS positioning reliability of the current acquisition period to calculate the center coordinates of the UAV take-off and landing surface at the last acquisition moment of the current acquisition period: In the formula, , , These represent the x-axis, y-axis, and altitude coordinates of the center of the UAV's take-off and landing surface at the last data acquisition moment of the current data acquisition cycle. , These are the x-axis and y-axis coordinates of the GPS positioning at the last acquisition time of the current acquisition cycle, respectively. , These are the x-axis and y-axis coordinates of the RTK positioning at the last acquisition moment of the current acquisition cycle. It is the height axis coordinate value of the RTK positioning coordinate at the last acquisition moment of the current acquisition cycle.

[0055] Thus, the coordinates of the center of the UAV take-off and landing surface at the last acquisition moment of the current acquisition cycle are obtained. Similarly, the coordinates of the center of the UAV take-off and landing surface at the last acquisition moment of each acquisition cycle can be obtained.

[0056] Step 6: Load the digital elevation model data of the inspection area, define the rectangular inspection area, input the target inspection height, and perform a reasonableness check.

[0057] Inspection Area Selection: The digital elevation model (DEM) data of the inspection area is pre-loaded to extract the actual terrain elevation corresponding to the two-dimensional coordinate points. Then, the operator clicks twice on the vehicle's map interface to define the diagonal vertices A (screen coordinates (x1, y1)) and B (screen coordinates (x2, y2)) of the rectangular inspection area. Through a map coordinate mapping algorithm, the screen coordinates are converted to absolute coordinates A (X_A, Y_A) and B (X_B, Y_B) in the WGS-84 coordinate system, automatically generating the four vertices of the rectangular area (A, B, C(X_A, Y_B), D(X_B, Y_A)). The calculation of the map coordinate mapping algorithm is a well-known technique, and the specific calculation steps will not be elaborated here.

[0058] Inspection height input and verification: S1: Manual input: The vehicle interface provides a height input box, where the operator inputs the target inspection height H_target (relative to the ground). S2: Absolute Height Calculation: The algorithm automatically reads RTK elevation data and calculates the inspection absolute height H_abs = H_target - ; S3: Reasonableness check: Built-in height safety threshold (minimum 5m, maximum 120m). If the input H_target exceeds the threshold, a prompt will pop up and suggest a reasonable range. It will take effect after confirmation.

[0059] Step 7: Based on the center coordinates of the UAV take-off and landing surface at the last data collection moment of the data collection cycle when the vehicle arrives at the UAV take-off point, automatically generate the take-off route, serpentine inspection path, and landing route.

[0060] Departure route generation: The coordinates of the center of the UAV takeoff and landing surface at the last data acquisition moment of the data acquisition cycle in which the vehicle arrives at the UAV takeoff point. Using point S as the reference point, along the vehicle's heading angle Offset 8m (to avoid the fuselage obstructing the RTK signal) to determine the takeoff point ,in, , The calculation method is as follows: Takeoff path planning: Climb in a straight line from point S' at a rate of 0.5 m / s. After reaching the target inspection altitude H_abs, switch to a horizontal path and point to the nearest vertex P1 (X_P1, Y_P1, H_abs) at the edge of the inspection area. The connection between the climb and horizontal sections uses the minimum turning radius r of the UAV for a smooth transition, where r is determined by the UAV's factory settings.

[0061] Additionally, it should be noted that if the last data collection moment of the data collection period in which the vehicle arrives at the drone takeoff point is not collected, meaning that the data collection period in which the vehicle arrives at the drone takeoff point is not a complete data collection period (1 minute), the average data within the data collection period can be used to complete the data collection.

[0062] Snake-shaped inspection path generation: Based on the effective detection range R of the UAV sensor and the preset overlap rate K (default 30%), the horizontal spacing d between adjacent inspection paths is calculated as d = R × (1 - K). In this embodiment, a parallel path is generated along the main direction (where the main direction is the direction from the starting point to the point farthest from the starting point in the inspection area) starting from the vertex closest to the current UAV in the inspection area. All paths have a uniform flight altitude of H_abs. Adjacent paths are connected by arcs with a minimum turning radius r to form a serpentine full-coverage path. The algorithm automatically reads RTK elevation data and corrects the height of path nodes: if the terrain elevation of a node is higher than H_abs-5m, the height of the node is automatically raised to the terrain elevation +5m to avoid collisions. The method for generating the serpentine inspection path is a well-known technology, and its specific generation process will not be described in detail.

[0063] Landing path generation: Landing point setting: Based on the center coordinates of the UAV's takeoff and landing surface For landing point E, ensure that the landing path does not intersect with the takeoff path; Landing Path Planning: From the final point Pn (X_Pn, Y_Pn, H_abs) of the serpentine inspection area, fly in a straight line to the transition point E1 (X_E1, Y_E1, H_abs) 10m above the hangar. +10), then descend in a straight line to point E at a descent rate of 0.3 m / s, avoiding the takeoff path and surrounding obstacles during the descent.

[0064] The above technical features constitute the preferred embodiment of this application, which has strong adaptability and the best implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

Claims

1. A method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles (UAVs), characterized in that, The method includes the following steps: Step 1: Real-time data collection is performed using vehicle-mounted GPS, IMU, and RTK sensors to collect the vehicle's positioning coordinates, vehicle speed, vehicle acceleration, and hangar attitude angles for each collection cycle before the vehicle arrives at the UAV takeoff point, and the data is preprocessed. Step 2: Using real-time collected data, calculate the real-time GPS / RTK positioning deviation based on the difference between the actual displacement distance traveled by the vehicle per unit time and the GPS / RTK positioning distance; Step 3: Fit the positioning deviation within a collection cycle to a straight line in chronological order, and calculate the root mean square difference between the positioning deviation and the fitted value. Combine the slope of the fitted line with the root mean square difference to calculate the GPS / RTK positioning error variation for each collection cycle. Step 4: Based on the difference in positioning error variation between RTK and GPS in one acquisition cycle and the mean difference in positioning deviation at all acquisition times within the same acquisition cycle, calculate the GPS positioning reliability for each acquisition cycle. Step 5: Based on the GPS positioning reliability of a data acquisition cycle, perform weighted fusion of the GPS positioning coordinates and RTK positioning coordinates at the last data acquisition moment of the same data acquisition cycle to calculate the center coordinates of the UAV take-off and landing surface at the last data acquisition moment of each data acquisition cycle. Step 6: Load the digital elevation model data of the inspection area, define the rectangular inspection area, input the target inspection height, and perform a reasonableness check; Step 7: Based on the center coordinates of the UAV take-off and landing surface at the last data collection moment of the data collection cycle when the vehicle arrives at the UAV take-off point, automatically generate the take-off route, serpentine inspection path, and landing route.

2. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 1, characterized in that, The positioning deviation is positively correlated with the difference and negatively correlated with the actual displacement distance.

3. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 1, characterized in that, The variation in positioning error is positively correlated with the slope and root mean square error of the fitted straight line, respectively.

4. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 1, characterized in that, The location reliability is positively correlated with the difference in the location error variation and the difference in the mean, respectively.

5. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 4, characterized in that, The method for calculating the location reliability is further as follows: The normalized value of the difference between the positioning error variation between RTK and GPS in the current acquisition period is used as the first confidence weight of GPS in the current acquisition period. The normalized value of the difference between the mean of the positioning deviation at all RTK acquisition times and the mean of the positioning deviation at all GPS acquisition times in the current acquisition period is used as the second confidence weight of GPS in the current acquisition period. Calculate the sum of the first confidence weight and the second confidence weight, and then calculate the product of the sum with the GPS positioning confidence of the previous acquisition period. The minimum value between the product result and the preset maximum value of GPS positioning reliability is taken as the GPS positioning reliability for the current acquisition period.

6. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 1, characterized in that, The weights for weighting the GPS positioning coordinates and RTK positioning coordinates at the last acquisition time of the same acquisition cycle are: the GPS positioning reliability of the same acquisition cycle and 1 minus the GPS positioning reliability of the same acquisition cycle.

7. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 1, characterized in that, The method for selecting the inspection area includes: preloading the digital elevation model data of the inspection area, clicking on the diagonal vertex of the rectangular inspection area on the vehicle map interface, and converting the screen coordinates into absolute coordinates in the WGS-84 coordinate system.

8. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 7, characterized in that, The method for generating the takeoff route includes: taking the center coordinates of the UAV takeoff and landing surface at the last acquisition moment of the acquisition cycle when the vehicle arrives at the UAV takeoff point as the reference point, offsetting the takeoff point by a preset distance along the vehicle's heading angle, climbing in a straight line to the absolute height of the inspection, and then turning into a horizontal path pointing to the nearest vertex on the edge of the inspection area. The connection between the climbing segment and the horizontal segment is smoothly transitioned using the minimum turning radius of the UAV.

9. The method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in claim 8, characterized in that, The method for generating the serpentine inspection path includes: calculating the horizontal spacing between adjacent inspection paths based on the effective detection range of the UAV sensor and a preset overlap rate; generating parallel paths along the main direction starting from the vertex closest to the current UAV in the inspection area; and generating a serpentine full-coverage path by connecting the vertices with the minimum turning radius of the UAV. The main direction is the direction from the starting point to the point farthest from the starting point in the inspection area.

10. A method for automatically generating inspection routes for vehicle-mounted unmanned aerial vehicles as described in any one of claims 1-9, characterized in that, The preprocessing involves filling in the missing data for each acquisition cycle using linear interpolation.

Images