A method, system, electronic device and storage medium for intelligent self-testing of a compound-wing unmanned aerial vehicle before takeoff.

Abstract

The present invention discloses an intelligent self-check method, system, electronic device and storage medium for a compound-wing unmanned aerial vehicle before takeoff, relating to the technical field of unmanned aerial vehicles. The intelligent self-check method for the compound-wing unmanned aerial vehicle before takeoff comprises the following steps: S10, collecting in real time an appearance video of the target unmanned aerial vehicle fixed at a preset position, and preprocessing each frame of image of the appearance video; S20, respectively performing state detection on the propeller, control surface and folded wing of the target unmanned aerial vehicle in the preprocessed image by using a propeller detection channel, a control surface detection channel and a folded wing detection channel; S30, mapping the state detection result in S20 into a state vector with a fixed length, comparing the state vector with a pre-stored standard state logic table, determining that the self-check is qualified and outputting a takeoff permission instruction if all are consistent, otherwise determining that the self-check is abnormal and outputting a takeoff prohibition instruction. The beneficial effects of the present invention: the operation safety and operation and maintenance efficiency in the unattended scenario are improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and more specifically, to an intelligent self-test method, system, electronic device, and storage medium for a compound-wing UAV before takeoff. Background Technology With the widespread application of drone technology in industrial inspection, logistics distribution, and emergency rescue, unmanned hangars, as an integrated infrastructure capable of automating drone storage, charging / battery swapping, scheduling, and recovery, are gradually becoming the mainstream solution for large-scale deployment in the industry. In unmanned mode, the integrity and functional correctness of the drone's mechanical structure and key components after it leaves the hangar and before it executes a flight mission are the primary prerequisites for flight safety. Any undetected mechanical failure, such as damaged propeller blades, jammed control surfaces, or unreliable locking of folding wings, can quickly lead to loss of control of the aircraft after takeoff, causing serious safety accidents and economic losses.

[0002] Currently, the status confirmation of drones before takeoff mainly relies on the following two methods: 1. Manual Visual Inspection: This is the most traditional method, where maintenance personnel inspect each drone on-site before takeoff for damage to the propeller blades, smooth rotation, proper deflection of the control surfaces, and whether the folding wings are fully deployed and locked in place. However, this method is highly dependent on the experience and responsibility of personnel, is time-consuming and labor-intensive, highly subjective, and cannot meet the needs of all-weather, high-frequency unattended scenarios. In the actual operation of unattended hangars, manual on-site inspection is fundamentally contradictory to the hangar's automation and unmanned concept, making this method difficult to implement.

[0003] 2. Airborne Sensor Self-Check: Some existing unmanned hangars have introduced sensor self-check functions for their flight control systems. For example, after the system is powered on, the ESC (Electronic Speed ​​Controller) sends a feedback signal to confirm whether the motors are responding normally, and the servo motors send a feedback signal to confirm whether the control surfaces can move as instructed. While this method can verify the continuity of electrical links and the effectiveness of electronic signal transmission, it is essentially a detection of "electrical signals," not a direct verification of "mechanical motion status." Therefore, there are significant blind spots. For example, even if the propeller is incorrectly installed in reverse (causing adjacent motors to rotate in the same direction), the ESC feedback may still show that the motors are operating normally; even if there is jamming in the control surface transmission mechanism, the servo motor feedback may show that the control surfaces have reached the designated position; even if the locking mechanism of the folding wing is not fully locked, the limit switch may give an incorrect locking signal. These critical mechanical assembly and motion status anomalies may appear completely normal at the sensor data level, but after takeoff, they can directly lead to catastrophic consequences.

[0004] To this end, the present invention provides an intelligent self-inspection method, system, electronic device and storage medium for pre-takeoff of a compound-wing UAV, which can realize unmanned pre-takeoff safety quality inspection of the UAV throughout the entire process, and can detect the operating status of key moving parts of the UAV, thereby improving the operational safety and maintenance efficiency in unattended scenarios. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, this invention provides an intelligent self-inspection method, system, electronic device, and storage medium for pre-flight checks of a compound-wing unmanned aerial vehicle (UAV). This method enables unmanned pre-flight safety quality inspection of the UAV throughout the entire process and can detect the operating status of key moving parts of the UAV, thereby improving operational safety and maintenance efficiency in unattended scenarios.

[0006] The technical solution adopted by this invention to solve its technical problem is: an intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle (UAV), the improvement of which is that the intelligent self-test method for pre-takeoff of the compound-wing UAV includes the following steps: S10: Real-time acquisition of the appearance video of the target drone fixed at a preset position, and preprocessing of each frame of the appearance video. S20 uses the propeller detection channel, control surface detection channel, and folding wing detection channel to perform state detection on the propeller, control surface, and folding wing of the target UAV in the preprocessed image, respectively. S30 maps the state detection result in S20 into a fixed-length state vector and compares the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is deemed qualified and a takeoff permission command is output; otherwise, the self-test is deemed abnormal and a takeoff prohibition command is output.

[0007] Furthermore, in step S10, the specific method for preprocessing each frame of the appearance video is as follows: sequentially performing grayscale processing, contrast-limited adaptive histogram equalization processing, brightness compensation, and Gaussian filtering processing on each frame of the acquired appearance video.

[0008] Furthermore, after the image preprocessing is completed, a preset ROI coordinate configuration file needs to be loaded to mark the initial positions of each propeller area, each control surface area, and each folding wing area of ​​the target UAV.

[0009] Furthermore, in step S20, the state detection of the propeller of the target UAV in the preprocessed image using the propeller detection channel specifically includes propeller blade integrity detection and propeller blade rotation state, direction, and speed consistency detection. The specific method for propeller blade integrity detection is as follows: In the static state before the propeller motors of the target drone start, extract a frame of the current moment from the preprocessed image stream and crop out each propeller sub-image according to the ROI coordinates. Each propeller sub-image is binarized, and the contours of the binarized sub-images are extracted. The geometric features of the profile are calculated, and the integrity of the blade is double-verified by multi-blade lateral comparison and standard reference comparison. If any verification triggers an anomaly, the blade integrity is determined to be abnormal. The specific method for detecting the consistency of propeller blade rotation state, direction, and speed is as follows: Start the propeller motor and perform feature point extraction and optical flow tracking on the frame sequence continuously acquired after the propeller motor starts; The tangential velocity components of each feature point are calculated based on the optical flow tracking results, and the direction score is accumulated. The instantaneous rotation direction of adjacent frame pairs is determined by whether the dominance ratio between the accumulated direction scores exceeds the preset direction determination dominance ratio threshold. The instantaneous rotation direction of multiple adjacent frame pairs is comprehensively determined by multi-frame analysis. The category with the highest frequency is taken as the final direction code of the propeller. If the highest frequency category is no effective rotation or the frequencies of two effective rotation directions are tied, the propeller is determined not to be rotating normally. For a normally rotating propeller, the inter-frame angular velocity is calculated based on the tangential velocity components of the feature points and their distances to the propeller center, and then converted into rotational speeds per minute (RPM). The RPM of all propellers is checked for consistency. If the RPM of a propeller deviates from the median by more than the preset dynamic deviation threshold or is lower than the speed limit, the propeller speed is determined to be abnormal. The detection status code of each blade is output based on the detection results of the blade's rotation status, direction, and speed.

[0010] Furthermore, the specific method for performing state detection of the target UAV's control surfaces in the preprocessed image using the control surface detection channel is as follows: Before the flight controller issues the control surface deflection command, the current frame is extracted from the preprocessed image stream as the reference frame, and Canny edge detection is performed in the corresponding control surface ROI area to extract the control surface trailing edge feature point set. The RANSAC algorithm is used to fit a straight line to the trailing edge feature point set, and the angle between the fitted line and the pre-calibrated reference line for the zero deflection position of the control surface is calculated. ; After the flight controller issues the control surface deflection command, the action frame is extracted and the same edge detection and straight line fitting operations as the reference frame are performed on the action frame to obtain the angle between the fitted straight line after the action and the zero deflection position reference line. ; Through formula calculate and deflection between The measured deflection direction is determined based on the sign of the measured deflection amount Δθ. The measured deflection direction is then compared with the deflection direction expected by the flight control command. If the directions are opposite, it is directly judged as an abnormal direction; if the direction is correct, the amplitude is further determined. If the value is within 20% of the expected value, the amplitude verification passes; otherwise, the amplitude is deemed abnormal. Perform the above procedure on the aileron, elevator, and rudder respectively, and output the status code of each control surface.

[0011] Furthermore, the specific method for using the folded wing detection channel to perform state detection on the folded wings of the target UAV in the preprocessed image is as follows: After the flight controller issues the wing deployment command, the current frame is extracted from the preprocessed image stream, and edge detection and line fitting are performed within the preset wing ROI region to obtain several candidate line segments. The optimal line segment among the candidate line segments obtained from the fitting is selected and used as the main edge of the wing. The angle between the selected wing principal edge and the fuselage axis is taken as the wing deployment angle. ,like If the wing deployment angle is greater than or equal to the preset wing deployment angle threshold, the wing deployment is considered normal. If the wing folding angle is less than the preset threshold, it is determined that the wings are not unfolded. If the wing angle is between the wing deployment angle threshold and the wing folding angle threshold, the wing is determined to be in a semi-deployed and stuck state. Within the preset small ROI of the wing root locking mechanism, normalized correlation coefficient template matching is used to compare the current frame with the calibrated locking standard template. If the best matching correlation coefficient is greater than or equal to the preset matching threshold, the locking is determined to be in place; otherwise, the locking is determined to be in place. After confirming that the wings are deployed and locked in place, multiple frames of images are continuously acquired. For the micro-motion detection ROI in the far end area of ​​the wings, the inter-frame difference map is calculated frame by frame. The proportion of non-zero pixels in the difference map is counted. If the proportion of frames exceeding the preset proportion exceeds the preset micro-motion threshold, the locking is determined to be unsecured; otherwise, the locking is determined to be reliable. Based on the results of wing deployment angle detection, lock confirmation, and micro-motion detection, the detection status code for each wing is output.

[0012] Furthermore, an anomaly re-detection mechanism is set up during the state detection process of the folding wing. The anomaly re-detection mechanism is as follows: If the wing deployment angle detection determines that the wing is in a semi-deployed and stuck state, the wing retraction command will be automatically issued. After a delay, the deployment command will be issued again and the deployment angle detection will be re-executed. If the wing is still determined to be semi-deployed and stuck after two consecutive re-deployments, a hardware fault will be confirmed and an alarm will be issued. If the micro-motion detection determines that the locking is not secure, the micro-motion detection process will be automatically re-executed after a preset time. If the re-detection still determines that the locking is not secure, it will be confirmed as a hardware failure and an alarm will be triggered.

[0013] An intelligent self-check system for pre-takeoff operation of a compound-wing unmanned aerial vehicle (UAV) is improved by including: The image acquisition unit is used to acquire real-time video of the appearance of the target drone fixed at a preset position. An image preprocessing unit is used to preprocess each frame of the appearance video; The propeller detection channel is used to detect the state of the propellers of the target UAV in the preprocessed image; The control surface detection channel is used to detect the state of the control surfaces of the target UAV in the preprocessed image; The folded wing detection channel is used to detect the state of the folded wings of the target UAV in the preprocessed image; The integrated judgment unit is used to map the state detection results of the propeller detection channel, control surface detection channel and folding wing detection channel into a fixed-length state vector, and compare the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is judged to be qualified and a takeoff permission command is output; otherwise, the self-test is judged to be abnormal and a takeoff prohibition command is output.

[0014] An electronic device, improved in that it includes at least one processor and at least one memory, wherein, The memory stores computer-readable instructions; The computer-readable instructions are executed by one or more processors, enabling the electronic device to implement the intelligent self-test method for pre-takeoff of the compound-wing UAV as described above.

[0015] An improvement of a storage medium storing computer-readable instructions is that the computer-readable instructions are executed by one or more processors to implement the intelligent self-test method for pre-takeoff of a compound-wing UAV as described above.

[0016] The beneficial effects of this invention are as follows: By acquiring and preprocessing images of the UAV's exterior fixed at a preset position in real time, this invention establishes a non-contact visual inspection foundation. Utilizing propeller detection channels, control surface detection channels, and folding wing detection channels, it independently detects the state of the propeller, control surface, and folding wing in the preprocessed images, achieving direct visual verification of the motion state of key mechanical components of the UAV. This replaces manual visual inspection and can detect mechanical assembly and motion state anomalies that sensors cannot perceive. Finally, by mapping all detection results into fixed-length state vectors and comparing them element-by-element with a pre-stored standard state logic table, a takeoff or no-fly zone decision is automatically generated through global logic verification, forming a fully automated, unmanned intelligent quality inspection closed loop. This solves the dual problems of the inability to perform manual inspections and the existence of blind spots in sensors in existing technologies. Therefore, this invention can achieve fully automated, unmanned pre-takeoff safety quality inspection of UAVs and can detect the operating status of key moving parts of the UAV, thereby improving operational safety and maintenance efficiency in unattended scenarios. Attached Figure Description

[0017] Figure 1 This is an overall flowchart of an intelligent self-test method for a compound-wing unmanned aerial vehicle before takeoff, according to the present invention. Figure 2 This is a block diagram of an intelligent self-test system for a compound-wing unmanned aerial vehicle (UAV) before takeoff, according to the present invention. Figure 3 This is a hardware structure diagram of an electronic device as an example embodiment; Figure 4 A block diagram illustrating an electronic device as an example embodiment; Figure 5 This is a flowchart illustrating an intelligent self-test method for a compound-wing unmanned aerial vehicle (UAV) before takeoff, as an example embodiment. Detailed Implementation

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] The following will clearly and completely describe the concept, specific structure, and technical effects of the present invention in conjunction with embodiments and accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are all within the scope of protection of the present invention. Furthermore, all connections / linkages involved in the patent do not simply refer to direct contact between components, but rather to the ability to form a better connection structure by adding or reducing connecting accessories according to specific implementation conditions. The various technical features in this invention can be combined interactively without contradicting each other.

[0020] Reference Figure 1 and Figure 5 As shown, this invention discloses an intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle (UAV), which includes the following steps: S10, real-time acquisition of the appearance video of the target drone fixed at a preset position, and preprocessing of each frame of the appearance video; wherein, the specific method of preprocessing each frame of the appearance video is as follows: sequentially performing grayscale processing, contrast-limited adaptive histogram equalization processing, brightness compensation, and Gaussian filtering processing on each frame of the acquired appearance video.

[0021] It should be noted that in this embodiment, at least two high-definition industrial cameras with a resolution of no less than 1920×1080 and a frame rate of no less than 30fps are fixedly installed at the hangar exit to cover the front / top and side views of the UAV, respectively, to ensure complete visibility of the propeller, all control surfaces (ailerons, elevators, rudders), and the left and right folding wings. After the UAV is pushed out of the hangar and fixed in the preset position, the hangar management platform issues a unified detection start command, and the system begins to collect the appearance video stream in real time, and performs four preprocessing operations on each frame of the image in sequence: First, grayscale processing is performed, which converts the color image into a single-channel grayscale image to reduce computational costs, eliminate color interference, and improve algorithm compatibility; then, a contrast-limited adaptive histogram is used. The image is divided into 8×8 pixel sub-blocks for equalization. Histogram equalization is performed within each sub-block, and the contrast truncation threshold is set to 2.0. Then, bilinear interpolation is used to eliminate the boundaries between blocks, thereby enhancing local contrast and suppressing the effects of drastic changes in lighting inside and outside the hangar. Next, brightness compensation is performed by calculating the global average gray value of the entire frame. If it is lower than the preset low-light threshold (e.g., 40 gray levels), the brightness is automatically increased by a gain factor (e.g., 1.5 times) to ensure that the image is usable in nighttime or low-light environments. Finally, a Gaussian filter with a 5×5 convolution kernel and a standard deviation σ=1.0 is used to smooth and denoise the image, removing random noise caused by sensor thermal noise and environmental particles, while preserving edge information to provide stable input for subsequent steps. In addition, after image preprocessing is completed, a preset ROI coordinate configuration file needs to be loaded to mark the initial positions of each propeller region, each control surface region, and each folding wing region of the target UAV. This file accurately records the vertex coordinates of each propeller region (each propeller corresponds to a rectangular ROI), each control surface region (left / right aileron, left / right elevator, rudder), and the left / right folding wing region. The system uses these coordinates to crop out each sub-image from the preprocessed image, which serves as the input for the propeller detection channel, control surface detection channel, and folding wing detection channel, respectively, thus providing a unified and accurate image basis for the independent analysis of the three parallel channels.

[0022] S20 uses the propeller detection channel, control surface detection channel, and folding wing detection channel to perform state detection on the propeller, control surface, and folding wing of the target UAV in the preprocessed image, respectively. It should be noted that in this embodiment, a multi-channel parallel detection strategy is adopted, with the propeller detection channel, control surface detection channel, and folding wing detection channel operating independently. Each channel performs state analysis on the propeller, control surface, and folding wing respectively, and outputs standardized component state identifiers. Specifically: The propeller detection channel is used to perform state detection on the propellers of the target UAV in the preprocessed image. This includes propeller blade integrity detection and propeller blade rotation state, direction, and speed consistency detection. The blade integrity detection is performed in the static state before the propeller motor starts, and its specific execution method is as follows: First, a frame of the current time is extracted from the preprocessed image stream, and propeller sub-images are cropped based on ROI coordinates. Then, for each cropped propeller ROI sub-image, the Otsu method is used to automatically calculate a global threshold and perform binarization. Contour extraction is then performed on the binarized sub-images. During contour extraction, to eliminate background interference, the aspect ratio of the bounding rectangle of the contour is introduced as a geometric filtering constraint. Only contours with an aspect ratio greater than a preset threshold (3.0) are retained as candidates, and the contour with the largest area is selected as the main blade contour. Next, the following rotation- and translation-invariant geometric features are calculated for the extracted blade contours: Contour area A: The pixel area of ​​the blade on the image plane; Profile perimeter P: The length of the outer boundary of the blade profile; Circularity C: Measures the degree of deviation of the profile from a standard circle, and its calculation expression is: When the blades are standard circular, C≈1; when the blades are slender, C is much less than 1. Convex hull loss rate H: ,in, H represents the area of ​​the convex hull of the profile. The convex hull of a complete blade almost coincides with the profile, and H is close to zero. If the blade has a gap, break, or deformation, the convex hull area is significantly larger than the actual area, and H will increase significantly. Edge continuity index E: This index tracks abrupt changes in gradient magnitude at the edges of the profile. If the gradient magnitude at a point exceeds twice its local mean, it is marked as an abrupt change point. E represents the percentage of pixels at these abrupt changes in the total number of pixels in the profile. For a complete blade edge, the edge is smooth and continuous, and E tends to be 0. If cracks or defects exist, E will increase significantly.

[0023] Finally, the integrity of the propellers is double-verified through multi-blade lateral comparison and standard reference comparison. If any verification triggers an anomaly, the propeller integrity is determined to be abnormal. Among them, multi-blade lateral comparison involves calculating the median profile area of ​​all propellers of the current UAV. Median loss rate of convex hull As a dynamic benchmark for this test, if the convex hull defect rate H of a certain propeller is greater than 1.5... Or its outline area A is less than 0.9 If the blade is damaged or deformed, it is determined that the blade has defects or deformation. Standard reference comparison: A set of undamaged standard blade samples is pre-collected, and five geometric features are extracted: contour area A, convex hull defect rate H, perimeter P, circularity C, and edge continuity index E. Based on the statistical data of the sample set, the reference mean μ and standard deviation σ of each feature are calculated to construct a standard reference feature library. Each blade feature is compared one by one with the pre-calibrated standard blade reference values. For contour area A, perimeter P, and circularity C, if any feature value of a blade deviates from the reference mean by more than 3σ, the blade is determined to be abnormal. For unidirectional distribution features such as convex hull defect rate H and edge continuity index E, which are close to zero, absolute upper limit thresholds are set for convex hull defect rate and edge continuity index. and If the threshold is exceeded, it is considered abnormal; The consistency detection of blade rotation status, direction, and speed is performed after the blade integrity detection is passed. At this time, the system sends a command to the flight controller to start the propeller motor. After the propeller motor starts, the consistency detection of blade rotation status, direction, and speed begins. The specific execution method is as follows: First, within the propeller ROI region of the reference frame, the Shi-Tomasi corner detection algorithm is used to extract strong corners with significant gradient features. This algorithm improves upon traditional Harris corner detection: for the autocorrelation matrix M of the image gradient, its eigenvalues ​​λ1 and λ2 are calculated, and the minimum eigenvalue is directly used as the corner response score R=min(λ1,λ2). Only when the gray-level changes in both orthogonal directions are sufficiently drastic (i.e., both eigenvalues ​​are large) will the point be identified as a corner, thus making the detection results more robust to noise and edges. In actual calculations, the algorithm first calculates the response values ​​of all pixels, then performs non-maximum suppression to preserve local maxima, introduces a global quality threshold (taking the top 1% of corners by response value as candidates) and a minimum Euclidean distance constraint (spacing ≥ 10 pixels), and filters out feature points with the highest scores and uniform spatial distribution. Then, for the corner points extracted from the reference frame, the Pyramid Lucas-Kanade Optical Flow (PyrLK) algorithm is used for tracking and matching in the next frame to obtain the displacement vectors of each feature point. This algorithm constructs an overdetermined system of equations by performing a first-order Taylor expansion of the brightness constant equation Ix·u + Iy·v + It = 0 (where u and v are the pixel's motion velocities in the x and y directions, respectively), combined with the assumption of pixel spatial consistency within a local window (e.g., 3×3 or 5×5), and solves for the optimal optical flow vector using the least squares method. A four-layer image pyramid (scale factor 0.5) is used to achieve multi-scale tracking from coarse to fine, effectively solving the problem of single-scale LK's sensitivity to large displacements. Next, the tangential velocity components of each feature point are calculated based on the optical flow tracking results, and the directional scores are accumulated. Let the coordinates of the propeller center point be... The coordinates of the i-th feature point in the reference frame are The coordinates of the corresponding point in the tracking frame are The calculation steps are as follows: (a) Calculate and filter the radial component: Define the radial vector of the feature point in the reference frame relative to the propeller center. The calculation expression is: ; vector The expression for calculating the modulus is: ,when When the pixel size is less than 2 pixels, it is confirmed that the feature point is close to the center and its movement is mainly radial stretching rather than rotation, so it is filtered out.

[0024] (b) Calculate the tangential velocity component: displacement vector The calculation expression is: The tangential velocity component of feature point i relative to the center is calculated using the cross product of two-dimensional vectors. Its calculation expression is: :in, Indicates counterclockwise direction. It indicates a clockwise direction.

[0025] (c) Direction Score Statistics and Instantaneous Determination: For all valid feature points that pass radial filtering, direction scores are accumulated based on the tangential component. A tangential displacement threshold τ is set to 0.02 × R_pixel (R_pixel is the pixel length of the propeller radius in the image). Motion components below this value are considered static noise or subpixel jitter and are not included in the direction score. If the cumulative counterclockwise score is greater than or equal to τ, then the cumulative counterclockwise score is S. + = S + +1; if If the sum is less than or equal to τ, then the cumulative clockwise score S - = S - + 1. Introduce a direction determination advantage ratio threshold α (a dimensionless proportionality coefficient; in this embodiment, α is set to 1.3, but in actual implementation, it can be adjusted within the range of 1.2 to 1.5 according to camera resolution, frame rate, and image noise level). α=1.3 means that the vote count in one direction must exceed 30% of that in the other direction to be considered a valid rotation. When the vote counts in the two directions are close (such as when stationary or purely jittery), the system determines that there is no valid rotation, thereby filtering out the interference of random noise in both directions. The instantaneous direction determination rule for the current "reference frame → tracking frame" frame pair is: if S + >α·S - If S - >α·S +If the rotation direction of the propeller is clockwise in the frame pair, then the propeller is determined to be clockwise; otherwise, the propeller is determined to have no effective rotation in the frame pair.

[0026] Then, the instantaneous rotation direction of multiple adjacent frame pairs is comprehensively determined using a multi-frame analysis. The system employs a sliding frame pair strategy, continuously acquiring N frames of images after motor startup, sequentially forming N... For each pair of adjacent frames, perform the above steps independently to obtain N. The instantaneous direction determination results within group 1. Statistical analysis of these N... In the results of Group 1, the frequency of each category is used, and the category with the highest frequency is taken as the final direction code of the propeller. If the highest frequency category is "no effective rotation" or the two effective rotation directions have the same highest frequency, the propeller is determined not to be rotating normally, and its final direction code is recorded as 0; otherwise, the effective rotation direction with the higher frequency is used as the judgment result, and its final direction code is recorded as 1 (clockwise) or -1 (counterclockwise).

[0027] For a normally rotating propeller, calculate the inter-frame angular velocity based on the tangential velocity component of a feature point and its distance from the propeller center. Calculate the instantaneous angular velocity of this feature point. Its calculation expression is: To eliminate the influence of random noise and outliers in single-point optical flow estimation, for all effective feature points... Take the median to obtain the median inter-frame angular velocity of the propeller. .Will Multiplying by the video frame rate (fps) gives an approximate angular velocity (radians per second), which can then be converted to rotational speeds per minute (RPM). ; Finally, the RPM of all propellers is checked for consistency. After obtaining the estimated RPM values ​​for each propeller, the median RPM of all propellers is calculated. If the RPM of a propeller deviates from... If the dynamic deviation exceeds a preset threshold of 30%, or if its absolute rotational speed falls below the set lower speed limit, the propeller speed is determined to be abnormal. Based on the detection results of the propeller blade's rotational state, direction, and rotational speed, the corresponding status code is output: This indicates the rotation status code (0 for normal, 1 for abnormal rotation). This indicates the direction encoding (0 for normal direction, 1 for abnormal direction, and 2 for invalid direction). This indicates the rotational speed status code (0 for normal, 1 for abnormal, 2 for invalid).

[0028] The specific method for using the control surface detection channel to perform state detection on the control surfaces (including ailerons, elevators, and rudder) of the target UAV in the preprocessed image is as follows: First, before the flight controller issues the control surface deflection command, the current frame is extracted from the preprocessed image stream as the reference frame, and Canny edge detection is performed in the corresponding control surface ROI area to extract the control surface trailing edge feature point set. Then, the RANSAC algorithm is used to fit a straight line to the trailing edge feature point set, and the angle between the fitted straight line and the pre-calibrated reference line for the zero deflection position of the control surface is calculated. ; Next, after the flight controller issues the control surface deflection command, the action frame is extracted and the same edge detection and straight line fitting operations as the reference frame are performed on the action frame to obtain the angle between the fitted straight line after the action and the zero deflection position reference line. ; Then, through the formula calculate and deflection between The measured deflection direction is determined based on the sign of the measured deflection amount Δθ. The measured deflection direction is then compared with the deflection direction expected by the flight control command. If the directions are opposite, it is directly judged as an abnormal direction; if the direction is correct, the amplitude is further determined. If the value is within 20% of the expected value, the amplitude verification passes; otherwise, the amplitude is deemed abnormal. Finally, the above process is executed on the ailerons, elevators, and rudder respectively, and the status codes of each control surface are output. (Status code for left aileron) (Right aileron status code) (Status code of left elevator) (Right elevator status code) (Steering wheel status codes): A status code of 0 indicates normal operation, while a status code of 1 indicates an abnormal operation.

[0029] The folding wing detection channel adopts a three-level progressive judgment architecture: unfolding angle, locking confirmation, and micro-motion detection. The specific detection methods are as follows: First, the wing deployment angle is detected: After the flight controller issues the wing deployment command, the current frame is extracted from the preprocessed image stream. Edge detection and line fitting are performed within the preset wing ROI region to obtain several candidate line segments. The optimal line segment among the fitted candidate line segments is selected and used as the wing's principal edge. The angle between the selected wing principal edge and the fuselage axis is taken as the wing deployment angle. ,like If the wing deployment angle is greater than or equal to the preset wing deployment angle threshold, the wing deployment is considered normal. If the wing folding angle is less than the preset threshold, it is determined that the wings are not unfolded. If the wing angle is between the wing deployment angle threshold and the wing folding angle threshold, the wing is determined to be in a semi-deployed and stuck state. Next, the locking area is confirmed: within the preset small ROI of the wing root locking mechanism, Normalized Correlation Coefficient (NCC) template matching is used to compare the current frame with the calibrated locking-in standard template. The NCC calculation formula is:

[0030]

[0031] In the formula, Template image in pixel coordinates grayscale value at that location For the current frame ROI image at the corresponding position grayscale value at that location and These are the average grayscale values ​​of the template image and the current frame ROI image, respectively. For the offset position The correlation coefficient at the location, with a range of values ​​of 100,000. The closer the value is to 1, the more similar the two images are.

[0032] Set matching threshold (This embodiment uses 0.85; in actual implementation, it can be calibrated within the range of 0.75-0.90 based on the stability of on-site lighting and the quality of the template). If the optimal matching position is... If the locking mechanism is in place, proceed to the next level of testing; if If the locking mechanism is not in place, the abnormal handling process is triggered, and the process will not proceed to the next level of detection.

[0033] Then, perform micro-motion detection after locking: After confirming that the wing is deployed and locked in place, continuously acquire M frames of images, and calculate the inter-frame difference map for the micro-motion detection ROI in the far-end region of the wing frame by frame. The calculation expression is as follows: (x,y)=| (x,y) (x,y)|; The difference image is binarized and morphological opening is performed to remove isolated noise pixels with an area smaller than 5 pixels. The percentage of non-zero pixels in the processed difference image is as follows: =count( > ) / total_pixels; In the formula, δ is the difference threshold (usually taken as 15~25 gray levels), and total_pixels is the total number of pixels within the ROI. If there are more than T frames (T is set to M×40%) in M ​​frames. If the micro-motion threshold ε is exceeded (pre-calibrated based on camera resolution and wing geometry; 5% in this example), the locking is considered insecure; otherwise, the locking is considered reliable. Furthermore, this channel includes an abnormal re-check mechanism: if the wing is in a semi-deployed, stuck state, a wing retraction command is automatically issued, followed by a 2-second delay before a deployment command is issued again, and the deployment angle detection is re-executed. If the wing is still determined to be semi-deployed and stuck after two consecutive re-deployments, a hardware fault is confirmed and an alarm is triggered. If the micro-motion detection determines the locking is insecure, the micro-motion detection process is automatically re-executed after 2 seconds. If the re-detection still determines the locking is insecure, a hardware fault is confirmed and an alarm is triggered. Based on the results of the wing deployment angle detection, locking confirmation, and micro-motion detection, the detection status code for each wing is output. (Left wing deployment angle detection status code) (Right wing deployment angle detection status code) (Left wing lock detection status code) (Right wing lock detection status code) (Left wing locked after micro-motion detection status code) (Right wing locked after micro-motion detection status code), each detection status code takes a value between 0, 1, and 2, where 0 indicates normal, 1 indicates abnormal, and 2 indicates invalid.

[0034] S30 maps the state detection result in S20 into a fixed-length state vector and compares the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is deemed qualified and a takeoff permission command is output; otherwise, the self-test is deemed abnormal and a takeoff prohibition command is output.

[0035] It should be noted that, in this embodiment, after completing all tests in the propeller detection channel, control surface detection channel, and folding wing detection channel, each channel pushes its detection status and anomaly information to the integrated judgment layer. The integrated judgment layer aggregates the output results from all channels, performs global logical comparison and anomaly handling. Specifically, the integrated judgment layer reads the detection results from each channel through a standardized data interface. The detection results from each channel are then mapped into a fixed-length state vector according to a preset encoding rule. It consists of propeller subvectors rudder surface subvector and wing sub-vectors It is composed of multiple parts, and its representation is as follows: =[ , , ];in: =[ , , , , , , , , , , , , , , , ]; Subscripts 1 to 4 correspond to the four propeller numbers in this example (the actual number of propellers depends on the aircraft model configuration during deployment); =[ , , , , ].

[0036] =[ , , , , ].

[0037] Finally, the constructed measured state vector of the entire system will be... The system performs an element-by-element comparison with the standard state logic table for this aircraft model pre-stored in the database. If all elements in the measured vector match the standard state logic table, the UAV is deemed to have passed the self-test and a takeoff permission command is output. If any element is inconsistent, the UAV is deemed to have failed the self-test and a takeoff prohibition command is output. Based on the index position of the mismatched element, the abnormal component and abnormal type are parsed, and a structured fault report is generated and pushed to the hangar management platform.

[0038] Reference Figure 2 As shown, the present invention also discloses an intelligent self-test system 600 for pre-takeoff testing of a compound-wing unmanned aerial vehicle, comprising: Image acquisition unit 601 is used to acquire real-time images of the appearance of the target UAV fixed at a preset position; Image preprocessing unit 602 is used to preprocess the appearance image; Propeller detection channel 603 is used to detect the state of the propellers of the target UAV in the preprocessed image; Control surface detection channel 604 is used to perform state detection on the control surfaces of the target UAV in the preprocessed image; Folded wing detection channel 605 is used to detect the state of the folded wings of the target UAV in the preprocessed image; The integrated judgment unit 606 is used to map the state detection results of the propeller detection channel, control surface detection channel and folding wing detection channel into a fixed-length state vector, and compare the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is judged to be qualified and a takeoff permission command is output; otherwise, the self-test is judged to be abnormal and a takeoff prohibition command is output.

[0039] It should be noted that the intelligent self-test strategy for the compound wing UAV before takeoff provided in the above embodiments is only an example of the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed. That is, the internal structure of the intelligent self-test system 600 for the compound wing UAV before takeoff will be divided into different functional modules to complete all or part of the functions described above. Furthermore, the intelligent self-test system 600 for pre-flight testing of the compound-wing UAV provided in the above embodiments and the intelligent self-test method for pre-flight testing of the compound-wing UAV belong to the same concept. The specific way each module performs its operation has been described in detail in the method embodiments, and will not be repeated here.

[0040] Figure 3 A schematic diagram of the structure of an electronic device according to an exemplary embodiment is shown.

[0041] It should be noted that this electronic device is merely an example adapted to the present invention and should not be construed as providing any limitation on the scope of use of the present invention. Furthermore, this electronic device should not be interpreted as requiring or depending on having... Figure 3 One or more components of the exemplary electronic device 2000 shown.

[0042] The hardware structure of electronic devices 2000 can vary significantly due to differences in configuration or performance, such as... Figure 3 As shown, the electronic device 2000 includes: a power supply 210, an interface 230, at least one memory 250, and at least one central processing unit (CPU) 270.

[0043] Specifically, power supply 210 is used to provide operating voltage for various hardware devices on electronic device 2000.

[0044] Interface 230 includes at least one wired or wireless network interface 231 for interacting with external devices. Of course, in other examples adapted to this invention, interface 230 may further include at least one serial-to-parallel conversion interface 233, at least one input / output interface 235, and at least one USB interface 237, etc. Figure 3 As shown, this does not constitute a specific limitation.

[0045] The memory 250 serves as a carrier for resource storage and can be a read-only memory, random access memory, disk, or optical disk, etc. The resources stored on it include the operating system 251, application programs 253, and data 255, etc., and the storage method can be temporary storage or permanent storage.

[0046] The operating system 251 is used to manage and control the various hardware devices and application programs 253 on the electronic device 2000, so as to enable the central processing unit 270 to perform calculations and processing on the massive data 255 in the memory 250. It can be Windows Server™, Mac OS X™, Unix™, Linux™, FreeBSD™, etc.

[0047] Application 253 is a computer-readable instruction based on operating system 251 that performs at least one specific task, and may include at least one module ( Figure 3 (Not shown), each module can contain computer-readable instructions for electronic device 2000. For example, the intelligent self-test device before takeoff of the compound-wing UAV can be regarded as application program 253 deployed on electronic device 2000.

[0048] Data 255 may be signal information, etc., and is stored in memory 250.

[0049] The central processing unit 270 may include one or more processors and is configured to communicate with the memory 250 via at least one communication bus to read computer-readable instructions stored in the memory 250, thereby enabling the computation and processing of massive amounts of data 255 in the memory 250. For example, an intelligent self-test method before takeoff of a compound-wing UAV can be implemented by the central processing unit 270 reading a series of computer-readable instructions stored in the memory 250.

[0050] Furthermore, the present invention can also be implemented through hardware circuits or a combination of hardware circuits and software. Therefore, the implementation of the present invention is not limited to any specific hardware circuit, software, or combination thereof.

[0051] Please see Figure 4 This invention provides an electronic device 4000, which may include: a desktop computer, a laptop computer, a server, etc., with sensor recognition capabilities.

[0052] exist Figure 4 In this context, the electronic device 4000 includes at least one processor 4001 and at least one memory 4003.

[0053] The data interaction between the processor 4001 and the memory 4003 can be achieved through at least one communication bus 4002. This communication bus 4002 may include a path for transmitting data between the processor 4001 and the memory 4003. The communication bus 4002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0054] Optionally, the electronic device 4000 may further include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 4004 is not limited to one type, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present invention.

[0055] Processor 4001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 4001 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0056] The memory 4003 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program instructions or code in the form of instructions or data structures and accessible by the electronic device 4000, but not limited thereto.

[0057] The memory 4003 stores computer-readable instructions, and the processor 4001 can read the computer-readable instructions stored in the memory 4003 through the communication bus 4002.

[0058] The computer-readable instructions are executed by one or more processors 4001 to implement the intelligent self-test method for pre-flight takeoff of the compound-wing UAV in the above embodiments.

[0059] Furthermore, this embodiment of the invention provides a storage medium storing computer-readable instructions, which are executed by one or more processors to implement the intelligent self-test method for pre-takeoff of the compound-wing UAV as described above.

[0060] Compared with existing technologies, the beneficial effects of this invention are as follows: By acquiring and preprocessing images of the UAV's appearance fixed at a preset position in real time, a non-contact visual inspection foundation is established; and by utilizing propeller detection channels, control surface detection channels, and folding wing detection channels, the propeller, control surface, and folding wing in the preprocessed images are independently inspected for their states, achieving direct visual verification of the motion state of key mechanical components of the UAV, thus replacing manual visual inspection. Furthermore, it can detect mechanical assembly and motion state anomalies that sensors cannot perceive. Finally, by mapping all detection results into fixed-length state vectors and comparing them element-by-element with a pre-stored standard state logic table, a takeoff or no-fly zone decision is automatically generated through global logic verification, forming a fully unmanned intelligent quality inspection closed loop. This solves the dual problems of manual implementation and sensor blind spots in existing technologies. Therefore, this invention can achieve fully unmanned pre-takeoff safety quality inspection of UAVs and can detect the operating status of key moving parts of the UAV, thereby improving operational safety and maintenance efficiency in unattended scenarios.

[0061] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the embodiments described. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A method for intelligent self-testing of a compound-wing unmanned aerial vehicle before takeoff, characterized in that, The intelligent self-test method for the compound-wing UAV before takeoff includes the following steps: S10: Real-time acquisition of the appearance video of the target drone fixed at a preset position, and preprocessing of each frame of the appearance video. S20, the propeller detection channel, control surface detection channel, and folding wing detection channel are used to perform state detection on the propeller, control surface, and folding wing of the target UAV in the preprocessed image, respectively. Specifically, the propeller state detection using the propeller detection channel includes propeller blade integrity detection and propeller blade rotation state, direction, and speed consistency detection. The specific method for propeller blade integrity detection is as follows: In the static state before the propeller motors of the target drone start, extract a frame of the current moment from the preprocessed image stream and crop out each propeller sub-image according to the ROI coordinates. Each propeller sub-image is binarized, and the contours of the binarized sub-images are extracted. The geometric features of the profile are calculated, and the integrity of the blade is double-verified by multi-blade lateral comparison and standard reference comparison. If any verification triggers an anomaly, the blade integrity is determined to be abnormal. The specific method for detecting the consistency of propeller blade rotation state, direction, and speed is as follows: Start the propeller motor and perform feature point extraction and optical flow tracking on the frame sequence continuously acquired after the propeller motor starts; The tangential velocity components of each feature point are calculated based on the optical flow tracking results, and the direction score is accumulated. The instantaneous rotation direction of adjacent frame pairs is determined by whether the dominance ratio between the accumulated direction scores exceeds the preset direction determination dominance ratio threshold. The instantaneous rotation direction of multiple adjacent frame pairs is comprehensively determined by multi-frame analysis. The category with the highest frequency is taken as the final direction code of the propeller. If the highest frequency category is no effective rotation or the frequencies of two effective rotation directions are tied, the propeller is determined not to be rotating normally. For a normally rotating propeller, the inter-frame angular velocity is calculated based on the tangential velocity components of the feature points and their distances to the propeller center, and then converted into rotational speeds per minute (RPM). The RPM of all propellers is checked for consistency. If the RPM of a propeller deviates from the median by more than the preset dynamic deviation threshold or is lower than the speed limit, the propeller speed is determined to be abnormal. The detection status code of each blade is output based on the detection results of the blade's rotation status, direction, and speed. S30 maps the state detection result in S20 into a fixed-length state vector and compares the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is deemed qualified and a takeoff permission command is output; otherwise, the self-test is deemed abnormal and a takeoff prohibition command is output.

2. The intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle according to claim 1, characterized in that, In step S10, the specific method for preprocessing each frame of the appearance video is as follows: performing grayscale processing, contrast-limited adaptive histogram equalization processing, brightness compensation, and Gaussian filtering processing on each frame of the acquired appearance video in sequence.

3. The intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle according to claim 2, characterized in that, After image preprocessing is complete, a preset ROI coordinate configuration file needs to be loaded to mark the initial positions of each propeller area, each control surface area, and each folding wing area of ​​the target drone.

4. The intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle according to claim 3, characterized in that, The specific method for using the control surface detection channel to perform state detection on the control surfaces of the target UAV in the preprocessed image is as follows: Before the flight controller issues the control surface deflection command, the current frame is extracted from the preprocessed image stream as the reference frame, and Canny edge detection is performed in the corresponding control surface ROI area to extract the control surface trailing edge feature point set. The RANSAC algorithm is used to fit a straight line to the trailing edge feature point set, and the angle between the fitted line and the pre-calibrated reference line for the zero deflection position of the control surface is calculated. ; After the flight controller issues the control surface deflection command, the action frame is extracted and the same edge detection and straight line fitting operations as the reference frame are performed on the action frame to obtain the angle between the fitted straight line after the action and the zero deflection position reference line. ; Through formula calculate and deflection between The measured deflection direction is determined based on the sign of the measured deflection amount Δθ. The measured deflection direction is then compared with the deflection direction expected by the flight control command. If the directions are opposite, it is directly judged as an abnormal direction; if the direction is correct, the amplitude is further determined. If the value is within 20% of the expected value, the amplitude verification passes; otherwise, the amplitude is deemed abnormal. Perform the above procedure on the aileron, elevator, and rudder respectively, and output the status code of each control surface.

5. The intelligent self-test method for pre-takeoff of a compound-wing unmanned aerial vehicle according to claim 4, characterized in that, The specific method for performing state detection of the folded wings of a target UAV in a preprocessed image using the folded wing detection channel is as follows: After the flight controller issues the wing deployment command, the current frame is extracted from the preprocessed image stream, and edge detection and line fitting are performed within the preset wing ROI region to obtain several candidate line segments. The optimal line segment among the candidate line segments obtained from the fitting is selected and used as the main edge of the wing. The angle between the selected wing principal edge and the fuselage axis is taken as the wing deployment angle. ,like If the wing deployment angle is greater than or equal to the preset wing deployment angle threshold, the wing deployment is considered normal. If the wing folding angle is less than the preset threshold, it is determined that the wings are not unfolded. If the wing angle is between the wing deployment angle threshold and the wing folding angle threshold, the wing is determined to be in a semi-deployed and stuck state. Within the preset small ROI of the wing root locking mechanism, normalized correlation coefficient template matching is used to compare the current frame with the calibrated locking standard template. If the best matching correlation coefficient is greater than or equal to the preset matching threshold, the locking is determined to be in place; otherwise, the locking is determined to be in place. After confirming that the wings are deployed and locked in place, multiple frames of images are continuously acquired. For the micro-motion detection ROI in the far end area of ​​the wings, the inter-frame difference map is calculated frame by frame. The proportion of non-zero pixels in the difference map is counted. If the proportion of frames exceeding the preset proportion exceeds the preset micro-motion threshold, the locking is determined to be unsecured; otherwise, the locking is determined to be reliable. Based on the results of wing deployment angle detection, lock confirmation, and micro-motion detection, the detection status code for each wing is output.

6. The intelligent self-test method for a compound-wing unmanned aerial vehicle before takeoff according to claim 5, characterized in that, An anomaly re-detection mechanism is set up during the condition detection of the folding wing. The anomaly re-detection mechanism is as follows: If the wing deployment angle detection determines that the wing is in a semi-deployed and stuck state, the wing retraction command will be automatically issued. After a delay, the deployment command will be issued again and the deployment angle detection will be re-executed. If the wing is still determined to be semi-deployed and stuck after two consecutive re-deployments, a hardware fault will be confirmed and an alarm will be issued. If the micro-motion detection determines that the locking is not secure, the micro-motion detection process will be automatically re-executed after a preset time. If the re-detection still determines that the locking is not secure, it will be confirmed as a hardware failure and an alarm will be triggered.

7. An intelligent self-test system for pre-takeoff of a compound-wing unmanned aerial vehicle (UAV), used to implement the intelligent self-test method for pre-takeoff of the compound-wing UAV as described in any one of claims 1-6, characterized in that, include: The image acquisition unit is used to acquire real-time video of the appearance of the target drone fixed at a preset position. An image preprocessing unit is used to preprocess each frame of the appearance video; The propeller detection channel is used to detect the state of the propellers of the target UAV in the preprocessed image; The control surface detection channel is used to detect the state of the control surfaces of the target UAV in the preprocessed image; The folded wing detection channel is used to detect the state of the folded wings of the target UAV in the preprocessed image; The integrated judgment unit is used to map the state detection results of the propeller detection channel, control surface detection channel and folding wing detection channel into a fixed-length state vector, and compare the state vector with the pre-stored standard state logic table. If they are all consistent, the self-test is judged to be qualified and a takeoff permission command is output; otherwise, the self-test is judged to be abnormal and a takeoff prohibition command is output.

8. An electronic device, characterized in that, Includes at least one processor and at least one memory, wherein, The memory stores computer-readable instructions; The computer-readable instructions are executed by one or more processors, causing the electronic device to implement the intelligent self-test method for pre-takeoff of the compound-wing unmanned aerial vehicle as described in any one of claims 1-6.

9. A storage medium having computer-readable instructions stored thereon, characterized in that, The computer-readable instructions are executed by one or more processors to implement the intelligent self-test method for pre-takeoff of the compound-wing UAV as described in any one of claims 1-6.

Images