Unmanned aerial vehicle parking state monitoring method and device and readable storage medium

Abstract

The invention discloses an unmanned aerial vehicle parking state monitoring method and device and a readable storage medium, and the method comprises the steps: detecting an object entering an empty parking space, obtaining an image, inputting the image into a deep learning model to obtain a preliminary detection target and a corresponding bounding box, extracting feature data in the bounding box, and comparing the feature data with an unmanned aerial vehicle feature database; determining whether the preliminary detection target is a specified unmanned aerial vehicle according to the comparison result, if yes, identifying whether the unmanned aerial vehicle is completely parked at the central position of the berth, and if not, adjusting the unmanned aerial vehicle to reach the central position; if the unmanned aerial vehicle is parked at the central position, acquiring an ArUco code at the bottom of the fuselage of the unmanned aerial vehicle, and analyzing an identity identifier corresponding to the ArUco code; and judging whether the identity identifier is the identity identifier of the unmanned aerial vehicle in the navigation station, and if so, finishing the parking of the unmanned aerial vehicle. According to the invention, high-precision perception and discrimination of the identity and the spatial parking position of the unmanned aerial vehicle are realized, and the whole process is automatic.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) technology, specifically to a method, device, and readable storage medium for monitoring the docking status of a UAV. Background Technology

[0002] In recent years, with the gradual liberalization of low-altitude airspace management policies and the rapid development of technologies such as intelligent sensing, communication, and navigation, drone-related technological products have become an important component of the low-altitude economy. In various industrial scenarios such as logistics distribution, inspection and monitoring, agricultural plant protection, and emergency rescue, drones are evolving from "auxiliary tools" to "core operational units." However, to achieve large-scale commercial deployment and routine operation of drone systems, it is urgent to solve key bottlenecks such as high maintenance costs, frequent manual intervention, and limited operational efficiency. Among these, how to achieve "unmanned" operation and maintenance and significantly reduce the marginal cost per flight has become an important research topic in the current drone-related industrial field.

[0003] In the current practice of "unmanned operation and maintenance" of drones, automatic docking, as a core component, has made some technological progress, but still faces three key problems in actual operation, which seriously restrict the system's reliability and large-scale deployment:

[0004] 1. Inability to perform reliable identification: When a drone approaches a parking spot, it needs to quickly and accurately complete identification to ensure that only authorized devices can access the charging, data uploading, or mission scheduling systems. However, existing solutions often fail to identify drones due to communication interference, tag failure, protocol incompatibility, or environmental obstructions (such as rain, fog, or metal structures). Incorrect or missing identification can not only lead to the risk of unauthorized device access but may also cause the system to refuse docking to legitimate drones, resulting in mission interruption or drone idleness.

[0005] 2. Inability to Achieve High-Precision Positioning: Automated docking requires the UAV to achieve centimeter-level positioning accuracy in the final stage to align with charging contacts or mechanical locking mechanisms. However, in complex scenarios such as urban canyons, indoor environments, forests, or areas with strong electromagnetic interference, GNSS signals are easily blocked or distorted, while visual / laser-assisted positioning methods are susceptible to changes in lighting, dynamic obstacles, or wear and tear on berth markers, leading to positioning drift or even complete failure. Inaccurate positioning will directly cause landing deviation, collision with berths, or inability to complete docking, threatening flight safety.

[0006] 3. Inability to Achieve End-to-End Full Automation: Ideal unmanned operation and maintenance should cover a closed loop encompassing the entire process from mission completion and return, precise landing, energy replenishment, to status self-checking and standby. However, most current systems still require manual intervention—for example, manually resetting a stuck robotic arm, clearing foreign objects from the berth, handling reconnection after communication interruptions, or manually scheduling multiple machines in case of conflict. This "semi-automatic" mode not only increases operation and maintenance costs but also weakens the system's availability and scalability in 24 / 7 continuous operation scenarios.

[0007] In summary, unreliable identity recognition, unstable precise positioning, and incomplete process automation have become the three major bottlenecks restricting the automatic docking of drones from becoming truly "unmanned." Solving these problems is a key prerequisite for building an efficient, safe, and scalable ecosystem of low-altitude unmanned systems. Summary of the Invention

[0008] Therefore, this application provides a method, device, and readable storage medium for monitoring the docking status of unmanned aerial vehicles (UAVs) to solve the problems of inaccurate identification, inaccurate positioning, and incomplete automation of the process during UAV docking in the prior art.

[0009] To achieve the above objectives, this application provides the following technical solution:

[0010] Firstly, a method for monitoring the docking status of unmanned aerial vehicles (UAVs) includes:

[0011] Step 1: Use a pre-calibrated wide-angle global monitoring camera to detect objects entering empty berths and acquire images;

[0012] Step 2: Input the image into a pre-trained deep learning model for object detection to obtain preliminary detected targets and corresponding bounding boxes;

[0013] Step 3: Extract the feature data from the bounding box and compare it with the pre-established UAV feature database. Based on the comparison results, determine whether the preliminary detection target is the specified UAV.

[0014] Step 4: If the initial detection target is not the designated drone, issue an alert;

[0015] Step 5: If the initial detection target is a designated drone, then the vision system is used to identify whether the drone is completely docked in the center of the berth.

[0016] Step 6: If the drone is not completely centered in the berth, adjust it to the center position and secure it.

[0017] Step 7: If the drone is completely docked in the center of the berth, use the close-up recognition camera to obtain the ArUco code on the bottom of the drone fuselage and parse the identity identifier corresponding to the ArUco code.

[0018] Step 8: Determine whether the identification identifier is a drone identification identifier within the airport;

[0019] Step 9: If the identification identifier is not a drone identification identifier within the airport, issue an alarm;

[0020] Step 10: If the identification identifier is a drone identification identifier within the station, then the drone docking is complete.

[0021] Preferably, step 1, when pre-calibrating the wide-angle global monitoring camera, specifically includes: continuously acquiring video streams of empty berths through the wide-angle global monitoring camera, correcting distorted images generated in the video stream, adjusting the brightness and contrast of the video stream image, and then performing color balancing to standardize the image.

[0022] Preferably, in step 1, the wide-angle global monitoring camera is fixedly installed on the ceiling of the terminal, with the lens installed vertically downwards.

[0023] Preferably, in step 2, the deep learning model is the YOLO model.

[0024] Preferably, in step 6, when the UAV reaches the center position, the position of the UAV on the XY axis is adjusted by using the grid clamps on the station.

[0025] Preferably, in step 6, a clamping mechanism is used to fix the drone.

[0026] Preferably, in step 7, the close-up recognition camera is fixedly positioned directly below the drone's fuselage, with its field of view aligned with the theoretical center point of the parking space, the ArUco code position.

[0027] Preferably, in step 7, when parsing the identity identifier corresponding to the ArUco code, the ArUco code needs to be corrected by perspective transformation before decoding.

[0028] Secondly, a drone docking status monitoring device includes:

[0029] The monitoring module is used to detect objects entering empty berths using a pre-calibrated wide-angle global monitoring camera and to acquire images;

[0030] The object detection module is used to input the image into a pre-trained deep learning model for object detection, and obtain preliminary detected objects and corresponding bounding boxes;

[0031] The feature comparison module is used to extract feature data from the bounding box and compare it with a pre-established UAV feature database.

[0032] The first judgment module is used to determine whether the preliminary detection target is the designated drone based on the comparison result; if the preliminary detection target is not the designated drone, an alarm is issued.

[0033] The second judgment module is used to identify whether the drone is completely docked in the center of the berth if the preliminary detection target is a designated drone; if it is not completely docked in the center of the berth, the drone is adjusted to the center position and fixed.

[0034] The identification identifier determination module is used to obtain the ArUco code on the bottom of the drone fuselage using a close-up recognition camera if the drone is completely docked in the center of the berth, and to parse the identification identifier corresponding to the ArUco code.

[0035] The third judgment module is used to determine whether the identity identifier is a drone identity identifier within the terminal; if the identity identifier is not a drone identity identifier within the terminal, an alarm is issued; if the identity identifier is a drone identity identifier within the terminal, the drone docking is completed.

[0036] Thirdly, a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for monitoring the docking status of an unmanned aerial vehicle (UAV).

[0037] Compared with the prior art, this application has at least the following beneficial effects:

[0038] This application provides a method for monitoring the docking status of unmanned aerial vehicles (UAVs), comprising: using a wide-angle global monitoring camera to detect objects entering an empty berth and acquiring images; inputting the images into a deep learning model for target detection to obtain preliminary detected targets and corresponding bounding boxes; extracting feature data from the bounding boxes and comparing them with a UAV feature database, and determining whether the preliminary detected target is a designated UAV based on the comparison results; if it is a designated UAV, identifying whether the UAV is completely docked at the center of the berth through a vision system; if it is not docked at the center, adjusting the UAV to the center and fixing it; if it is docked at the center, acquiring the ArUco code on the bottom of the UAV fuselage using a close-up recognition camera and parsing the identity identifier corresponding to the ArUco code; determining whether the identity identifier is the UAV identity identifier within the terminal, and if so, the UAV docking is complete, thus achieving high-precision perception and identification of the UAV's identity and its spatial docking position, and the entire process is fully automated. Attached Figure Description

[0039] To more intuitively illustrate the prior art and this application, exemplary drawings are provided below. It should be understood that the specific shapes and structures shown in the drawings should not generally be regarded as limiting conditions for implementing this application; for example, based on the technical concept disclosed in this application and the exemplary drawings, those skilled in the art are able to easily make conventional adjustments or further optimizations to the addition / reduction / classification, specific shapes, positional relationships, connection methods, size ratios, etc. of certain units (components).

[0040] Figure 1 This is a basic flowchart of a method for monitoring the docking status of an unmanned aerial vehicle (UAV) according to Embodiment 1 of this application.

[0041] Figure 2 The flowchart of a method for monitoring the docking status of a drone provided in Embodiment 1 of this application is shown. Detailed Implementation

[0042] The present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0043] In the description of this application: unless otherwise stated, "a plurality of" means two or more. The terms "first," "second," "third," etc., in this application are intended to distinguish the objects referred to and do not have any special meaning in terms of technical connotation (e.g., they should not be construed as an emphasis on importance or order). Expressions such as "including," "comprising," and "having" also mean "not limited to" (certain units, components, materials, steps, etc.).

[0044] The terms used in this application, such as "upper," "lower," "left," "right," and "middle," are generally used to indicate the general relative positional relationship for the purpose of intuitive understanding by referring to the accompanying drawings, and are not absolute limitations on the positional relationship in the actual product.

[0045] Example 1

[0046] Please see Figure 1 and Figure 2 This embodiment provides a method for monitoring the docking status of a drone, including:

[0047] S1: Detect objects entering empty berths using a pre-calibrated wide-angle global monitoring camera and acquire images;

[0048] Specifically, in this embodiment, the wide-angle global monitoring camera is deployed on the ceiling of the terminal, with a downward-looking angle, installed almost vertically downwards to ensure there are no blind spots. The wide-angle global monitoring camera is used to detect the entry and coarse positioning of drones, and to monitor the overall status of the entire berth area.

[0049] After the wide-angle global monitoring camera is installed, it needs to be calibrated. Specifically, when the drone is powered off, powered on but not yet taken off, and after landing, the image distortion caused by the wide-angle lens is corrected based on the video stream continuously captured by the wide-angle global monitoring camera at the berth to ensure the accuracy of the image geometry. Then, the brightness and contrast of the image are adjusted, and color balance is performed to standardize the image and eliminate the influence of changes in lighting.

[0050] After the wide-angle global monitoring camera completes calibration, it detects moving or newly appearing objects that are different from the background of the empty berth in real time. Once a large object is detected entering the berth, the subsequent recognition process is immediately triggered.

[0051] S2: Input the image into a pre-trained deep learning model for object detection to obtain preliminary detected targets and corresponding bounding boxes;

[0052] Specifically, in this embodiment, the deep learning model uses the YOLO model. Training the YOLO model requires calibrating a large number of images of the target drone under various environmental conditions and parking directions, as well as background images of the "empty parking spaces" under various environmental conditions and parking directions. Deep learning is then performed using the YOLO model. When each drone enters the warehouse, images are taken from multiple angles, and features are extracted and stored in the database.

[0053] This step uses a pre-trained deep learning YOLO model to identify the "drone" as a whole in the image and give its bounding box.

[0054] S3: Extract the feature data in the bounding box and compare it with the pre-established UAV feature database. Based on the comparison results, determine whether the preliminary detection target is the specified UAV.

[0055] Specifically, this step requires detecting whether there are unique appearances on the drone within the bounding box and comparing them with a pre-established drone feature database. Based on the comparison results, it is determined whether the drone is a drone and whether the model is correct (i.e., whether it is the specified drone).

[0056] S4: If the initial detection target is not the designated drone, issue an alert;

[0057] S5: If the initial detection target is a designated drone, the vision system will identify whether the drone is completely parked in the center of the berth.

[0058] Specifically, this step uses a vision system to identify whether the drone is completely docked in the center of the designated berth, checks the docking quality, and determines whether the drone is parked within the allowable tolerance range.

[0059] S6: If the drone is not completely centered in the berth, adjust the drone to the center position and secure it.

[0060] Specifically, if the drone is not completely docked in the center of the berth, the drone's position on the XY axis is adjusted using the grid clamps on the terminal to bring it to the designated position. Then, the drone is clamped and secured by the clamping mechanism so that it is in the same planar position before each launch from the terminal and after each recovery.

[0061] S7: If the drone is completely parked in the center of the berth, the close-up recognition camera will be used to obtain the ArUco code on the bottom of the drone and the corresponding identity identifier will be parsed.

[0062] Specifically, in this embodiment, the close-up recognition camera is deployed directly below the drone's fuselage, with its field of view aligned with the theoretical center point of the berth, the ArUco code position.

[0063] This step involves using a high-definition camera with close-up recognition to record the ArUco code affixed to the bottom of the drone's fuselage, performing perspective transformation correction, and then decoding it to directly obtain its built-in unique identifier. This identifier is strongly bound to the drone's serial number in the system.

[0064] S8: Determine if the identifier is a drone identifier within the station;

[0065] S9: If the identification identifier is not the identification identifier of the drone within the station, issue an alarm;

[0066] S10: If the identification identifier is the same as the drone identification identifier within the station, then the drone docking is complete.

[0067] The UAV docking status monitoring method provided in this embodiment achieves high-precision perception and identification of UAV identity and its spatial docking location, and the entire process is fully automated, solving three core problems in UAV docking: unreliable identity recognition, unstable accurate positioning, and incomplete process automation.

[0068] The UAV docking status monitoring method provided in this embodiment requires, during actual deployment, uniform and shadowless lighting to avoid glare and shadows. High-precision camera calibration is essential; otherwise, attitude estimation results will be inaccurate. Large airfields may require multiple cameras to cover different areas, necessitating solutions to coordinate system issues between cameras. It also needs to handle situations such as partial occlusion, sudden changes in lighting, and motion blur. When deployed on embedded devices, real-time performance must be ensured.

[0069] This embodiment provides a method for monitoring the docking status of unmanned aerial vehicles (UAVs, parking spaces), serving as a crucial bridge connecting the physical world (UAVs, parking spaces) and the digital world (data from the control system). Its application areas collectively point to a future: an efficient, reliable, and scalable ecosystem of unmanned systems, characterized by highly automated machines, intelligent infrastructure, and a central intelligent brain working in tandem, as well as UAV air stations and UAV swarm operations.

[0070] Example 2

[0071] This embodiment provides a drone docking status monitoring device, including:

[0072] The monitoring module is used to detect objects entering empty berths using a pre-calibrated wide-angle global monitoring camera and to acquire images;

[0073] The object detection module is used to input the image into a pre-trained deep learning model for object detection, and obtain preliminary detected objects and corresponding bounding boxes;

[0074] The feature comparison module is used to extract feature data from the bounding box and compare it with a pre-established UAV feature database.

[0075] The first judgment module is used to determine whether the preliminary detection target is the designated drone based on the comparison result; if the preliminary detection target is not the designated drone, an alarm is issued.

[0076] The second judgment module is used to identify whether the drone is completely docked in the center of the berth if the preliminary detection target is a designated drone; if it is not completely docked in the center of the berth, the drone is adjusted to the center position and fixed.

[0077] The identification identifier determination module is used to obtain the ArUco code on the bottom of the drone fuselage using a close-up recognition camera if the drone is completely docked in the center of the berth, and to parse the identification identifier corresponding to the ArUco code.

[0078] The third judgment module is used to determine whether the identity identifier is a drone identity identifier within the terminal; if the identity identifier is not a drone identity identifier within the terminal, an alarm is issued; if the identity identifier is a drone identity identifier within the terminal, the drone docking is completed.

[0079] For details on the specific implementation of each module in a UAV docking status monitoring device, please refer to the above description of the UAV docking status monitoring method; further details will not be provided here.

[0080] Example 3

[0081] This embodiment provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of a method for monitoring the docking status of an unmanned aerial vehicle (UAV).

[0082] The technical features of the above embodiments can be combined in any way (as long as there is no contradiction in the combination of these technical features). For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described; these embodiments not explicitly written should also be considered to be within the scope of this specification.

Claims

1. A method for monitoring the docking status of unmanned aerial vehicles (UAVs), characterized in that, include: Step 1: Use a pre-calibrated wide-angle global monitoring camera to detect objects entering empty berths and acquire images; Step 2: Input the image into a pre-trained deep learning model for object detection to obtain preliminary detected targets and corresponding bounding boxes; Step 3: Extract the feature data from the bounding box and compare it with the pre-established UAV feature database. Based on the comparison results, determine whether the preliminary detection target is the specified UAV. Step 4: If the initial detection target is not the designated drone, issue an alert; Step 5: If the initial detection target is a designated drone, then the vision system is used to identify whether the drone is completely docked in the center of the berth. Step 6: If the drone is not completely centered in the berth, adjust it to the center position and secure it. Step 7: If the drone is completely docked in the center of the berth, use the close-up recognition camera to obtain the ArUco code on the bottom of the drone fuselage and parse the identity identifier corresponding to the ArUco code. Step 8: Determine whether the identification identifier is a drone identification identifier within the airport; Step 9: If the identification identifier is not a drone identification identifier within the airport, issue an alarm; Step 10: If the identification identifier is a drone identification identifier within the station, then the drone docking is complete.

2. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 1, the pre-calibration of the wide-angle global monitoring camera specifically includes: continuously acquiring video streams of empty berths through the wide-angle global monitoring camera, correcting distorted images generated in the video stream, adjusting the brightness and contrast of the video stream image, and then performing color balancing to standardize the image.

3. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 1, the wide-angle global monitoring camera is fixedly installed on the ceiling of the terminal, with the lens installed vertically downwards.

4. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 2, the deep learning model used is the YOLO model.

5. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 6, when the UAV reaches the center position, the position of the UAV on the XY axis is adjusted by using the grid clamps on the station.

6. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 6, a clamping mechanism is used to fix the drone.

7. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 7, the close-up recognition camera is fixedly installed directly below the drone's fuselage, with its field of view aligned with the theoretical center point of the parking space, the ArUco code position.

8. The method for monitoring the docking status of a UAV according to claim 1, characterized in that, In step 7, when parsing the identity identifier corresponding to the ArUco code, the ArUco code needs to be corrected by perspective transformation before decoding.

9. A device for monitoring the docking status of unmanned aerial vehicles (UAVs), characterized in that, include: The monitoring module is used to detect objects entering empty berths using a pre-calibrated wide-angle global monitoring camera and to acquire images; The object detection module is used to input the image into a pre-trained deep learning model for object detection, and obtain preliminary detected objects and corresponding bounding boxes; The feature comparison module is used to extract feature data from the bounding box and compare it with a pre-established UAV feature database; The first judgment module is used to determine whether the preliminary detection target is the designated drone based on the comparison results; If the initial detection target is not the designated drone, an alert will be issued; The second judgment module is used to identify whether the drone is completely docked in the center of the berth if the preliminary detection target is a designated drone; if it is not completely docked in the center of the berth, the drone is adjusted to the center position and fixed. The identification identifier determination module is used to obtain the ArUco code on the bottom of the drone fuselage using a close-up recognition camera if the drone is completely docked in the center of the berth, and to parse the identification identifier corresponding to the ArUco code. The third judgment module is used to determine whether the identity identifier is a drone identity identifier within the terminal; if the identity identifier is not a drone identity identifier within the terminal, an alarm is issued; if the identity identifier is a drone identity identifier within the terminal, the drone docking is completed.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.

Images