Unmanned aerial vehicle automatic mounting method, equipment and medium

Abstract

The invention discloses an unmanned aerial vehicle automatic mounting method, equipment and a medium. The method comprises the following steps: acquiring a satellite signal of a target platform through a positioning system of a specified unmanned aerial vehicle; determining coarse particle position information of the target platform based on the satellite signal, so that the specified unmanned aerial vehicle approaches the target platform based on the coarse particle position information; obtaining fusion information of the specified unmanned aerial vehicle relative to the target platform, wherein the fusion information comprises pose information, height information and motion information; and determining fine particle position information of the target platform based on the fusion information so as to mount the specified unmanned aerial vehicle to the target platform based on the fine particle position information. The whole process autonomy is achieved through a coarse-to-fine two-stage positioning mechanism, the necessity of manual remote control or monitoring is eliminated fundamentally, the mounting failure risk and potential safety hazards caused by operator skill difference, judgment delay or environmental interference are remarkably reduced, and the mounting reliability is improved. And the unmanned aerial vehicle can stably and efficiently complete the mounting task in a complex dynamic environment.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicles, and in particular to an unmanned aerial vehicle automatic mounting method, device and medium. BACKGROUND

[0002] In recent years, unmanned aerial vehicle technology has shown great potential and value in the fields of logistics transportation, emergency rescue, equipment deployment, military applications, etc. One of the key application scenarios is to achieve autonomous mounting, landing or docking of unmanned aerial vehicles on mobile or fixed platforms (such as vehicles, ships, and specific ground facilities) to complete tasks such as energy supply, material transfer, sensor replacement, or data backhaul.

[0003] Existing unmanned aerial vehicle mounting technology is generally highly dependent on human intervention. During the mounting process, the operator usually needs to remotely monitor the status information of the target platform and the unmanned aerial vehicle in real time, and controls the attitude, position and speed of the unmanned aerial vehicle through manual remote control or complex guidance instructions, so that it is finally accurately aligned and stably landed on the mounting point of the target platform. This process not only requires the operator to have superb operation skills and on-site judgment ability, but also faces many challenges in actual application. SUMMARY

[0004] One or more embodiments of the present specification provide an unmanned aerial vehicle automatic mounting method, device and medium to solve the technical problems raised in the background.

[0005] One or more embodiments of the present specification adopt the following technical solutions:

[0006] One or more embodiments of the present specification provide an unmanned aerial vehicle automatic mounting method, which comprises:

[0007] acquiring a satellite signal of a target platform through a positioning system of a specified unmanned aerial vehicle;

[0008] determining coarse-grained position information of the target platform based on the satellite signal, so that the specified unmanned aerial vehicle approaches the target platform based on the coarse-grained position information;

[0009] acquiring fusion information of the specified unmanned aerial vehicle relative to the target platform, the fusion information including pose information, height information and motion information;

[0010] determining fine-grained position information of the target platform based on the fusion information, so that the specified unmanned aerial vehicle is mounted to the target platform based on the fine-grained position information.

[0011] It should be noted that the application realizes full-process autonomy through a two-stage positioning mechanism (coarse-grained satellite positioning guide approach, fine-grained precise positioning control mounting by fusing multi-source information), which fundamentally eliminates the necessity of manual remote control or monitoring, significantly reduces the risk of mounting failure and safety hazards caused by differences in operator skills, delays in judgment or environmental interference, and enables the unmanned aerial vehicle to complete the mounting task stably and efficiently in a complex dynamic environment.

[0012] Further, the acquisition of the fusion information of the target platform comprises:

[0013] acquiring image information of the target platform containing a preset visual marker through an image acquisition system of the specified unmanned aerial vehicle;

[0014] determining, based on the image information of the preset visual marker, pose information of the specified unmanned aerial vehicle relative to the target platform, the pose information comprising position information and running attitude information of the specified unmanned aerial vehicle relative to the target platform;

[0015] acquiring height information of the specified unmanned aerial vehicle relative to the target platform through a ranging system of the specified unmanned aerial vehicle;

[0016] acquiring motion information of the specified unmanned aerial vehicle relative to the target platform through an inertial measurement system of the specified unmanned aerial vehicle.

[0017] It should be noted that the application fuses three sensors of visual marker positioning, active ranging and inertial motion measurement, so that the unmanned aerial vehicle can autonomously generate a three-dimensional relative pose-motion comprehensive model with centimeter-level precision under complex lighting and motion conditions only by the passive marker of the target platform, greatly improving the spatial situation awareness accuracy and dynamic adaptability in the near-range mounting stage, and fundamentally overcoming the traditional single-sensor technical bottlenecks such as monocular vision drift, ranging blind area or inertial cumulative error, thereby laying a self-verifiable perception foundation for robust contact control of the unmanned driving system on a shaking platform.

[0018] Further, the preset visual marker is a specified two-dimensional code, and the acquisition of the image information of the target platform containing the preset visual marker through the image acquisition system of the specified unmanned aerial vehicle comprises:

[0019] determining a shooting height of the image acquisition system based on the size of the specified two-dimensional code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system and the field of view angle of the image acquisition system, the shooting height being the distance for accurately identifying the specified two-dimensional code in the image information;

[0020] The specified unmanned aerial vehicle is driven to a specified position based on the shooting height, and image information of the target platform containing the specified two-dimensional code is acquired by the image acquisition system.

[0021] It should be noted that the application can ensure that a complete, clear and non-distorted two-dimensional code image can be acquired at one time under different platform scales or environmental light interference by pre-calculating the quantitative relationship (size-pixel-field of view) of two-dimensional code recognizability and image system parameters, and actively controlling the unmanned aerial vehicle to accurately hover at a theoretically optimal recognition height, establishing a high-confidence data input basis for subsequent pose solving, avoiding recognition failure or coordinate solving drift caused by improper shooting distance in traditional visual positioning from the source, and significantly enhancing the robustness and system stability of the marker perception link.

[0022] Further, the shooting height of the image acquisition system is determined based on the size of the specified two-dimensional code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system, comprising:

[0023] The size of the specified two-dimensional code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system are input into the formula The shooting height of the image acquisition system is determined, wherein;

[0024] y is the actual distance corresponding to a single pixel, FOV is the field of view of the image acquisition system, DPI is the resolution of the image acquisition system, and h is the shooting height of the image acquisition system.

[0025] Further, the pose information of the specified unmanned aerial vehicle relative to the target platform is determined based on the image information of the preset visual marker, comprising:

[0026] The three-dimensional position coordinates and running attitude information of the specified unmanned aerial vehicle relative to the world coordinate system of the two-dimensional code are determined based on the PnP algorithm, and the running attitude information includes a rotation matrix and a translation vector.

[0027] It should be noted that the application can generate a six-degree-of-freedom pose reference (a rotation matrix maps a three-dimensional attitude, and a translation vector quantifies a spatial distance) strictly aligned with the physical space of the target platform in a dynamic environment by rigidly binding a two-dimensional code marker with a known physical size to the world coordinate system and using the PnP algorithm to solve the accurate geometric projection relationship between the optical center of the unmanned aerial vehicle and the marker in real time, completely replacing the manual estimation and correction of the unmanned aerial vehicle pitch angle, horizontal offset and contact distance by the operator, providing a mathematical abstract description that can directly drive a precise pose closed loop for the flight control system, and enabling the unmanned system to obtain spatial relationship recognition accuracy and anti-interference control capability beyond human beings.

[0028] Further, the motion information of the specified UAV relative to the target platform is acquired by an inertial measurement system of the specified UAV, including:

[0029] acquiring the last time motion information of the specified UAV relative to the target platform;

[0030] processing the last time motion information based on a pre-created kinematic model to predict the current time motion information of the specified UAV relative to the target platform.

[0031] It should be noted that the application inputs the real relative motion state at the last time into the pre-modeled UAV rigid body kinematic system to deduce the six-degree-of-freedom motion state prediction value (including linear acceleration, angular velocity and its derivative) at the current time in real time, so that the flight control system obtains the forward-looking motion trend judgment ability within the key time window before the actual measurement data output of the sensor, effectively compensates for the control lag caused by image processing delay and ranging sampling interval, forms the machine-level motion continuity guarantee with higher precision and faster response similar to the pre-judgment behavior of human operators, and significantly suppresses the attitude oscillation or trajectory deviation risk caused by information update delay in the dynamic mounting process.

[0032] Further, if the target platform is a dynamic target platform, the fine-grained position information of the target platform is determined based on the fusion information, including:

[0033] acquiring real-time motion state feedback information of the dynamic target platform;

[0034] determining the fine-grained position information of the target platform based on the fusion information and the real-time motion state feedback information.

[0035] It should be noted that the application introduces the high-frequency motion state feedback (such as acceleration / angular velocity in the carrier coordinate system) of the target platform in real time, and constructs an antagonistic coupling solution with the fusion information perceived by the UAV, generates a reference unified and spatiotemporal aligned relative pose stream in a dynamic environment, completely eliminates the coordinate system drift error and visual-inertial measurement false deviation caused by target motion, enables the UAV control system to obtain the spatial synchronization cognition ability for the shaking target, and realizes the full-autonomous servo mounting operation under the millisecond-level tracking precision, which has better stability and robustness than the manual compensation limit of the trajectory prediction of the dynamic platform by artificial operation.

[0036] Further, if the specified UAV is provided with multiple times, the specified UAV is mounted to the target platform based on the fine-grained position information, including:

[0037] The first unmanned aerial vehicle is used as a master node, and based on the fine-grained position information, the specified unmanned aerial vehicle is mounted to the target platform, and the mounting position information of the master node is released to the remaining unmanned aerial vehicles, so that the remaining unmanned aerial vehicles are mounted to the target platform based on the mounting position information.

[0038] It should be noted that the present application solidifies the first successfully mounted unmanned aerial vehicle as a master node carrying physical space true value, so that the fine-grained pose solution completed in the mechanical contact moment is upgraded to the origin anchoring process of the global reference coordinate system. The subsequent unmanned aerial vehicle only needs to directly inherit the millimeter-level pose template verified by the entity and convert to its own coordinate system, so as to realize atomic-level alignment of the group mounting pose, fundamentally solve the problems of cumulative pose deviation, motion trajectory conflict and mounting interface stress concentration caused by independent perception in multi-machine cooperative operation, and extend the single-machine-level precise mounting capability of the unmanned aerial vehicle cluster to unlimited scale without loss.

[0039] One or more embodiments of the present specification provide an unmanned aerial vehicle automatic mounting device, comprising:

[0040] at least one processor; and

[0041] a memory in communication connection with the at least one processor; wherein

[0042] The memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to:

[0043] acquire satellite signals of the target platform through the positioning system of the specified unmanned aerial vehicle;

[0044] determine coarse-grained position information of the target platform based on the satellite signals, so that the specified unmanned aerial vehicle approaches the target platform based on the coarse-grained position information;

[0045] acquire fusion information of the specified unmanned aerial vehicle relative to the target platform, the fusion information including pose information, height information and motion information;

[0046] determine fine-grained position information of the target platform based on the fusion information, so that the specified unmanned aerial vehicle is mounted to the target platform based on the fine-grained position information.

[0047] One or more embodiments of the present specification provide a non-volatile computer storage medium, which stores computer executable instructions, and the computer executable instructions can realize:

[0048] acquire satellite signals of the target platform through the positioning system of the specified unmanned aerial vehicle;

[0049] determine coarse-grained position information of the target platform based on the satellite signals, so that the specified UAV approaches the target platform based on the coarse-grained position information;

[0050] acquire fusion information of the specified UAV relative to the target platform, the fusion information including pose information, height information, and motion information;

[0051] determine fine-grained position information of the target platform based on the fusion information, so that the specified UAV is mounted to the target platform based on the fine-grained position information.

[0052] The above at least one technical solution adopted by the embodiments of the present specification can achieve the following beneficial effects:

[0053] The present application realizes full-process autonomy through a two-stage positioning mechanism from coarse to fine (coarse-grained satellite positioning guiding approach, fine-grained precise positioning control mounting based on fusion of multi-source information), fundamentally eliminates the necessity of manual remote control or monitoring, significantly reduces the risk of mounting failure and safety hazards caused by differences in operator skills, delays in judgment, or environmental interference, and enables the UAV to complete the mounting task stably and efficiently in a complex dynamic environment. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the embodiments of the present specification or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments described in the present specification, and other drawings can also be obtained according to these drawings without creative labor. In the drawings:

[0055] Figure 1 a flowchart of a UAV automatic mounting method provided for one or more embodiments of the present specification;

[0056] Figure 2 a flowchart of GPS coarse positioning provided for one or more embodiments of the present specification;

[0057] Figure 3 a coordinate system diagram provided for one or more embodiments of the present specification;

[0058] Figure 4 a flowchart of multi-modal data fusion provided for one or more embodiments of the present specification;

[0059] Figure 5 a structural diagram of a UAV automatic mounting device provided for one or more embodiments of the present specification. DETAILED DESCRIPTION

[0060] The embodiment of the specification provides a method and device for automatic mounting of a UAV and a medium.

[0061] In recent years, the technology of unmanned aerial vehicles (UAVs) has developed rapidly, and its application fields have expanded rapidly from traditional military reconnaissance, aerial photography and surveying to logistics distribution, infrastructure inspection, agricultural plant protection, emergency rescue, public safety and many other civilian fields. With the increasing richness of UAV application scenarios and the improvement of task complexity, higher requirements are put forward for the endurance, operation efficiency, autonomy and interaction ability with external systems of UAVs.

[0062] Many advanced application scenarios, such as long-haul patrol, large-scale logistics network, operation on mobile platforms (vehicles, ships, etc.), and tasks requiring frequent replacement of payloads or replenishment of energy, all face a core challenge: how to reliably, efficiently and autonomously connect (mount) and disconnect (detach) the UAV with the ground station, charging pile, mobile carrier or other target platform during or after task execution. Such physical connection may be used for:

[0063] Automatic charging / refueling: solving the endurance bottleneck of UAVs to achieve long-time or continuous operation.

[0064] Automatic data transmission: quickly download task data or upload new task instructions.

[0065] Automatic payload exchange: replace sensors, cameras, delivery objects, etc., to improve operation flexibility and efficiency.

[0066] Parking and recovery: safe parking and fixing on a specific platform (such as the top of a mobile vehicle, the deck of a ship, a fixed site).

[0067] Traditional solutions often rely on manual operation or use simple ground facilities. However, manual operation has high environmental requirements, low efficiency and safety risks; and simple ground facilities usually lack the ability to accurately dock with UAVs. Therefore, developing an automatic mounting and detaching device that can achieve high precision, high robustness and high autonomy between UAVs and target platforms has become one of the key technical bottlenecks for promoting the development of UAV applications to deeper and broader fields.

[0068] Currently, automatic mounting and detaching of UAVs mainly rely on manual operation, GPS navigation, visual identification or specific customized systems, but these methods have many shortcomings:

[0069] Manual operation depends on skills and is inefficient: it is highly dependent on operator experience and environmental conditions, difficult to perform stably in complex or harsh environments, has low success rate, and cannot achieve large-scale automation and unattended operation.

[0070] GPS precision deficiency or signal loss: The positioning precision provided by GPS (meter level) is far lower than the centimeter or even millimeter level required for physical docking; in indoor, underground, urban canyons, or areas with obstructions, the signal is weak or completely lost, and effective high-precision navigation cannot be provided.

[0071] Poor robustness and incomplete information in visual recognition: Methods based on visual identification are easily disturbed by environmental factors such as light changes, rain, snow, fog, and identification pollution or obstruction, leading to recognition failure or precision decline; they can usually only provide two-dimensional plane information, making it difficult to obtain accurate three-dimensional positions and complete attitude (6 degrees of freedom) information, which is insufficient to support physical docking that requires complex alignment.

[0072] Poor generalization of customized systems: Highly customized automated docking stations are usually only suitable for specific models of unmanned aerial vehicles or fixed platforms, with complex structures and high costs, making it difficult to flexibly adapt to diverse unmanned aerial vehicle models or dynamically changing target platforms, limiting deployment and promotion.

[0073] Although existing technologies provide some solutions for the landing and ground interaction of unmanned aerial vehicles, to achieve efficient, normalized, and all-weather automated operation of unmanned aerial vehicle systems, especially in application scenarios that require precise physical connection (mounting) and disconnection (separation) with target platforms, there are still many challenges and urgent needs:

[0074] (1) The need for high-precision positioning and attitude alignment: The positioning precision of existing technologies (such as GPS, relying on simple visual identification) is usually insufficient to support centimeter or even millimeter-level precise docking, and cannot meet the strict alignment requirements of complex physical mounting on position and attitude (including roll, pitch, and yaw).

[0075] (2) Environmental adaptability and robustness: Many existing solutions are easily affected by factors such as light changes, weather (rain, snow, fog), environmental obstructions, and identification pollution, leading to system failure or performance decline, and lack the ability to work stably in complex, dynamic, or harsh environments.

[0076] (3) Level of autonomy and intelligence: Existing solutions have limited automation and often require human intervention or rely on overly simplified docking mechanisms, making it difficult to achieve a completely autonomous and intelligent mounting and separation process, especially in handling unexpected situations.

[0077] (4) Generalization and flexibility: Most automated platforms or docking systems are customized for specific models of unmanned aerial vehicles or specific application scenarios, with poor generalization and difficulty in quickly adapting to different types of unmanned aerial vehicles or diverse target platforms (fixed, mobile, different shapes).

[0078] Therefore, there is an urgent need in the field for an automatic mounting and dismounting technology that can overcome the above-mentioned technical challenges and achieve high precision, high robustness, high autonomy, high versatility and high security between UAVs and target platforms.

[0079] With the rapid development of drone technology and the continuous expansion of its application fields, drones have been widely used in aerial photography, inspection, logistics, surveying, security, and many other areas. However, drones typically need to carry different equipment to complete tasks. The efficiency and degree of automation of these operations directly affect the overall performance and routine operation capability of the drone system. In existing technologies, the mounting and detachment of drones mainly rely on the following methods:

[0080] (1) The operator (pilot) uses a ground remote controller and relies on their visual observation (either by eye or through images transmitted from a first-person view camera) and operational experience to precisely control the drone's flight attitude (pitch, roll, yaw) and position (forward / backward, left / right, up / down). The operator needs to concentrate highly to judge the relative position and attitude deviation between the drone and the docking point on the target platform, and continuously fine-tune the control commands to make the drone slowly and smoothly approach the target point until the docking mechanism on the drone makes physical contact with the docking mechanism on the platform and completes the connection (such as a latch, magnetic attraction, plug insertion, etc.). The separation process is the reverse operation, where the drone is manually controlled to unlock the docking mechanism and take off safely. This is the most basic and direct method.

[0081] (2) GPS-based autonomous approach: The UAV uses its built-in GPS receiver to obtain its own global positioning information (latitude, longitude, and altitude). The location of the target platform is known in advance, usually input into the UAV's flight control system or mission planning in the form of GPS coordinates. The UAV calculates the deviation between its current GPS position and the target platform's GPS position, uses an autonomous navigation algorithm to control its flight, gradually reducing the deviation, and flies towards the target's GPS coordinates. This is typically used to guide the UAV to a position approximately above the target area.

[0082] (3) Visual Identification-Based Assisted or Autonomous Docking: One or more visual identifiers with unique patterns (e.g., a black and white checkerboard, a specific QR code, or a simple H-shaped mark) are pre-placed on the ground or walls near the target platform. The UAV is equipped with a downward-facing or forward-facing camera. As the UAV approaches the target platform, the camera captures an image containing the identifier. An onboard or ground computer runs an image processing algorithm to identify the identifier in the image and calculates the camera's (and thus the UAV's) two-dimensional planar position (offset on the X and Y axes) and yaw angle relative to the identifier based on the identifier's known size and perspective distortion in the image. This relative position and angle information is fed back to the UAV's flight control system to adjust the UAV's flight trajectory and attitude in real time, ensuring it flies precisely to the center of the identifier and aligns with the yaw angle.

[0083] (4) Dedicated automated docking station or interaction platform: This is a highly integrated system, typically a fixed ground station, charging station, or a customized platform mounted on a specific vehicle (such as the top of a logistics vehicle). This type of system integrates multiple sensors (potentially including high-precision vision, lidar, ultra-wideband, or even RTK-GPS as auxiliary sensors), precise navigation and positioning algorithms, and specially designed physical docking mechanisms (such as guide rails, robotic arms, customized charging interfaces, data interfaces, or payload grabbing / releasing mechanisms). The system can detect the arrival of the drone, guide it into the designated docking area, use its high-precision sensors (such as near-field vision or LiDAR) for centimeter-level or even millimeter-level precise relative positioning and attitude adjustment, and then activate the physical docking mechanism to complete the connection (such as guiding the drone into the charging slot, the robotic arm grabbing and placing the payload, and automatic plug insertion). The separation process is controlled by the system to unlock and guide the drone to leave safely.

[0084] In summary, while these existing solutions have played a role in their respective fields, none can fully meet the requirements for high-precision, robust, and versatile automatic attachment and detachment between UAVs and diverse target platforms in complex environments. This is precisely the core problem that this project aims to solve by integrating innovative technologies.

[0085] In achieving automated mounting and dismounting between UAVs and target platforms, existing technologies mainly include manual control, GPS-based coarse approach, visual identification-based assisted docking, and highly customized dedicated automated docking stations. However, these solutions generally suffer from significant shortcomings: they lack the centimeter-level or even millimeter-level high-precision positioning and attitude alignment capabilities required for physical connection; they have poor robustness to environmental changes (such as lighting, weather, and obstruction), making them susceptible to interference and failure; their versatility and flexibility are limited, making them difficult to adapt to different UAV models or diverse target platforms; and they often require highly customized facilities or advanced human skills. Therefore, existing technologies struggle to meet the reliability, autonomy, and efficiency requirements of automated mounting and dismounting processes for UAVs in complex and dynamic environments. Specifically, the disadvantages of existing technologies include:

[0086] 1. Manual control of mounting and dismounting:

[0087] It demands extremely high precision control skills from operators, especially in complex environments such as windy, swaying, or poorly lit conditions. Achieving centimeter-level or even millimeter-level precise alignment and connection is extremely difficult, easily leading to docking failure, equipment damage, or even accidents. It is inefficient, time-consuming, and cannot support large-scale, high-frequency automated operations. It also poses high safety risks, with a high risk of collisions.

[0088] 2. GPS-based autonomous approach:

[0089] GPS positioning accuracy is typically only at the meter level, far from sufficient to support physical attachment and detachment operations between drones and target platforms (such as charging stations, mobile vehicle rooftops, and aerial refueling points) that require centimeter-level or even millimeter-level precision alignment. It cannot provide effective approach navigation in indoor environments with weak or lost GPS signals, urban canyons, or areas subject to interference. Furthermore, it cannot obtain the drone's attitude (roll, pitch, yaw) information relative to the target platform, while precise attitude is crucial for physical docking.

[0090] 3. Visual identifier-based assisted or autonomous integration:

[0091] It heavily relies on the clarity and visibility of visual identifiers, and is susceptible to environmental factors such as changes in lighting, shadows, dirt, partial occlusion, rain, and snow, leading to recognition failures or decreased positioning accuracy and poor robustness. It typically struggles to provide precise 3D position and complete attitude (roll, pitch, yaw) information, insufficient to support the mounting or dismounting of complex physical connection mechanisms requiring precise multi-degree-of-freedom alignment. For mobile target platforms, additional complex algorithms are needed for real-time tracking and prediction, increasing system complexity. Furthermore, it requires specific visual identifiers for the target platform, limiting its versatility.

[0092] 4. Dedicated automated docking station or interactive platform:

[0093] These systems are typically highly customized, lack versatility, and are difficult to adapt to different models of UAVs or different types of target platforms (such as mobile vehicle tops, aerial refueling points, and non-standard ground stations). Their docking mechanisms are often fixed, making it difficult to cope with real-time changes in the target platform's position and attitude or environmental interference. Development and deployment costs are high, and flexibility is limited.

[0094] In summary, existing technologies generally suffer from several key drawbacks in achieving automatic attachment and separation between UAVs and target platforms. These include a lack of centimeter / millimeter-level high-precision positioning and attitude alignment capabilities, poor environmental robustness, limited versatility and flexibility, and insufficient autonomy. Consequently, they are unable to meet the demands for efficient and reliable operations in complex and dynamic environments.

[0095] The main objective of this invention is to provide a technology that overcomes the aforementioned shortcomings of existing technologies and achieves high-precision, high-robustness, high-autonomy, high-versatility, and high-safety automatic attachment and detachment between UAVs and target platforms. Specifically, this invention aims to achieve the following main objectives:

[0096] 1. Achieve high-precision positioning and attitude alignment:

[0097] This invention aims to propose a technical solution that overcomes the limitations of existing technologies in terms of precision, achieving centimeter-level or even sub-centimeter-level high-precision relative positioning and attitude (including position and complete attitude angles) alignment of the UAV body or its docking mechanism with respect to the docking point or marker on the target platform. This means that the system can accurately calculate and control the UAV's six degrees of freedom in space, enabling it to stably and accurately align its own docking mechanism with the target platform's docking mechanism and complete the physical connection, thereby ensuring the success rate and safety of docking and meeting the needs of various complex physical docking mechanisms.

[0098] 2. Improve environmental adaptability and robustness:

[0099] This invention aims to propose an automated mounting and dismounting scheme with higher environmental adaptability and robustness. By employing multi-sensor fusion technology (e.g., combining vision, lidar, IMU, and other sensing methods), or utilizing sensing technologies insensitive to environmental changes, the system's dependence on a single sensor or specific environmental conditions is reduced. The goal is to enable the UAV to maintain stable and accurate relative positioning and attitude awareness in complex, dynamic, and even harsh environments (such as insufficient light, wind, slight target platform sway, and the presence of obstructions or interference), thereby ensuring the success rate and safety of the automated docking process and achieving reliable operation in all weather and all scenarios.

[0100] 3. Enhance autonomy and intelligence levels:

[0101] This invention aims to significantly improve the autonomy and intelligence of the automatic loading and unloading process of unmanned aerial vehicles (UAVs). The goal is to enable UAVs to autonomously complete the entire process after receiving loading / unloading commands, including autonomous navigation to the target area, autonomous search and identification of the target platform, autonomous high-precision relative positioning and attitude alignment, autonomous execution of physical docking / disconnection operations, and autonomous and safe departure upon completion. This requires the system to possess intelligent capabilities such as environmental perception, status assessment, path planning, decision control, and anomaly handling, minimizing or eliminating reliance on human intervention, improving operational efficiency, and supporting the automated deployment and operation of large-scale, distributed UAV systems.

[0102] 4. Enhance versatility and flexibility:

[0103] This invention aims to propose an automated mounting and dismounting scheme with greater versatility and flexibility. The goal is to design a universal docking mechanism that can adapt to different types of UAVs (possibly through software configuration or a few compatible parts) and diverse target platforms (including fixed and mobile platforms, which may differ in shape and size, and require little or no modification). By improving the system's versatility, development, deployment, and maintenance costs can be significantly reduced, expanding the application scope and enabling the automated UAV mounting and dismounting technology to be more widely applied in various scenarios, thereby improving the overall system's economy and practicality.

[0104] This invention aims to solve the following technical problems existing in the automatic attachment and detachment of existing drones from target platforms (such as robots, charging stations, vehicle tops, aerial refueling platforms, etc.) after mission execution:

[0105] Relying on manual operation is inefficient and poses safety hazards: Existing drones typically require manual intervention after completing their missions for operations such as landing, charging, or changing payloads. This manual operation is not only inefficient and increases the operation cycle, but also poses high safety risks in complex environments, at night, or in severe weather conditions due to limited visibility and increased operational difficulty, making it difficult to guarantee the safety of personnel and equipment.

[0106] Manual control lacks precision and stability, making accurate docking difficult. This is especially true in applications requiring precise docking between drones and mobile or fixed platforms (such as autonomous landing on top of mobile vehicles, docking with charging stations, and aerial refueling), where extremely high demands are placed on the precision and stability of manual control. Existing methods relying on manual or simple control cannot achieve centimeter-level or even millimeter-level high-precision alignment and stable docking, resulting in a high docking failure rate and failing to meet mission requirements.

[0107] The current technology struggles to support the automation and routine operation requirements of drones: Due to the aforementioned reliance on manual labor and insufficient docking accuracy, existing technologies cannot achieve fully autonomous operation of drones, including autonomous return to home, autonomous docking with platforms, and autonomous completion of subsequent operations (such as charging and payload replacement). This limits the routine and large-scale application of drones in fields such as industrial inspection, logistics delivery, and emergency rescue, and increases operating costs.

[0108] Insufficient robustness in complex environments: Existing docking methods relying on single sensors or simple controls suffer significant reliability degradation in positioning, navigation, and control under complex real-world conditions (such as varying lighting, partial obstruction, interference, and target platform movement), impacting docking success rate and safety. This invention employs multi-sensor fusion technology (vision, inertial measurement unit (IMU), lidar, Global Navigation Satellite System (GPS), etc.) to improve the system's perception and positioning robustness under these complex conditions. This ensures stable system operation even with limited sensor information, reliably addressing the aforementioned accuracy, safety, and automation issues.

[0109] Therefore, it can be said that the present invention mainly solves the problems of reliance on manual labor, low efficiency, safety risks, insufficient docking accuracy / stability, and difficulty in automation and normalization, and ensures that the solution to these problems is reliable in real complex environments by improving robustness.

[0110] In summary, the main technical problems solved by this invention cover multiple key dimensions such as efficiency, safety, accuracy, degree of automation, and reliability in complex environments, rather than just the aforementioned points. Robustness is one of the key technical challenges in achieving high-precision, safe, and automated docking.

[0111] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0112] Figure 1 This diagram illustrates a process flow for an automatic drone mounting method provided in one or more embodiments of this specification. This process can be executed by an automatic drone mounting system. Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0113] The method flow steps of the embodiments in this specification are as follows:

[0114] S101 acquires satellite signals from the target platform through the positioning system of the designated UAV.

[0115] In the embodiments described in this specification, the global navigation satellite system receiving module mounted on the designated UAV is activated to continuously monitor the positioning signal broadcast by the satellite signal transmitter installed on the target platform. When the target platform is within the satellite signal coverage area, the satellite signal is received and decoded in real time via a wireless communication link.

[0116] S102, determine the coarse-grained position information of the target platform based on the satellite signal, so that the designated UAV can approach the target platform based on the coarse-grained position information.

[0117] In the embodiments described in this specification, the satellite signal obtained in S101 is input into the UAV flight control system to calculate the two-dimensional planar coordinates (longitude and latitude) and altitude estimate of the target platform in the global coordinate system, generating coarse-grained position information with an accuracy of meters. Based on this position point, the UAV autonomously plans its flight path and controls its propulsion system to fly to a hovering position in an airspace centered on the target platform with a radius less than a set threshold (e.g., 20 meters).

[0118] S103, obtain the fusion information of the designated UAV relative to the target platform, the fusion information including pose information, altitude information and motion information.

[0119] In the embodiments described in this specification, multi-sensor synchronous data acquisition is enabled at the hovering position:

[0120] Pose information acquisition: The vision camera on the drone is activated to capture a preset QR code mark on the target platform. The six-degree-of-freedom pose (three-dimensional position + three-axis attitude) of the drone relative to the QR code coordinate system is output through machine vision algorithms (such as PnP calculation).

[0121] Altitude information acquisition: Activate the laser rangefinder to emit a beam vertically downwards and measure the absolute altitude of the UAV to the surface of the target platform.

[0122] Motion information acquisition: Read the body's three-axis acceleration and angular velocity output in real time from the inertial measurement unit (IMU).

[0123] The above information is aligned by timestamp and encapsulated into a fusion information packet, which is then updated to the flight control system at a fixed frequency.

[0124] S104, determine the fine-grained position information of the target platform based on the fused information, so as to complete the mounting of the designated UAV to the target platform based on the fine-grained position information.

[0125] In the embodiments described in this specification, the fused information from S103 is input into the positioning engine. Through spatial coordinate transformation and motion compensation algorithms (such as Kalman filtering), the millimeter-level fine-grained position (including attitude tolerance) of the target platform mounting point in the UAV body coordinate system is calculated. The flight control system generates three-dimensional spatial trajectory control commands, driving the power system and servos to descend along a precision-optimized path until the UAV mounting mechanism and the target platform mechanical interface complete physical engagement (such as electromagnetic adsorption locking), without any manual intervention throughout the process.

[0126] It should be noted that this application achieves full-process autonomy through a two-stage positioning mechanism from coarse to fine (coarse-grained satellite positioning to guide approach, and fine-grained precise positioning and control of mounting by fusing multi-source information). This fundamentally eliminates the need for manual remote control or monitoring, significantly reduces the risk of mounting failure and safety hazards caused by differences in operator skills, judgment delays, or environmental interference, and enables UAVs to complete mounting tasks stably and efficiently even in complex dynamic environments.

[0127] Furthermore, when acquiring the fusion information of the target platform, the image acquisition system of the designated UAV can be used to acquire image information of the target platform containing preset visual markers; based on the image information of the preset visual markers, the pose information of the designated UAV relative to the target platform can be determined, the pose information including the position information and running attitude information of the designated UAV relative to the target platform; the altitude information of the designated UAV relative to the target platform can be acquired through the ranging system of the designated UAV; and the motion information of the designated UAV relative to the target platform can be acquired through the inertial measurement system of the designated UAV.

[0128] It should be noted that the above content can be implemented through the following specific implementation plan:

[0129] Visual marker image acquisition and pose calculation:

[0130] The global shutter camera mounted on the underside of the designated drone is activated to acquire real-time images of the QR code marker affixed to the top of the target platform at a preset detection altitude (pre-calculated based on the physical size of the QR code and camera parameters). The captured images undergo distortion correction and grayscale preprocessing to extract the pixel coordinates of the four corner vertices of the QR code. Combining the preset physical size of the QR code (e.g., side length) with the camera's intrinsic parameters (focal length, optical center), a perspective geometry algorithm is used to output six-DOF pose information.

[0131] Location information: The three-dimensional coordinates (X, Y, Z axis offsets) of the UAV's optical center in the QR code coordinate system;

[0132] Attitude information: The rotation matrix (pitch angle, roll angle, yaw angle) of the UAV body coordinate system relative to the QR code coordinate system.

[0133] Accurate measurement of altitude information: The laser ranging modules on both sides of the UAV landing gear are activated simultaneously to emit modulated laser beams vertically toward the surface of the target platform; the reflected light signals are received and the time of flight (ToF) is calculated, and the absolute altitude value (H) is calculated by combining atmospheric compensation parameters; the altitude data is aligned with the pose information timestamp as a redundancy verification benchmark for the Z-axis position.

[0134] Inertial motion state capture: Reading the raw data stream from the fuselage's core IMU module (three-axis gyroscope + three-axis accelerometer); solving in real time using the fuselage kinematic model (rigid body rotation and translation constraints):

[0135] Angular velocity: The instantaneous rotational rate of the body about the X / Y / Z axes;

[0136] Linear acceleration: The instantaneous acceleration vector of an organism in three dimensions (including the gravitational component).

[0137] Using the target platform as the motion reference frame, the relative motion vector (velocity, acceleration) of the UAV relative to the platform is output through coordinate transformation.

[0138] Information fusion encapsulation and timing synchronization: A unified time stamp (μs-level accuracy) is applied to pose information, altitude information, and motion information; the three types of data are encapsulated into a structured fusion information packet at a fixed frequency (e.g., 50Hz), including: position information (X, Y, Z); attitude information (rotation matrix); altitude information (H); and motion information (three-axis angular velocity, three-axis acceleration), which is transmitted to the flight control system's fusion positioning engine via a high-speed serial bus.

[0139] It should be noted that this application, through the fusion of three sensing technologies—visual marker positioning, active ranging, and inertial motion measurement—enables the UAV to autonomously generate a three-dimensional relative pose-motion integrated model with centimeter-level accuracy under complex lighting and motion conditions, requiring only passive markers preset by the target platform. This significantly improves the accuracy of spatial situational awareness and dynamic adaptability during the short-range mounting phase, fundamentally overcoming the bottlenecks of traditional single-sensor technologies such as monocular visual drift, ranging blind spots, or inertial cumulative errors. This lays a self-verifiable perception foundation for robust contact control of unmanned driving systems on swaying platforms.

[0140] Furthermore, the preset visual marker is a designated QR code. When acquiring image information of the target platform containing the preset visual marker through the image acquisition system of the designated UAV, the shooting height of the image acquisition system can be determined based on the size of the designated QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system. The shooting height is the distance for accurately identifying the designated QR code in the image information. Based on the shooting height, the designated UAV is driven to a designated position, and the image information of the target platform containing the designated QR code is acquired through the image acquisition system.

[0141] It should be noted that the above content can be implemented through the following specific implementation plan:

[0142] Shooting height calculation: Retrieve the pre-stored physical parameters (actual side length and size) of the target platform QR code; combine the inherent properties of the image acquisition system (resolution, field of view) and the single-pixel imaging size model (i.e., the actual physical distance corresponding to 1 pixel in the image); based on the preset minimum effective recognition pixel ratio of the QR code (e.g., the mark needs to occupy 20% of the image width), calculate the theoretical minimum shooting height to ensure complete recognition and the maximum shooting height to avoid distortion; take the median value of the height range as the optimal shooting height.

[0143] Precise altitude hovering control: The calculated optimal shooting altitude is input into the altitude control loop of the UAV flight control system; the power system is activated for vertical ascent and descent, and the altitude deviation is fed back in real time through the barometer and laser rangefinder; the UAV is controlled to move along the vertical axis until the relative altitude with the target platform surface is constant and equal to the optimal shooting altitude.

[0144] Horizontal position maintenance: Simultaneously read visual positioning data during ascent and descent to lock the drone's horizontal position directly above the target platform (XY plane deviation < 0.5 meters); trigger position holding mode to resist wind disturbance and maintain hovering.

[0145] Image acquisition execution: After confirming that the UAV meets both the altitude tolerance (±5%) and the horizontal position tolerance, activate the image acquisition system; adjust the camera pitch angle so that the optical axis is perpendicular to the QR code plane, and acquire at least 3 frames of RAW format images under constant lighting parameters; output the image information stream with timestamps to the vision processing module.

[0146] It should be noted that this application pre-calculates the quantitative relationship between QR code recognizability and image system parameters (size-pixel-field of view), and actively controls the drone to hover precisely at the theoretically optimal recognition height. This ensures that complete, clear, and distortion-free QR code images can be acquired in one go under different platform scales or ambient light interference. This establishes a high-confidence data input foundation for subsequent pose calculation, avoiding recognition failure or coordinate calculation drift caused by improper shooting distance in traditional visual positioning from the source. This significantly enhances the robustness and system stability of the marker perception process.

[0147] Furthermore, when determining the shooting height of the image acquisition system based on the size of the specified QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system, the size of the specified QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system can be input into the formula. Determine the shooting height of the image acquisition system, where y is the actual distance corresponding to a single pixel, FOV is the field of view of the image acquisition system, DPI is the resolution of the image acquisition system, and h is the shooting height of the image acquisition system.

[0148] Furthermore, when determining the pose information of the designated UAV relative to the target platform based on the image information of the preset visual marker, the three-dimensional position coordinates and running posture information of the designated UAV relative to the world coordinate system of the QR code can be determined based on the PnP algorithm. The running posture information includes rotation matrix and translation vector.

[0149] It should be noted that this application rigidly binds a QR code marker with known physical dimensions to the world coordinate system and uses the PnP algorithm to calculate the precise geometric projection relationship between the UAV's optical center and the marker in real time. In a dynamic environment, it autonomously generates a six-degree-of-freedom pose reference (rotation matrix maps three-dimensional attitude, translation vector quantizes spatial spacing) that is strictly aligned with the physical space of the target platform. This completely replaces the operator's visual estimation and manual correction of the UAV's pitch angle, horizontal offset, and contact distance. It provides a mathematical abstract description for the flight control system that can directly drive a precise pose closed loop, enabling the unmanned system to achieve spatial relationship cognition accuracy and anti-interference control capabilities that surpass those of humans.

[0150] Furthermore, when acquiring the motion information of the designated UAV relative to the target platform through the inertial measurement system of the designated UAV, the motion information of the designated UAV relative to the target platform at the previous moment can be acquired first; the motion information at the previous moment can be processed based on a pre-created kinematic model to predict the current motion information of the designated UAV relative to the target platform.

[0151] It should be noted that this application inputs the actual relative motion state of the previous moment into the pre-modeled UAV rigid body kinematics system to deduce the predicted value of the six degrees of freedom motion state (including linear acceleration, angular velocity and their derivatives) in real time. This enables the flight control system to obtain the ability to predict the motion trend in the critical time window before the actual measurement data of the sensors is output. This effectively compensates for the control lag caused by image processing delay and ranging sampling interval, forming a machine-level motion continuity guarantee that is similar to the predictive behavior of human operators but with higher accuracy and faster response. It significantly suppresses the risk of attitude oscillation or trajectory deviation caused by information update delay during dynamic loading.

[0152] Furthermore, if the target platform is a dynamic target platform, when determining the fine-grained position information of the target platform based on the fused information, the real-time motion state feedback information of the dynamic target platform can be obtained first; the fine-grained position information of the target platform can be determined based on the fused information and the real-time motion state feedback information.

[0153] It should be noted that the above content can be implemented through the following specific implementation plan:

[0154] Dynamic platform motion status request: When the fused information detects that the displacement of the target platform continues to exceed the limit (such as speed > threshold), a wireless communication request to the dynamic target platform is triggered; a point-to-point data link is established, and the target platform is instructed to send back its real-time motion status raw data stream.

[0155] Platform motion data synchronous reception: Real-time reception of raw sensor data carried by the target platform through a low-latency communication module (such as UWB), including: three-axis acceleration and angular velocity of the platform carrier coordinate system (from the platform IMU), global position correction value output by the platform positioning module (such as GNSS differential data), and adding UAV-end reception timestamp to the data packet.

[0156] Forced alignment of dual-end spatiotemporal references: Based on a pre-synchronized atomic clock source, the transmission delay deviation between the timestamp in the platform data packet and the local clock of the UAV is corrected; the motion parameters of the platform carrier coordinate system are mapped to the UAV visual positioning reference system (such as the origin of the QR code coordinate system) through the coordinate system rotation matrix.

[0157] Dynamic coupling error filtering: The spatiotemporally aligned platform motion data is input into the motion decoupler. The platform acceleration / angular velocity is used as input to calculate the projection compensation amount of the entrained motion on the visual pose. The global position drift in the fusion information is calibrated based on the platform GNSS correction value. The corrected anti-disturbance fusion information is output.

[0158] Fine-grained position calculation activation: The corrected fusion information is input into the fine-grained positioning engine; based on the platform-induced motion compensation results and the body fusion data, the real-time pose of the mount point with millimeter-level accuracy is generated; additional motion trend prediction output is added to guide flight control pre-compensation.

[0159] It should be noted that this application introduces the high-frequency motion state feedback of the target platform itself in real time (such as acceleration / angular velocity in the carrier coordinate system) and constructs an adversarial coupling solution with the fused information perceived by the UAV. In the dynamic environment, a relative pose flow with unified reference and spatiotemporal alignment is generated, which completely eliminates the coordinate system drift error and visual-inertial measurement false deviation caused by the target motion. This enables the UAV control system to obtain the ability to spatially synchronize with the swaying target and achieve fully autonomous servo-mounted operation with millisecond-level tracking accuracy. Its stability and robustness exceed the limits of manual operation for trajectory prediction and manual compensation of dynamic platforms.

[0160] Furthermore, if multiple designated drones are configured, during the process of mounting the designated drones to the target platform based on the fine-grained position information, the first drone is used as the master node. The designated drones are mounted to the target platform based on the fine-grained position information, and the mounting position information of the master node is released to the other drones so that the other drones can be mounted to the target platform based on the mounting position information.

[0161] It should be noted that the above content can be implemented through the following specific implementation plan:

[0162] Master node qualification activation: When multiple drones enter the pre-load airspace, the first drone that arrives directly above the target platform is selected as the candidate master node through the airborne decision module; the candidate master node independently executes standard fine-grained position calculation and loading actions until the mechanical lock completes physical engagement.

[0163] Mounting position information encapsulation: At the moment the master node is successfully mounted, a high-precision positioning module (such as a laser tracker) is triggered to measure in real time the precise spatial relationship (6DoF pose) between the master node's body coordinate system and the target platform's mounting interface; this spatial relationship and the dynamic compensation parameters (such as vibration tolerance) during mounting are encapsulated into a structured position information packet.

[0164] Cluster broadcasting and receiving: The master node broadcasts a data frame containing location information through a wireless ad hoc network; the remaining drones (slave nodes) receive and parse the data frame while maintaining safe hovering, extracting two core parameters: the global coordinates of the mount point and the tolerance range of the mechanical interface.

[0165] From node coordinate transformation: Each slave node, based on its current position, calculates the relative spatial vector from the host machine to the mount point; loads the mechanical tolerance parameters shared by the master node, and generates customized trajectory control commands for the host machine.

[0166] Collaborative mounting execution: The nodes descend sequentially along the planned path. Before contacting the target platform: the local sensing data is compared with the master node's pose template in real time; the attitude is finely adjusted to physically align the mounting mechanism with the interface; the locking device is triggered to complete the mounting, and the status is fed back to the cluster.

[0167] It should be noted that this application solidifies the first successfully mounted UAV as the master node carrying the true value of the physical space, thereby elevating the fine-grained pose calculation completed at the moment of mechanical contact to the origin anchoring process of the global reference coordinate system. Subsequent UAVs only need to directly inherit the millimeter-level pose template verified by the entity and transform it to their own coordinate system to achieve atomic-level alignment of the group's mounted pose. This fundamentally solves the problems of cumulative pose deviation, motion trajectory conflict and stress concentration at the mounting interface caused by independent perception in multi-UAV collaborative operations, enabling the UAV swarm to extend the single-UAV-level precise mounting capability to an infinite scale of lossless replication.

[0168] It should be noted that the attachment and detachment of UAVs from target platforms (such as charging stations, vehicles, and aerial platforms) is a crucial step in achieving autonomous UAV operations and expanding their application scenarios. Accurate and efficient docking helps improve mission execution efficiency, reduce operating costs, and ensure operational safety in complex environments. Current UAV docking with target platforms typically relies on manual operation, which suffers from low efficiency, high safety risks, and difficulty in guaranteeing accuracy, especially under harsh conditions or when docking with mobile platforms. To address these shortcomings, this invention proposes an automatic UAV attachment / detachment device and system. First, coarse positioning is performed using satellite signals such as GPS to guide the UAV closer to the target platform area. Then, a fine positioning stage based on visual recognition is initiated, where the onboard camera captures the target (e.g., a QR code), and the PnP algorithm is used to calculate the high-precision pose of the UAV relative to the target platform. To further improve the accuracy and robustness of the final docking, the solution employs multi-sensor data fusion, integrating data from various sensors such as cameras, IMUs, and LiDAR (or point cloud matching), and utilizes algorithms such as Kalman filtering or graph-optimized SLAM to achieve high-precision real-time pose estimation of the UAV relative to the target platform. Based on precise pose information, the flight control system performs accurate adjustments to guide the UAV to align with the mechanism on the target platform. Automatic mounting or dismounting is accomplished using a locking system, which can be achieved through the following implementation scheme:

[0169] 1. Coarse GPS signal positioning

[0170] GPS coarse positioning uses Global Positioning System (GPS) satellite signals to determine the approximate geographical location of a drone or aircraft. It typically provides positioning accuracy in the range of a few meters to tens of meters, sufficient to guide the aircraft to the vicinity of the target area. In automated docking or landing schemes, GPS coarse positioning serves as an initial navigation step, used to approach the target over a wide area, paving the way for subsequent, more precise positioning methods (such as vision, laser, etc.).

[0171] 2. Precise positioning based on visual recognition

[0172] The drone uses its onboard camera to photograph pre-defined visual markers on the target platform. Through image processing and algorithms such as Perspective-n-Point (PnP), the drone's real-time, high-precision three-dimensional pose (position and attitude) relative to the target platform is accurately calculated. This process provides relative positioning information at the centimeter or even millimeter level.

[0173] 3. Multimodal data fusion

[0174] Multi-sensor data fusion integrates data acquired from different sensors on a UAV (such as GPS and IMU). Its purpose is to combine the strengths of each sensor and overcome the limitations of a single sensor, such as the low accuracy of GPS or the susceptibility of vision to environmental influences. Through fusion algorithms (such as Kalman filtering), more accurate, stable, and reliable real-time pose (position and attitude) estimates for the UAV can be generated than any single sensor.

[0175] 4. Locking system

[0176] The robust locking system builds upon this foundation, providing a strong, reliable, and status-clear physical connection. Its design prioritizes ensuring a secure connection that can withstand various external loads and environmental changes without requiring custom designs for each drone or platform (simply ensuring interface compatibility is sufficient).

[0177] This solution employs a combination of multi-sensor data fusion (including GPS, vision, IMU, etc.) and high-precision visual recognition to achieve automatic and precise attachment and separation between the UAV and the target platform. Compared to traditional manual operation or methods relying on a single sensor (such as coarse GPS positioning), this solution significantly improves the automation level and success rate of docking, achieving relative positioning accuracy at the centimeter or even millimeter level, ensuring high real-time performance and robustness in the attachment and separation process. By fusing data from different sensors, it overcomes the limitations of single sensors in complex environments (such as low light or no GPS signal), effectively avoiding positioning drift and error accumulation. This method dynamically fuses sensor information through intelligent algorithms, adapting to different lighting conditions, backgrounds, and platform motion states, ensuring stable docking in various complex scenarios. This provides accurate and reliable automatic attachment and separation capabilities for UAVs in autonomous operations, material delivery, and equipment maintenance in complex dynamic environments.

[0178] The implementation of this solution will effectively improve the efficiency and reliability of autonomous mission execution by drones, providing key technical support for future scenarios such as drone logistics, infrastructure inspection, and mobile charging pile docking. Through the collaborative work of visual recognition and multi-sensor fusion, this solution can not only handle precise docking with static platforms but also adapt to mobile platforms or disturbed environments, significantly improving the autonomous operation capability and mission success rate of drones in complex dynamic environments, and providing strong support for the widespread application and intelligent development of drone systems.

[0179] The positioning process of this patent is explained in detail below:

[0180] 1. Coarse GPS signal positioning

[0181] GPS coarse positioning is a key technology for the system in the initial stage of a mission and for long-distance navigation. It uses GPS satellite signals to calculate the approximate position of the UAV in the global coordinate system, helping the UAV navigate from its current position to the approximate area or airspace of the target docking platform. Figure 2This is a flowchart of GPS coarse positioning. Here, "coarse" means that its positioning accuracy (usually meter-level) is insufficient to directly achieve the final centimeter or millimeter-level precise physical docking. By analyzing the GPS satellite signals received by the UAV's GPS terminal, when the UAV receives a sufficient number of satellite signals (usually greater than or equal to 4), it can measure the pseudorange with each satellite. The UAV terminal typically receives multiple satellite signals. To accurately calculate the UAV's position in three-dimensional space and the receiver clock deviation, the pseudorange information of these satellites needs to be used for calculation. Because low-elevation satellite signals are unstable due to atmospheric delay, multipath effects, and observation noise, and contribute little to positioning accuracy, to improve positioning accuracy and stability, low-elevation satellites are usually filtered out based on a set elevation angle threshold (generally between 5° and 15°), and satellites with elevation angles higher than the threshold are considered valid satellites. Then, using the ephemeris data (position information) of these valid satellites and the measured pseudorange, the three-dimensional coordinates are solved according to Formula 1, where (x... s ,y s ,z s (x, y, z) represents the satellite coordinates, (x, y, z) represents the UAV coordinates, δt represents the clock offset, and ρ represents the pseudorange. This allows for the acquisition of rough positional information at the meter level.

[0182] 2. Precise positioning based on visual recognition

[0183] After the drone reaches the vicinity of the target platform (within meter-level accuracy) via GPS coarse positioning, the system switches to visual recognition-based precise positioning mode to complete centimeter- or even millimeter-level precise mounting / dismounting tasks. First, we need to determine the conditions and distances required for precise positioning. The core is ensuring the camera can clearly and effectively capture the QR code on the instrument for high-precision pose calculation. We need to determine at what height the QR code is completely within the camera's field of view while occupying enough pixels to meet the subsequent high-precision positioning requirements. Positioning is achieved using nested QR codes of varying sizes, ensuring the QR code is within the field of view while maximizing the number of pixels it occupies to improve navigation accuracy. Influencing factors include: camera shooting height h, camera field of view (FOV), image resolution (DPI), and QR code side length w. When the camera is shooting at a height h above the object, its FOV determines the actual width it can cover. This actual width divided by the image's pixel resolution gives the actual distance represented by each pixel.

[0184] First, calculate the actual width (or height) y1 that the camera can cover at a distance h.

[0185] At a distance h, the total width y1 visible to the camera can be calculated using trigonometric functions:

[0186]

[0187] therefore, Given camera parameters, what is the area of ​​the ground that the aircraft can cover when viewed from different altitudes? To ensure that the QR code is always within the field of view, y1 must be greater than or equal to the maximum size (or diagonal length) of the QR code.

[0188] Calculate the actual distance y corresponding to a single pixel:

[0189] If the image resolution is DPI, then:

[0190] Integration formula: Therefore, y is the actual world distance corresponding to a single pixel (unit: meters / pixel).

[0191] This formula illustrates that the farther the camera is from the QR code (the larger h is), or the smaller the camera's field of view, the greater the actual distance represented by each pixel. This means the image's detail resolution will decrease. To improve positioning accuracy, we need y to be as small as possible, which means the camera needs to be closer to the QR code.

[0192] The relationship between the number of pixels occupied by the side length of the QR code in the image and its height:

[0193] Assuming the side length of the QR code is w, from the formula The number of pixels (p) occupied by the side length of the QR code can be obtained. Now, substituting y and w = 0.5 into the equation, we can obtain the relationship between the number of pixels occupied by the QR code's side length in the image and its height:

[0194] By substituting this standard into the formula, the optimal altitude or timing for initiating high-precision visual positioning can be calculated or determined in real time. To improve navigation accuracy, the QR code should occupy as many pixels as possible in the image, as it provides finer feature point coordinates. As the aircraft's altitude h increases, the number of pixels occupied by the QR code decreases, affecting recognition robustness and positioning accuracy. Therefore, this formula helps designers determine the minimum number of pixels the QR code needs to occupy at different altitudes to meet positioning accuracy requirements, and accordingly design the QR code size or switch positioning methods (e.g., the distance from satellite positioning to visual positioning).

[0195] Once we reach the predetermined altitude, the system uses the aircraft's onboard camera to capture images of the QR codes on the instruments to be carried. Image processing methods such as OpenCV are then used to identify the corner positions of the QR codes in a pixel coordinate system. This is combined with the intrinsic parameter matrix obtained beforehand using the Zhang Zhengyou calibration method (printing a chessboard image and taking multiple photos of it to obtain the pixel coordinates (u, v) of each corner point; fixing the world coordinate system on the chessboard with the origin at the chessboard plane; knowing the size of each square; and obtaining the coordinates (X, Y, 0) of each corner point in the world coordinate system; using the pixel coordinates and physical coordinates of each corner point in the world coordinate system for camera calibration), and the known physical coordinates of the QR code in a preset world coordinate system (with the center of the QR code as the origin), such as... Figure 3 The coordinate system diagram shown.

[0196] The system uses the PnP algorithm to calculate the precise three-dimensional position and attitude (i.e., extrinsic parameter matrices R and T) of the camera (representing the aircraft) relative to the world coordinate system of the QR code. This provides high-precision relative attitude information of the aircraft relative to the instrument, offering precise navigation commands to the flight control system and guiding the aircraft through final alignment, slow landing, or separation operations. The four corner points (X...) on the QR code... w ,Y w Z w The position of a point in the world coordinate system is known, and the corresponding coordinates (u, v) in the pixel coordinate system can be determined by QR code recognition and positioning. The conversion relationship between the two is as follows:

[0197]

[0198] In the formula Z c Let point Z be in the coordinate system of the airborne camera. c The coordinate values ​​of the axes; K is the camera's intrinsic parameter matrix, measured during camera calibration; Let R be the required camera extrinsic matrix; R and T are the rotation matrix and translation vector, respectively.

[0199] 3. Multimodal data fusion

[0200] In the final stage of automated UAV mounting / unmounting, achieving centimeter-level or even sub-centimeter-level high-precision positioning and attitude alignment is crucial for success. While visual recognition can provide high-precision relative pose, it is susceptible to environmental factors such as lighting and occlusion; laser rangefinders provide accurate altitude but have limited information dimensions; and inertial measurement units (IMUs) provide high-frequency motion information but suffer from integral drift. To overcome the limitations of single sensors, this solution, based on precise visual positioning, introduces extended Kalman filtering (EKF) technology based on multi-sensor information fusion to estimate the UAV's six-degree-of-freedom (6-DoF) pose relative to the target platform in real time, with high precision and robustness.

[0201] State variables: position (X, Y, Z) and velocity (V) of the aircraft x V y V z ), attitude quaternion (representing the orientation of the aircraft relative to the world coordinate system). These are the targets that the Kalman filter needs to estimate in real time.

[0202] Observations: These are the data sources used by the filter to correct predictions. QR code pose calculation, laser rangefinder, and inertial navigation data are the observations of this system.

[0203] QR code pose calculation: After the camera recognizes the QR code, the aircraft's pose (position and attitude) relative to the QR code's world coordinate system is calculated using the PnP algorithm. This is a high-precision, but only usable, local observation when the QR code is visible.

[0204] Laser rangefinder: Provides distance information from the aircraft to the ground, typically vertical distance. This effectively constrains altitude estimation, especially maintaining stability during descent.

[0205] Inertial navigation data: The IMU (Inertial Measurement Unit) provides acceleration and angular velocity. While IMU data itself is often used for state prediction (through integration), its raw readings or preprocessed data can also be used as observations to correct attitude, velocity, and even position in filters. IMUs typically update at extremely high frequencies, providing continuous short-term motion information.

[0206] This method uses pose data from QR code visual calculations, altitude data from a laser rangefinder, and motion information from an IMU as observations, and estimates the aircraft's position, velocity, and attitude as state variables. A Kalman filter, by combining the system's motion model predictions and real-time sensor measurements, optimally fuses this noisy data, thereby overcoming the limitations of a single sensor, improving the accuracy and robustness of pose estimation, and ensuring the successful execution of automated docking missions. Figure 4 The flowchart shown is for multimodal data fusion.

[0207] This flowchart presents the Kalman filter workflow for aircraft landing scenarios, used to fuse multi-source data to achieve high-precision state estimation. The process starts with the aircraft's position, velocity, and attitude from the previous moment. Using a kinematic model (e.g., a uniform linear motion model), it predicts the state for the next moment, obtaining the prior state and prior covariance. Subsequently, it uses QR code analysis to calculate attitude, laser ranging, inertial navigation, and other multi-source observation data. This data is then passed through an observation model to the update stage, where observation residuals and Kalman gain are fused to correct the prior state, outputting a posterior state estimate that integrates multi-source information. This state serves as both the current state estimation result and, in a closed loop, as the input for the next moment's process, continuously optimizing the estimation accuracy of the aircraft's position, velocity, and attitude quaternions during the landing phase. This balances computational efficiency with accuracy requirements, adapting to the state awareness requirements of the landing scenario.

[0208] In short, the process is as follows: the IMU provides rapid predictions, while other sensors provide relatively accurate observations. The extended Kalman filter continuously uses IMU data for rapid predictions and utilizes observation data from sensors such as vision and laser to correct prediction drift and errors, thereby obtaining a more accurate and stable UAV pose estimate than any single sensor, ensuring the accuracy and success rate of the mounting / dismounting process.

[0209] 4. Locking system

[0210] The robust locking system is the core component of the entire device responsible for the stability of the physical connection. After a sophisticated mechanical mechanism guides the drone to its aligned position with the target platform, the locking system activates, firmly securing the drone to the platform until separation is required.

[0211] Its "robustness" is reflected in its ability to reliably complete locking and unlocking operations under various complex environments and working conditions, and to ensure that the connection will not be accidentally disconnected in the locked state.

[0212] A robust locking system is the core guarantee for ensuring a reliable and secure physical connection between the UAV and the target platform. It employs a positive locking mechanism, a mechanical or electromechanical mechanism whose locked state is maintained not by friction, adhesion, or clamping force generated by elastic deformation, but by preventing separation or relative movement of the connecting parts through physical interlocking, interference, or constraint. Unlocking requires a specific, intentional unlocking operation to release this physical interference.

[0213] Meanwhile, the system is designed with environmental adaptability in mind, and can cope with challenges such as temperature changes, dust, rain and snow. It also provides key safety information to the control system by providing clear lock / unlock status feedback. When necessary, fault-tolerant and redundant designs can be introduced to further improve reliability in harsh application scenarios.

[0214] Secondly, the robust locking system, through its design features, strongly supports the versatility and flexibility of the entire solution. It establishes a standardized physical locking interface, allowing different models of drones and platforms to dock via this standard interface without requiring customized locking mechanisms for each combination. Simultaneously, the system possesses a certain load-bearing capacity and tolerance for minor alignment errors. This means that the same locking system can be compatible with a range of drone weights and can handle minor positional deviations in actual operation, significantly improving the system's adaptability to diverse drones and fixed / mobile platforms, greatly expanding its application scope and practical value.

[0215] It should be noted that the embodiments in this specification, through the above content, have the following beneficial effects:

[0216] 1. High-precision pose estimation method based on multi-sensor fusion:

[0217] This paper proposes an innovative multi-sensor data fusion algorithm (such as extended Kalman filtering or graph-optimized SLAM) that can fuse information from sensors based on different principles, including vision (camera), inertial measurement unit (IMU), lidar, and global navigation satellite system (GPS / RTK), in real time and with high accuracy. This method aims to overcome the limitations of single sensors in complex environments, providing centimeter-level or even sub-centimeter-level six-degree-of-freedom (position and attitude) relative pose estimation of the UAV relative to the target platform, which is a core technology for achieving precise physical docking.

[0218] 2. Phased and progressive autonomous docking control strategy:

[0219] This paper designs and implements a phased autonomous docking control strategy for unmanned aerial vehicles (UAVs), encompassing a stage from long-range coarse GPS positioning and navigation, to short-range precise positioning based on visual markers, and finally to precise docking control after high-precision pose estimation through multi-sensor fusion. This strategy enables UAVs to gradually transition from large-scale autonomous navigation to centimeter-level or even millimeter-level precise alignment and physical connection, achieving full-process automation and intelligence, and minimizing or eliminating human intervention.

[0220] 3. Precise positioning technology based on visual sign recognition:

[0221] After the drone approaches the target platform, a downward-looking camera is used to identify pre-set visual positioning markers (such as QR codes) on the target platform, and the initial relative pose of the drone with respect to the markers is quickly calculated. This technology, as a key step in the transition from satellite positioning to high-precision fusion positioning, provides a good initial value for subsequent high-precision pose estimation, and can adapt to recognition needs at different distances through the design of nested markers of different sizes.

[0222] 4. Multi-source sensing solutions to improve environmental adaptability and robustness:

[0223] By carefully selecting and fusing information from complementary multi-source sensors such as vision, inertial navigation, and lidar, this invention significantly improves the positioning accuracy and system robustness of the entire automatic docking system in complex dynamic environments such as changes in lighting, adverse weather conditions such as rain, snow, and fog, signal obstruction, and slight movement or shaking of the target platform. It ensures stable and reliable operation in all weather conditions and all scenarios, effectively solving the problem of easy failure of existing technologies in complex environments.

[0224] 5. Automatic drone mounting / unmounting system suitable for multiple platforms and application scenarios:

[0225] This invention provides an integrated system comprising a drone equipped with multiple sensors and a target platform with standardized positioning markers and a compatible mounting / detachment mechanism. Through the aforementioned technical solution, this system can adapt to different drone models (through software configuration or a few compatible parts) and diverse target platforms (including fixed, mobile, and different shapes, requiring little or no modification), achieving high versatility and flexibility. It significantly reduces deployment and maintenance costs and expands the application scenarios of drones in a wide range of scenarios, such as autonomous charging, payload exchange, multi-drone collaboration, and flying car assembly and disassembly.

[0226] Figure 5 A schematic diagram of an automatic mounting device for a drone, provided for one or more embodiments of this specification, includes:

[0227] At least one processor; and,

[0228] A memory communicatively connected to the at least one processor; wherein,

[0229] The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to:

[0230] The satellite signals of the target platform are obtained by using the positioning system of the designated drone;

[0231] Based on the satellite signals, coarse-grained position information of the target platform is determined, so that the designated UAV can approach the target platform based on the coarse-grained position information;

[0232] Obtain fused information of the designated UAV relative to the target platform, the fused information including pose information, altitude information and motion information;

[0233] Based on the fused information, the fine-grained location information of the target platform is determined, so that the designated UAV can be mounted onto the target platform based on the fine-grained location information.

[0234] This specification provides one or more embodiments of a non-volatile computer storage medium storing computer-executable instructions, which, when executed by a computer, can perform the following:

[0235] The satellite signals of the target platform are obtained by using the positioning system of the designated drone;

[0236] Based on the satellite signals, coarse-grained position information of the target platform is determined, so that the designated UAV can approach the target platform based on the coarse-grained position information;

[0237] Obtain fused information of the designated UAV relative to the target platform, the fused information including pose information, altitude information and motion information;

[0238] Based on the fused information, the fine-grained location information of the target platform is determined, so that the designated UAV can be mounted onto the target platform based on the fine-grained location information.

[0239] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0240] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the apparatus embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0241] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0242] In the embodiments provided in this application, it should be understood that the disclosed apparatus / network devices and methods can be implemented in other ways. For example, the apparatus / network device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0243] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0244] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The aforementioned units can be implemented in hardware or software.

[0245] If the integrated module / unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0246] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An automatic mounting method for unmanned aerial vehicles (UAVs), characterized in that, The method includes: The satellite signals of the target platform are obtained by using the positioning system of the designated drone; Based on the satellite signals, coarse-grained position information of the target platform is determined, so that the designated UAV can approach the target platform based on the coarse-grained position information; Obtain fused information of the designated UAV relative to the target platform, the fused information including pose information, altitude information and motion information; Based on the fused information, the fine-grained location information of the target platform is determined, so that the designated UAV can be mounted onto the target platform based on the fine-grained location information.

2. The method according to claim 1, characterized in that, The acquisition of the fusion information of the target platform includes: The image acquisition system of the designated UAV acquires image information of the target platform containing preset visual markers; Based on the image information of the preset visual markers, the pose information of the designated UAV relative to the target platform is determined, and the pose information includes the position information and operating attitude information of the designated UAV relative to the target platform. The altitude information of the designated UAV relative to the target platform is obtained through the ranging system of the designated UAV. The motion information of the designated UAV relative to the target platform is obtained through the inertial measurement system of the designated UAV.

3. The method according to claim 2, characterized in that, The preset visual marker is a designated QR code. The step of acquiring image information of the target platform containing the preset visual marker through the image acquisition system of the designated UAV includes: Based on the size of the specified QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system, the shooting height of the image acquisition system is determined. The shooting height is the distance at which the specified QR code in the image information can be accurately identified. Based on the shooting altitude, the designated drone is driven to the designated location, and the image acquisition system is used to acquire image information of the target platform containing the designated QR code.

4. The method according to claim 3, characterized in that, The step of determining the shooting height of the image acquisition system based on the size of the specified QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system includes: Input the size of the specified QR code in the target platform, the actual distance corresponding to a single pixel, the resolution of the image acquisition system, and the field of view of the image acquisition system into the formula. Determine the shooting height of the image acquisition system, wherein; y is the actual distance corresponding to a single pixel, FOV is the field of view of the image acquisition system, DPI is the resolution of the image acquisition system, and h is the shooting height of the image acquisition system.

5. The method according to claim 3, characterized in that, The image information determined based on the preset visual markers, and the pose information of the designated UAV relative to the target platform, include: The three-dimensional position coordinates and operational attitude information of the specified UAV relative to the world coordinate system of the QR code are determined based on the PnP algorithm. The operational attitude information includes a rotation matrix and a translation vector.

6. The method according to claim 1, characterized in that, The step of acquiring the motion information of the designated UAV relative to the target platform through the inertial measurement system of the designated UAV includes: Obtain the motion information of the designated UAV relative to the target platform at the previous moment; Based on a pre-created kinematic model, the motion information of the previous moment is processed to predict the current motion information of the designated UAV relative to the target platform.

7. The method according to claim 1, characterized in that, If the target platform is a dynamic target platform, determining the fine-grained location information of the target platform based on the fused information includes: Obtain the real-time motion status feedback information of the dynamic target platform; The fine-grained position information of the target platform is determined based on the fused information and the real-time motion state feedback information.

8. The method according to claim 1, characterized in that, If multiple designated drones are configured, the process of mounting the designated drones onto the target platform based on the fine-grained position information includes: The first drone is used as the master node. Based on the fine-grained position information, the designated drone is mounted to the target platform. The master node's mounting position information is then released to the other drones so that the remaining drones can be mounted to the target platform based on the mounting position information.

9. An automatic mounting device for unmanned aerial vehicles (UAVs), characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: The satellite signals of the target platform are obtained by using the positioning system of the designated drone; Based on the satellite signals, coarse-grained position information of the target platform is determined, so that the designated UAV can approach the target platform based on the coarse-grained position information; Obtain fused information of the designated UAV relative to the target platform, the fused information including pose information, altitude information and motion information; Based on the fused information, the fine-grained location information of the target platform is determined, so that the designated UAV can be mounted onto the target platform based on the fine-grained location information.

10. A non-volatile computer storage medium, characterized in that, It stores computer-executable instructions, which, when executed by a computer, can achieve the following: The satellite signals of the target platform are obtained by using the positioning system of the designated drone; Based on the satellite signals, coarse-grained position information of the target platform is determined, so that the designated UAV can approach the target platform based on the coarse-grained position information; Obtain fused information of the designated UAV relative to the target platform, the fused information including pose information, altitude information and motion information; Based on the fused information, the fine-grained location information of the target platform is determined, so that the designated UAV can be mounted onto the target platform based on the fine-grained location information.

Images