Unmanned aerial vehicle landing control method and system, unmanned aerial vehicle take-off and landing platform and storage medium

Abstract

The invention discloses an unmanned aerial vehicle landing control method and system, an unmanned aerial vehicle take-off and landing platform and a storage medium, and relates to the technical field of unmanned aerial vehicle take-off and landing, and the method comprises the following steps: calculating and generating state data representing platform physical structure dynamic space constraints; distributing a monotonically increasing unique version number for each generated state snapshot; based on a landing trajectory planning request of an unmanned aerial vehicle end, obtaining a state snapshot version number declared in the request as a to-be-verified version number; comparing the version number to be verified with the latest version number in the currently released constraint state sequence; allowing the unmanned aerial vehicle to execute landing trajectory planning or control based on the state snapshot corresponding to the latest version number; if the version number to be verified is earlier than the latest version number, a response indicating that the version state is inconsistent is returned to the unmanned aerial vehicle end, and a landing process based on the version number to be verified is terminated; according to the invention, the landing success rate and stability of the system in a complex marine environment are obviously improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) take-off and landing technology, specifically to UAV landing control methods, systems, UAV take-off and landing platforms, and storage media. Background Technology

[0002] With the rapid development of the low-altitude economy at sea, drones are increasingly widely used in the field of marine engineering, especially in scenarios such as offshore wind power inspection, maritime rescue, and maritime patrol and law enforcement. Drones, with their high-altitude wide-area vision and flexible maneuverability, have become key equipment for improving operational efficiency. At present, the cumulative grid-connected installed capacity of offshore wind power in China has exceeded 44.2 million kilowatts, accounting for more than half of the global total installed capacity. Moreover, the distance of offshore wind power from the shore is generally more than 50km. Daily inspections mainly rely on drones to take off from ship-borne platforms to carry out tasks. Unmanned and intelligent full-process inspection has become an inevitable trend in the industry.

[0003] Because ships at sea are subject to violent and irregular swaying caused by waves and wind, the take-off and landing of drones poses a great challenge. Traditional manual launch and recovery methods are costly and risky.

[0004] To address this issue, marine intelligent dynamic compensation technology has emerged. Based on this technology, a stable take-off and landing platform for unmanned aerial vehicles (UAVs) and a marine dynamic compensation UAV hangar have been developed. Through the combination of servo mechanisms and intelligent algorithms, they can sense changes in ship attitude in real time and compensate quickly, enabling the take-off and landing platform to remain level even when the ship is swaying, thus providing a relatively stable foundation for UAV take-off and landing.

[0005] However, the marine environment is extremely complex. In existing shipborne UAV landing systems, the UAV relies on static or outdated data of the platform's physical structure when planning its landing trajectory. In the process of maintaining a horizontal position, the physical structures of the shipborne dynamic compensation platform, such as guardrails and hangar roofs, undergo dynamic spatial position changes as the actuators perform compensation actions. Furthermore, the constraint information generated by these dynamic changes has not been integrated with the UAV's planning process through a real-time, mandatory version consistency verification and coordination mechanism. This results in a deviation between the UAV's planned landing trajectory and the platform's actual physical constraints, which in turn leads to collision risks or landing failures. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a drone landing control method, system, drone take-off and landing platform, and storage medium.

[0007] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for controlling the landing of an unmanned aerial vehicle (UAV), comprising the following steps: Condition monitoring steps: Monitor the attitude changes of the take-off and landing platform and the actions of the actuators. If the update conditions are met, calculate and generate state data representing the dynamic spatial constraints of the platform's physical structure based on the platform's 3D model and kinematic model, as a state snapshot. Versioned sequence construction steps: Assign a monotonically increasing unique version number to each generated state snapshot to form a versioned constraint state sequence; Planning request response steps: Based on the landing trajectory planning request from the UAV, obtain the state snapshot version number declared in the request as the version number to be verified; Version consistency verification steps: Compare the version number to be verified with the latest version number in the currently released constraint state sequence; Authorization and execution steps: The UAV is allowed to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number only when the version number to be verified is consistent with the latest version number; Process termination steps: If the version number to be verified is earlier than the latest version number, a response indicating a version inconsistency will be returned to the drone terminal, and the landing process based on the version number to be verified will be terminated.

[0008] As a preferred technical solution of the present invention, the updating conditions include the absolute value change of the roll angle or pitch angle of the take-off and landing platform exceeding the angle threshold, or the extension and retraction amount change of any of its actuators exceeding the displacement threshold. The generation of status snapshots specifically includes: calculating the real-time coordinate set of at least one physical structure in space, such as the platform guardrail and the hangar roof; obtaining the current travel distance and remaining travel margin of each actuator; and calculating the real-time minimum distance between the UAV and the physical structure.

[0009] As a preferred technical solution of the present invention, the planning request response step and the process termination step are executed by the status version controller deployed on the take-off and landing platform side; In the planning request response step, the status version controller receives a planning request from the drone, which encapsulates a version number to be verified. In the process termination step, when the status version controller determines that the version number to be verified is earlier than the latest version number, it carries the latest version number in the returned response indicating that the version status is inconsistent.

[0010] As a preferred embodiment of the present invention, after the authorization and execution steps, a state locking step is further included: Obtain the altitude of the drone relative to the landing surface of the take-off and landing platform; When the altitude first drops below the preset locked altitude threshold, the current latest version number is marked as the locked version number; During the period from when the altitude is below the locked altitude threshold until the end of this landing attempt, the status monitoring step and the versioned sequence construction step are suspended. For subsequent planning requests initiated by the UAV, the authorization and execution steps are based on the status snapshot data corresponding to the locked version number. During the lockout period, if the remaining travel margin of the actuator of the take-off and landing platform is determined to be lower than the limit threshold based on the status snapshot corresponding to the lockout version number, the process termination step is executed to terminate the current landing process and control the drone to enter the hovering state.

[0011] As a preferred technical solution of the present invention, when the UAV performs landing trajectory planning based on the state snapshot, the minimum safe distance threshold used is dynamically determined according to the remaining travel margin obtained in the state snapshot. The rule is that the average value of the remaining travel margin is negatively correlated with the minimum safe distance threshold.

[0012] As a preferred embodiment of the present invention, after the versioned sequence construction step, a data fusion processing step is further included: Monitor the update frequency of the version number of the status snapshot; When the update frequency exceeds the preset frequency threshold, the platform physical structure spatial location data and actuator state data in multiple state snapshots generated within a continuous time period are fused and calculated to generate a state snapshot representing the comprehensive constraint state within that time period. Assign a new version number to the state snapshot of the integrated constraint state and update the constraint state sequence.

[0013] As a preferred embodiment of the present invention, after the process termination step, an exception handling step is further included: If the drone receives more than a preset number of inconsistent response versions within a preset time period, the drone will be controlled to ascend to a specified safe altitude and enter a hovering waiting state. During the hovering wait period, continuously listen for version update events of the constraint state sequence; Only when a version update event is detected and the time interval between two consecutive updates is greater than a set stability threshold, the drone is controlled to re-initiate a landing trajectory planning request based on the latest version number.

[0014] The present invention also provides a drone landing control system, comprising: The status monitoring module is used to monitor the attitude changes and actuator actions of the dynamically compensated sea surface take-off and landing platform. If the update conditions are met, it calculates and generates status data representing the dynamic spatial constraints of the platform's physical structure based on the platform's 3D model and kinematic model, as a status snapshot. The versioned sequence building module is used to assign a monotonically increasing unique version number to each generated state snapshot, forming a versioned constraint state sequence. The planning request response module is used to respond to the landing trajectory planning request from the UAV and obtain the state snapshot version number declared in the request as the version number to be verified. The version consistency verification module is used to compare the version number to be verified with the latest version number in the currently released constraint state sequence; The authorization and execution module is used to allow the UAV to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number only when the version number to be verified is consistent with the latest version number; The process termination module is used to return a response indicating a version inconsistency to the drone if the version number to be verified is earlier than the latest version number, and to terminate the landing process based on the version number to be verified.

[0015] The present invention also provides a drone take-off and landing platform, characterized in that it includes a platform body, a dynamic compensation actuator disposed on the platform body, a sensor for monitoring the platform attitude, and a control unit.

[0016] The present invention also provides a computer-readable storage medium, characterized in that it stores a computer program thereon, which, when executed by a processor, implements a method for controlling the landing of an unmanned aerial vehicle.

[0017] The beneficial effects of this invention are: 1. In this invention, by versioning the physical constraint state of the platform and verifying the version consistency when making planning requests, it is ensured that each landing trajectory planning of the UAV is based on the latest state snapshot, which completely eliminates the cognitive gap between the data used for planning and the actual state of the platform, significantly reduces the risks of insufficient airspace and trajectory deviation caused by outdated data, and improves the matching accuracy between the landing trajectory and the actual constraints of the platform.

[0018] 2. This invention pauses state updates after the UAV descends to the locked altitude threshold, keeping the constraint state on which the UAV makes decisions during the critical landing phase stable and avoiding planning oscillations or emergency obstacle avoidance risks caused by sudden state changes. At the same time, the minimum safe distance threshold is dynamically adjusted in a negative correlation with the remaining travel margin of the actuator, automatically adopting a more conservative planning strategy when the platform approaches its physical limits, thus avoiding risky landing operations when the platform's capabilities are insufficient, and further ensuring landing safety under extreme sea conditions.

[0019] 3. In this invention, a state snapshot of the comprehensive constraint state is generated by fusion calculation, which avoids continuous interruption of UAV planning due to frequent state updates; when the abnormal handling step receives continuous version inconsistency responses, the UAV is controlled to rise to a safe altitude and hover and wait, and planning is restarted only after the platform state stabilizes. This effectively copes with complex scenarios such as severe turbulence and significantly improves the landing success rate and stability of the system in complex marine environments. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0021] Figure 1 This is a schematic diagram illustrating the workflow of the UAV landing control method of the present invention.

[0022] Figure 2 This is a schematic diagram of the overall structure of the UAV landing control system of the present invention. Detailed Implementation

[0023] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0024] like Figure 1 As shown, the drone landing control method includes the following steps: Status monitoring steps: Monitor the attitude changes of the take-off and landing platform and the actions of the actuators. If the update conditions are met, calculate and generate status data representing the dynamic spatial constraints of the platform's physical structure based on the platform's three-dimensional model and kinematic model, as a status snapshot. This step continuously monitors the attitude changes of the take-off and landing platform and the actions of its actuators. The take-off and landing platform usually refers to a UAV take-off and landing platform with dynamic compensation function installed on the ship deck. It includes a platform body that can actively extend and retract through actuators such as electric cylinders to counteract the ship's sway. The specific objects monitored include, but are not limited to: the roll and pitch angles of the platform obtained through the inertial measurement unit (IMU), and the extension and retraction of each actuator obtained through sensors such as encoders; When the absolute value of the detected attitude angle change exceeds the preset angle threshold (e.g., 0.5 degrees), or the extension and retraction change of any actuator exceeds the preset displacement threshold (e.g., 1 mm), it is determined that the platform has undergone a meaningful attitude adjustment or compensation action, and the update conditions are met. Once the update conditions are met, the system will calculate and generate a state snapshot in real time based on the pre-established platform 3D model and kinematic model. This state snapshot is a set of state data characterizing the dynamic spatial constraints of the platform's physical structure. Its core contents include: the set of spatial coordinates of the platform's key physical structures (such as guardrails and hangar roofs) after changes due to compensation actions, the current travel and remaining travel margin of each actuator, and the estimated real-time minimum distance between the UAV and these physical structures. For example, when the ship rolls to starboard, the port electric cylinder extends and the starboard cylinder retracts to keep the deck level. This step will calculate the actual spatial position of the top of the port guardrail at this time, which has been raised and shifted compared to the horizontal stationary state.

[0025] Versioned sequence construction steps: Assign a monotonically increasing unique version number to each generated state snapshot. This step discretizes the continuous physical state change process into a series of definite states with clear sequence identifiers. All state snapshots with version numbers are arranged in the order of generation to form a versioned constraint state sequence. The version number is essentially a logical marker for each major change event in the lifting platform (hereinafter referred to as the platform). For example, the first state snapshot version number is V1. When the platform generates a new state snapshot after compensation due to new shaking, the version number V2 is assigned, and so on. The latest entry in this sequence represents the most authoritative and accurate physical constraint reality of the platform at the current moment.

[0026] Planning request response steps: Based on the landing trajectory planning request from the UAV, obtain the state snapshot version number declared in the request as the version number to be verified; When the control module on the UAV side (such as the landing envelope verification and adapter) needs to perform landing trajectory planning or replanning, it will send a landing trajectory planning request to the platform side. This request must encapsulate a key parameter, namely the version number of the state snapshot currently held by the UAV side module or used in the last successful planning. This version number is extracted as the version number to be verified in this step. This means that when the UAV side requests planning authorization, it must declare which historical version of state cognition its decision is based on.

[0027] Version consistency verification steps: Compare the version number to be verified with the latest version number in the currently released constraint state sequence; The purpose of the comparison is to verify whether the drone's intended state perception is synchronized with the platform's latest physical reality.

[0028] Authorization and Execution Steps: This step generates branch logic based on the verification result. Only when the version number to be verified is consistent with the latest version number, the UAV is allowed to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number. At this point, the system determines that the drone's perception is up-to-date, and therefore authorizes the drone to perform subsequent landing trajectory planning or control based on the state data (such as the latest structural coordinates and remaining travel margin) contained in the state snapshot corresponding to this latest version number. For example, the drone planning algorithm will use the guardrail coordinates in this latest state snapshot as the obstacle boundary to calculate a safe landing path.

[0029] Process termination step: If the version number to be verified is earlier than the latest version number, then return a response indicating that the version status is inconsistent to the drone terminal, and terminate the landing process based on the version number to be verified. If, during the version consistency verification step, it is found that the version number to be verified is earlier than the latest version number, it indicates that since the last time the drone terminal obtained the status information, the platform has performed a new compensation action and generated an updated status snapshot. At this time, the status snapshot corresponding to the version number to be verified held by the drone terminal is outdated, and there is a security risk in planning based on this old state. Therefore, the system will return a response indicating an inconsistency in the version status to the drone and terminate the landing process based on the version number to be verified. This means that the current landing attempt or planning cycle is forcibly terminated, and the drone must obtain the latest status information before restarting the planning process.

[0030] By iteratively executing the above steps, this method constructs a closed-loop collaborative control logic. It ensures that every effective landing decision of the UAV is strictly bound to the latest physical state generated after a specific compensation action of the platform at the business level. This solves the core problem of the disconnect between planning and execution consequences in dynamic scenarios and significantly improves the reliability and safety of autonomous landing of UAVs on mobile platforms.

[0031] Furthermore, the update conditions include the absolute value change of the roll or pitch angle of the take-off and landing platform exceeding an angle threshold, or the extension / retraction change of any of its actuators exceeding a displacement threshold. The first type of condition is for the overall attitude angle change of the take-off and landing platform. The setting of the angle threshold (e.g., 0.5 degrees) is intended to filter out minor swaying of the ship that does not affect the substantial change of the platform's physical structure spatial position. Only when the attitude angle change is significant enough is it determined to be an event that may significantly change the platform's structural spatial constraints, thereby triggering a state update. For example, when operating at sea, a gust of wind and waves causes the ship to roll by 3 degrees. This change far exceeds the 0.5-degree angle threshold, and the system immediately determines that the update conditions are met. The second type of condition refers to the change in the motion of the actuators (such as servo electric cylinders) of the lifting platform. That is, the change in the extension or retraction of any actuator exceeds a preset displacement threshold. The change in extension or retraction can be calculated by the encoder feedback value of the servo motor. The setting of the displacement threshold (e.g., 1 mm) reflects the system's sensitivity to the small movements of the actuators. Even if the change in attitude angle does not reach the angle threshold, if an actuator produces an extension or retraction exceeding the displacement threshold in order to fine-tune the platform attitude, it also means that the platform structure has undergone measurable spatial displacement. Therefore, it is also necessary to trigger a state update to ensure the accuracy of the constraint state description. These two conditions ensure that any significant change in the ship's attitude or a minor adjustment in the platform's actuators, as long as its physical consequences may affect the safety of the drone's landing, can be detected in a timely manner and the status synchronization process can be initiated.

[0032] The generation of the state snapshot specifically includes: calculating the real-time coordinate set of at least one physical structure in the space, such as the platform guardrail and the hangar roof, obtaining the current travel and remaining travel margin of each actuator, and calculating the real-time minimum distance between the UAV and the physical structure; The platform's 3D model predefines the geometry, dimensions, and relative positional relationships of all key constraint components (such as guardrails and hangar roofs) to the platform's kinematic joints (such as electric cylinder hinge points). When the update condition is triggered, the system performs forward kinematics calculations based on the target or actual extension / retraction of all actuators. This calculation does not only output the attitude of the deck plane but also accurately calculates the 3D world coordinates of each key point or boundary line on the physical structure of interest under the current platform configuration. For example, the coordinates of the center points of the top edges of the four guardrails and the coordinates of the center point of the front edge of the hangar roof opening are calculated. These coordinate sets dynamically depict the real-time positions of obstacles that the drone needs to avoid during landing.

[0033] By acquiring the current stroke and remaining stroke margin of all actuators, the system can assess the platform's potential reserve of ability to maintain level or compensate when encountering further sway. The system reads the feedback value of the encoder of each actuator (electric cylinder) from the actuator's driver to obtain its current stroke. At the same time, the system can store or query the physical maximum and minimum stroke of each actuator. The remaining stroke margin usually refers to the proportion or absolute value of the amount of stroke that the actuator can continue to operate in the current extension or retraction direction (extension or retraction) relative to the total available stroke. For example, if an electric cylinder with a maximum extension of 500 mm has now extended to 450 mm, its remaining stroke margin in the extension direction is very small.

[0034] Calculating the real-time minimum distance between the UAV and the physical structure is a comprehensive safety assessment information in the state snapshot. Its calculation requires combining the real-time coordinate set of the physical structure calculated above, as well as the current or estimated position and geometric information of the UAV. The position information of the UAV can be obtained through its downlink GPS, RTK or visual positioning data. At the same time, the system has a three-dimensional shape model of the UAV (especially the diameter and position of the rotor disk). By calculating the spatial Euclidean distance between the UAV's outline (usually the outer edge of the rotor) and each key coordinate point of the physical structure, and taking the minimum value among them, the real-time minimum distance between the UAV and the physical structure can be obtained. This distance value intuitively reflects the most pressing spatial interval between the drone and the platform's moving structure at the current moment, and is an important basis for judging whether there is an immediate risk of collision on the landing trajectory.

[0035] Through the comprehensive calculation of the above three steps, the state snapshot generated by the system fully reflects the physical constraint state of the platform under the current compensation action, providing an accurate and comprehensive data foundation for subsequent version consistency verification and landing authorization.

[0036] Furthermore, the version consistency verification step, the planning request response step, and the process termination step are all executed by the status version controller deployed on the take-off and landing platform side; The core functions of version management and consistency adjudication are centrally deployed on the elevator platform side based on the following considerations: First, the elevator platform is the root cause of all physical constraint state changes. The computing unit on the elevator platform side calculates and maintains the versioned constraint state sequence in real time, which can ensure the uniqueness and authority of the source of state data. Second, centralized verification avoids the state divergence and synchronization delay problems that may occur in distributed systems. In practical applications of offshore wind power inspection, ships may be located in sea areas more than 50 kilometers away from the coast, where communication conditions are complex. Deploying critical version consistency verification logic on the platform side can effectively reduce reliance on remote communication links such as satellites or 4G / 5G, and improve the system's reliability in harsh communication environments.

[0037] In the planning request response step, the status version controller receives a planning request from the UAV, and the planning request encapsulates the version number to be verified. When a drone needs to plan its landing trajectory, its onboard computing unit constructs a standardized planning request data packet. This data packet follows a predefined communication protocol, and the key field is the version number to be verified. This version number is the version identifier of the platform's physical constraint state that the drone last successfully obtained. It represents the data version on which the drone's current decision is based. The request data packet is transmitted to the shipboard platform through a dedicated data link between the ship and the drone (usually using a low-latency communication module in the 2.4GHz or 5.8GHz band). After receiving the request, the status version controller first parses the data packet, extracts the version number to be verified encapsulated in it, and then compares it with the latest version number maintained internally. This mechanism ensures that a rigorous version check is performed before each landing decision, avoiding the arbitrariness of drones planning directly based on cached or historical data in traditional systems.

[0038] In the process termination step, when the status version controller determines that the version number to be verified is earlier than the latest version number, it carries the latest version number in the returned response indicating that the version status is inconsistent. The status version controller executes strict version number comparison logic. If it determines that the version number to be verified is less than the latest version number, it constructs a response data packet containing a "status expired" error code. At the same time, it fills the current latest version number into a specific field of the data packet. This response data packet is returned to the drone through the same dedicated data link. After receiving the response, the drone can not only know that the current planning request has been rejected due to version inconsistency, but also immediately know which version number to re-initiate the request based on, thus avoiding the process of repeated trial and error. In severe sea conditions where the ship rolls more than 15 degrees, the physical constraint state of the platform may change significantly within milliseconds. This mechanism ensures that the UAV can immediately replan based on the latest platform state after receiving a rejection response, without having to wait for the next periodic state update.

[0039] Furthermore, following the authorization and execution steps, a state locking step is also included: The height of the UAV relative to the landing surface of the take-off and landing platform is obtained in real time. The height is the vertical distance between the UAV and the designated landing surface on the take-off and landing platform (usually the center area of ​​the deck plane that is kept horizontal after dynamic compensation). When the height first drops below a preset locked height threshold, the current latest version number is marked as the locked version number; The lock-on altitude threshold is a preset system parameter, determined based on the UAV rotor size, platform structure height, and typical descent speed, and is usually set to 5 meters. For example, the system captures the current latest version number (e.g., V15) the instant the drone's altitude drops from 5.1 meters to 4.9 meters and marks it as the locked version number. This marking operation is achieved by modifying the version management table inside the state version controller, ensuring that subsequent operations can recognize and use the locked version.

[0040] During the period from when the altitude is below the locked altitude threshold until the end of this landing attempt, the execution of the status monitoring step and the versioned sequence construction step is suspended, and for all subsequent planning requests initiated by the UAV, the authorization and execution steps are all executed based on the status snapshot data corresponding to the locked version number; Once locked, the system will pause the normal dynamic status update process. Specifically, during this period, regardless of the attitude changes or actuator actions of the platform caused by the ship's rolling, the system will no longer respond to these changes to trigger new state monitoring steps and version sequence construction steps. That is, it will no longer generate and publish state snapshots with new version numbers. The platform's own dynamic compensation control will still run to keep the deck level, but the resulting spatial consequences of its actions (i.e., structural position changes) will no longer be calculated in real time and updated to the collaborative decision-making system. Meanwhile, for any landing trajectory planning request initiated by the UAV during this period, in the authorization and execution steps, the system will ignore the current version number, which may have changed, and instead force the use of the data (including structural coordinates, remaining travel margin, etc.) in the state snapshot bound to the locked version number to respond to the request, authorizing the UAV to perform trajectory planning or control based on this fixed constraint data.

[0041] During the locking period, if the remaining travel margin of the actuator of the take-off and landing platform is determined to be lower than the limit threshold according to the status snapshot corresponding to the locking version number, the process termination step is executed to terminate the current landing process and control the UAV to enter the hovering state. The state lockout process not only provides stability but also includes proactive safety monitoring. During the lockout, the system continuously or periodically checks the remaining travel margin of the actuators recorded in the locked state snapshot. Typically, a limit threshold (e.g., 10%) is set, which indicates that the platform's ability to resist further shaking is nearly exhausted and there is a risk of instability. If, during the lockout, the check finds that the remaining travel margin of the actuators in the state snapshot is lower than this limit threshold, it indicates that even based on the frozen state, the platform is in a dangerous state of insufficient capability. At this point, the system will immediately trigger a safety response: execute the process termination step, send a termination command to the drone terminal, forcibly terminate the current landing process, and control the drone to stop descending, enter a hovering state, and wait for subsequent instructions (such as go-around or waiting for the sea conditions to ease). This achieves proactive perception and immediate risk avoidance of the platform's capability degradation risk in the final stage.

[0042] The design of this state-locking step fully considers the special characteristics of the final stage of UAV landing. In offshore wind farm inspection missions more than 50 kilometers from the shore, communication conditions are complex and sea conditions are changeable. Traditional solutions often cause UAVs to frequently adjust their trajectories in the final stage due to continuous changes in platform status, increasing the risk of collision. This invention creates a stable and consistent decision-making environment in the critical stage through the state-locking mechanism, enabling UAVs to focus on precise ground contact without having to deal with real-time changes in platform physical constraints.

[0043] Furthermore, when the UAV performs landing trajectory planning based on the state snapshot, the minimum safe distance threshold used is dynamically determined according to the remaining travel margin obtained from the state snapshot; The rule is that the average value of the remaining travel margin is negatively correlated with the minimum safe distance threshold.

[0044] The remaining travel margin reflects the remaining ability of the take-off and landing platform's actuators to continue to extend and retract in all directions to counteract further swaying of the ship in the current attitude. The smaller the average value of the remaining travel margin, the less dynamic compensation capacity the platform has, and the more limited its potential to maintain stability or cope with sudden swaying. In this state, the risk of unpredictable and significant positional adjustments to the platform due to subsequent shaking increases. If the drone continues to fly close to the conventional fixed safety boundary, the risk of collision will increase significantly. Therefore, it is necessary to establish a risk-adaptive safety strategy.

[0045] The specific mathematical relationship of "negative correlation" can be designed according to system requirements, for example, by using a linear function relationship: ; Among them, D base The basic safety distance is a static reference value set according to the drone model, control precision, and general safety requirements. α is the elasticity coefficient, used to adjust the sensitivity of the safety distance as the capability decreases; R avg The average remaining travel margin is calculated from the state snapshot, and its value is between 0 and 1, when R avg At higher values ​​(e.g., 0.8), the platform has strong capabilities, (1-R) avg The value is small, D safeSlightly higher than D base The planning is relatively efficient, when R avg At lower values ​​(e.g., 0.2), the platform capability is weak, (1-R) avg The value is large, D safe Significantly increased, planning tends to be conservative.

[0046] For example, in actual offshore wind power inspection scenarios, when the vessel is in calm sea conditions, the average remaining travel margin R of the platform actuators is... avg It is usually maintained above 0.7, at which point the minimum safe distance threshold D is reached. safe The calculation is 0.5 + 0.3 * (1 - 0.7) = 0.59 meters. The drone can plan an efficient landing trajectory based on this relatively compact safety distance. As sea conditions worsened, the ship's roll angle increased to 15 degrees, and the platform's actuators continuously increased their travel to maintain levelness. avg It gradually decreases to 0.3, at which point D safe The automatic adjustment is 0.5 + 0.3 * (1 - 0.3) = 0.71 meters. The drone landing envelope verification and adapter replan the trajectory based on this increased diameter safety zone to ensure sufficient distance from structures such as guardrails and hangar roofs. When a ship encounters extreme sea conditions and its roll reaches nearly 20 degrees, the average remaining stroke margin R of the platform actuators... avg When it drops to 0.1, D safe The distance is further increased to 0.77 meters, which is almost 1.5 times the basic safety distance. At this time, the trajectory planned by the drone will be significantly away from the edge of the platform. If the space is insufficient to meet this requirement, the system will automatically trigger the hovering or go-around process to avoid a forced landing when the platform's capacity is critical.

[0047] Furthermore, following the versioned sequence construction step, a data fusion processing step is also included: Monitor the update frequency of the version number of the aforementioned status snapshot; The system continuously monitors the generation and publication events of the state snapshots and calculates the number of times the new version number appears per unit time, i.e., the version number update frequency. This frequency directly reflects the degree of change in the platform's physical world due to compensation actions. For example, in calm sea conditions, the frequency may be below 1 Hz, while when subjected to continuous large waves, the frequency may rise sharply to 10 Hz or even higher.

[0048] When the update frequency exceeds a preset frequency threshold, the platform physical structure spatial location data and actuator state data in multiple state snapshots generated within a continuous time period are fused and calculated to generate a state snapshot representing the comprehensive constraint state within that time period. The frequency threshold is determined based on the processing capacity, communication bandwidth, and allowable state delay of the UAV planning system. The frequency threshold can be set to 5 Hz. When the monitored update frequency exceeds this threshold, it indicates that the platform status is changing too fast. If every minute change continues to be published as an independent status, the drone-side planning module may be unable to form a stable trajectory due to frequent status updates and "status expiration" responses. At this time, the system starts the data fusion mechanism. The fusion operation is performed on multiple state snapshots (assuming there are 5, version numbers Vn to Vn+4) generated within a continuous time period (e.g., the most recent 0.5 seconds). The fusion calculation mainly includes two aspects: Fusion of spatial location data of platform physical structure: For the same physical structure (such as a specific guardrail vertex), extract its spatial coordinate sequence in multiple snapshots, and the fusion algorithm (such as calculating the coordinate envelope range, average position or using a filtering algorithm to generate a representative trajectory) generates a fused coordinate that can characterize the movement range or typical position of the structure within the time period. Fusion of actuator status data: Extract actuator travel margin data from multiple snapshots, and use a fusion algorithm (such as taking the minimum value, average value, or predicting based on the trend) to generate a fused travel margin that can characterize the overall level of the platform's compensation capability during the time period.

[0049] The fused coordinate set of all structures obtained from the above fusion calculation, the fused travel margin set, and the optional calculated fused minimum clearance distance are collectively encapsulated into a new data object, called the state snapshot of the comprehensive constraint state. This snapshot represents the statistical characteristics of the platform constraint state in the most recent time period.

[0050] Assign a new version number to the state snapshot of the integrated constraint state and update the constraint state sequence; To this end, a new, monotonically increasing version number (e.g., Vn+5) is assigned to the newly generated state snapshot of the integrated constraint state. Then, this is used as the latest valid state to update the versioned constraint state sequence and issue an update notification. This means that during high-speed updates, the UAV will receive a sequence of state snapshots based on the fused state with a lower update frequency, rather than a series of overly fragmented instantaneous states.

[0051] For example, in scenarios where ships encounter irregular and continuous large waves, causing the platform to frequently perform large-scale and high-frequency compensation, if integration is not performed, the state version controller may release a series of states such as version number V101, V102, V103, etc. in a short period of time. The UAV planning module may have just started planning based on V101 when it receives an update notification for V102, causing the planning to be judged as being based on an expired state and interrupted. This cycle continues, making it impossible to make effective progress. After enabling the data fusion processing step, when the system detects that the update frequency exceeds the threshold (e.g., 5Hz), it will merge the five states V101 to V105 into a comprehensive state within a 0.2-second time window and publish it as V106. The UAV may only receive two valid state updates, V100 and V106, thus having enough time to complete a complete trajectory planning and execution based on V106, which greatly improves the collaborative stability and mission success rate of the system in extreme dynamic environments.

[0052] Furthermore, following the process termination step, an exception handling step is also included: If the drone receives more than a preset number of inconsistent response versions of the instruction within a preset time period, the drone is controlled to ascend to a specified safe altitude and enter a hovering waiting state. This is the trigger and initial security procedure for this step. The system sets two key parameters: a preset time period (e.g., 10 seconds) and a preset number of times (e.g., 5 times). These two parameters together define an anomaly detection window. When the drone initiates a landing trajectory planning request in a continuous landing attempt, and each of its requests is rejected in the process termination step due to version inconsistency, and it continuously receives more than a preset number of responses indicating version inconsistency, such as 5 times, and these responses all occur within a preset time period (such as 10 seconds), the system determines that the current collaborative environment is in an extremely unstable abnormal state. In this state, continuing to attempt a landing is highly likely to fail and consume the drone's battery. Therefore, the system immediately issues a control command, ordering the drone to stop the current landing process and ascend vertically to a designated safe altitude. This designated safe altitude is a pre-set hovering altitude that is far away from the platform structure and ensures absolute safety (e.g., more than 20 meters away from the platform). After reaching this altitude, the drone enters a hovering waiting state, stops all active landing planning attempts, and waits for the environment to improve.

[0053] During the hovering wait period, continuously monitor the version update events of the constraint state sequence; In the hovering and waiting state, the UAV is not completely passive. Its onboard system or ground station module maintains a continuous connection with the platform-side state version controller and listens for update events of the versioned constraint state sequence, that is, listens for whether a new state snapshot version number is published. By listening for version update events instead of directly requesting data, it can continuously perceive the dynamic change trend of the platform world with low communication overhead. Only when a version update event is detected and the time interval between two consecutive updates is greater than a set stability threshold, the drone is controlled to re-initiate a landing trajectory planning request based on the latest version number; This is the recovery condition and action for this step. The system sets a set stability threshold (e.g., 2 seconds), which represents the degree of change in the platform state. During the monitoring process, the drone not only records whether an update has occurred, but more importantly, it calculates the time interval between two adjacent version update events. The recovery conditions are: the drone must detect a version update event (indicating that the platform is still working and in a released state), and the calculated time interval between two consecutive updates must be greater than the set stability threshold. For example, if the platform is shaking violently, it may release multiple new versions per second with an interval of much less than 2 seconds. Only when the shaking slows down, the platform's state changes more slowly, and the release interval gradually lengthens and stably exceeds 2 seconds, can it be considered that the platform's movement has entered a relatively stable or predictable stage, and it has the basis to try safe collaboration again. Once the above conditions are met, the system controls the drone to re-initiate a landing trajectory planning request based on the latest version number. That is, the drone uses the latest version number it has been listening to to initiate a new planning request to the platform side, restarting the normal landing process that begins with the planning request response step.

[0054] For example, when a ship encounters unusually violent and irregular continuous large waves, the platform may perform several large compensations per second, causing the status snapshot version number to be updated extremely rapidly (such as V100, V101, V102... being released continuously within 2 seconds). If the drone planning module cannot keep up with this change, it may receive five consecutive responses within 2 seconds indicating inconsistent version status, such as "based on V100 expired", "based on V101 expired", etc. At this time, the exception handling steps are triggered, and the drone automatically rises to a safe altitude and hovers to avoid risks in the chaos. During hovering, it continuously monitors version number updates. In the initial stage, it may still hear a dense series of V103, V104, etc., with update intervals of very short intervals (0.3 seconds). As the wind and waves subside, the ship's swaying slows down, the platform's compensation actions become slower, and the version update frequency decreases. The time interval between V105 and V106 detected by the drone increases to 3 seconds, exceeding the set stability threshold of 2 seconds. The system determines that the environment has become stable, and then controls the drone to re-initiate the landing request based on the latest version number V106 detected, safely continuing the landing mission.

[0055] This step endows the system with self-protection and adaptive recovery capabilities under extreme and harsh conditions, significantly improving the overall system robustness and task completion rate.

[0056] Furthermore, such as Figure 2 As shown, the present invention also provides a drone landing control system, comprising: The status monitoring module is used to monitor the attitude changes and actuator actions of the dynamically compensated sea surface take-off and landing platform. If the update conditions are met, it calculates and generates status data representing the dynamic spatial constraints of the platform's physical structure based on the platform's 3D model and kinematic model, as a status snapshot. This module is the system's data acquisition and status triggering unit. The versioned sequence construction module is used to assign a monotonically increasing unique version number to each generated state snapshot, forming a versioned constraint state sequence. This module is the system's state version management unit. The planning request response module is used to respond to the landing trajectory planning request from the UAV and obtain the state snapshot version number declared in the request as the version number to be verified. This module is the interface unit for collaborative interaction between the system and the UAV. The version consistency verification module is used to compare the version number to be verified with the latest version number in the currently released constraint state sequence. This module is the core unit for the system to make collaborative logic decisions. The authorization and execution module is used to allow the UAV to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number only when the version number to be verified is consistent with the latest version number. This module executes the corresponding control branch according to the verification result of the version consistency verification module. The process termination module is used to return a response indicating version inconsistency to the drone terminal if the version number to be verified is earlier than the latest version number, and terminate the landing process based on the version number to be verified. This module is responsible for handling the safety measures when the versions are inconsistent.

[0057] During system operation, the status monitoring module and the versioned sequence construction module constitute the status publishing end. Once the platform experiences a meaningful shock or compensation, the status monitoring module generates a new status snapshot, and the versioned sequence construction module immediately assigns it a new version number and updates the sequence, forming a new factual version.

[0058] The planning request response module, version consistency verification module, authorization and execution module, and process termination module constitute the request adjudication end. When the UAV initiates a planning request, the planning request response module receives and extracts its version declaration; the version consistency verification module compares the declared version with the latest version of the system; if they match, the authorization and execution module allows the request; if they do not match, the process termination module intervenes to terminate the process and provides feedback.

[0059] Furthermore, the present invention also provides a drone take-off and landing platform, including a platform body, a dynamic compensation actuator disposed on the platform body, a sensor for monitoring the platform attitude, and a control unit.

[0060] The platform body is the main mechanical structure of the take-off and landing platform, usually a rigid frame or deck, which forms the supporting foundation for the final landing plane of the UAV and the surrounding physical structures (such as guardrails and hangar frames).

[0061] The dynamic compensation actuator is the core power component for achieving active stability of the platform. It is usually composed of multiple (such as three or four) high-performance servo electric cylinders, one end of which is hinged to the ship deck (or intermediate transition base), and the other end is hinged to a specific point on the platform body. The dynamic compensation actuator receives instructions from the control unit and, through the precise and rapid extension and retraction movements of each electric cylinder, jointly drives the platform body to perform multi-degree-of-freedom translation and rotation in space. This actively counteracts the swaying caused by the ship's rolling, pitching, heave, and other movements, ensuring that the landing plane on the platform body remains basically horizontal and stable in the inertial coordinate system.

[0062] The sensors used to monitor the platform's attitude are sensing units that perceive the motion state of the platform and ship. They mainly include high-precision inertial measurement units (IMUs). The IMUs are fixedly installed on the platform body or in parts rigidly connected to it, and are used to measure the attitude angles (roll angle, pitch angle) of the platform body relative to inertial space, as well as information such as angular velocity and acceleration in real time. These sensor data are the basic inputs for the control unit to drive the dynamic compensation actuator for compensation control, and are also the key basis for determining whether the update conditions are met.

[0063] The control unit is the coordinating hub of the UAV take-off and landing platform. The control unit is an embedded computer system containing a processor, memory and necessary input / output interfaces, which is specifically configured (i.e. through hardware logic design or by loading and running specific software programs) to execute the UAV landing control method.

[0064] Furthermore, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a drone landing control method.

[0065] A computer-readable storage medium is a non-transient physical storage medium, such as flash memory, read-only memory (ROM), solid-state drive (SSD), optical disk, etc., on which a computer program is stored. Here, "storage" means that the computer program is persistently recorded on the medium in the form of instruction code that can be recognized and executed by a processor. The computer program contains a series of instructions for controlling the processor to perform specific operations.

[0066] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling the landing of an unmanned aerial vehicle (UAV), characterized in that, Includes the following steps: Status monitoring steps: Monitor the attitude changes of the take-off and landing platform and the actions of the actuators. If the update conditions are met, calculate and generate status data representing the dynamic spatial constraints of the platform's physical structure based on the platform's three-dimensional model and kinematic model, as a status snapshot. Versioned sequence construction steps: Assign a monotonically increasing unique version number to each generated state snapshot to form a versioned constraint state sequence; Planning request response steps: Based on the landing trajectory planning request from the UAV, obtain the state snapshot version number declared in the request as the version number to be verified; Version consistency verification steps: Compare the version number to be verified with the latest version number in the currently released constraint state sequence; Authorization and execution steps: The UAV is allowed to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number only when the version number to be verified is consistent with the latest version number; Process termination step: If the version number to be verified is earlier than the latest version number, a response indicating a version inconsistency is returned to the drone terminal, and the landing process based on the version number to be verified is terminated.

2. The UAV landing control method according to claim 1, characterized in that, The update conditions include the absolute value change of the roll or pitch angle of the take-off and landing platform exceeding the angle threshold, or the extension / retraction change of any of its actuators exceeding the displacement threshold. The generation of the state snapshot specifically includes: calculating the real-time coordinate set of at least one physical structure in space, such as the platform guardrail and the hangar roof; obtaining the current travel distance and remaining travel margin of each actuator; and calculating the real-time minimum distance between the UAV and the physical structure.

3. The UAV landing control method according to claim 1, characterized in that, The planning request response step and the process termination step are executed by the status version controller deployed on the take-off and landing platform side; In the planning request response step, the status version controller receives a planning request from the UAV, and the planning request encapsulates the version number to be verified. In the process termination step, when the status version controller determines that the version number to be verified is earlier than the latest version number, it carries the latest version number in the returned response indicating that the version status is inconsistent.

4. The UAV landing control method according to claim 1, characterized in that, Following the authorization and execution steps, a state locking step is also included: Obtain the altitude of the UAV relative to the landing surface of the take-off and landing platform; When the height first drops below a preset locked height threshold, the current latest version number is marked as the locked version number; During the period from when the altitude is below the locked altitude threshold until the end of this landing attempt, the execution of the status monitoring step and the versioned sequence construction step is suspended, and for subsequent planning requests initiated by the UAV, the authorization and execution steps are both based on the status snapshot data corresponding to the locked version number. During the locking period, if the remaining travel margin of the actuator of the take-off and landing platform is determined to be lower than the limit threshold based on the status snapshot corresponding to the locking version number, the process termination step is executed to terminate the current landing process and control the UAV to enter the hovering state.

5. The UAV landing control method according to claim 2, characterized in that, When the UAV performs landing trajectory planning based on the state snapshot, the minimum safe distance threshold used is dynamically determined according to the remaining travel margin obtained from the state snapshot; The rule is that the average value of the remaining travel margin is negatively correlated with the minimum safe distance threshold.

6. The UAV landing control method according to claim 1, characterized in that, Following the versioned sequence construction step, a data fusion processing step is also included: Monitor the update frequency of the version number of the aforementioned status snapshot; When the update frequency exceeds a preset frequency threshold, the platform physical structure spatial location data and actuator state data in multiple state snapshots generated within a continuous time period are fused and calculated to generate a state snapshot representing the comprehensive constraint state within that time period. Assign a new version number to the state snapshot of the comprehensive constraint state and update the constraint state sequence.

7. The UAV landing control method according to claim 1, characterized in that, Following the process termination step, an exception handling step is also included: If the drone receives more than a preset number of inconsistent response versions of the instruction within a preset time period, the drone is controlled to ascend to a specified safe altitude and enter a hovering waiting state. During the hovering wait period, continuously monitor the version update events of the constraint state sequence; Only when a version update event is detected and the time interval between two consecutive updates is greater than a set stability threshold, the drone is controlled to re-initiate a landing trajectory planning request based on the latest version number.

8. A drone landing control system for implementing the drone landing control method according to any one of claims 1-7, characterized in that, include: The status monitoring module is used to monitor the attitude changes and actuator actions of the dynamically compensated sea surface take-off and landing platform. If the update conditions are met, it calculates and generates status data representing the dynamic spatial constraints of the platform's physical structure based on the platform's 3D model and kinematic model, as a status snapshot. The versioned sequence building module is used to assign a monotonically increasing unique version number to each generated state snapshot, forming a versioned constraint state sequence. The planning request response module is used to respond to the landing trajectory planning request from the UAV and obtain the state snapshot version number declared in the request as the version number to be verified. The version consistency verification module is used to compare the version number to be verified with the latest version number in the currently released constraint state sequence. The authorization and execution module is used to allow the UAV to perform landing trajectory planning or control based on the state snapshot corresponding to the latest version number only when the version number to be verified is consistent with the latest version number; The process termination module is used to return a response indicating a version inconsistency to the drone terminal if the version number to be verified is earlier than the latest version number, and terminate the landing process based on the version number to be verified.

9. A drone take-off and landing platform, characterized in that, The system includes a platform body, a dynamic compensation actuator mounted on the platform body, sensors for monitoring the platform's attitude, and a control unit configured to perform the method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the method as described in any one of claims 1-7.

Images