A Distributed Architecture-Based Cooperative Integrated Control Method and System for Unmanned Aerial Vehicles

Abstract

This application relates to the field of unmanned aerial vehicle (UAV) control technology, and more particularly to a UAV collaborative integrated control method and system based on a distributed architecture. The proposed scheme involves constructing a near-field three-dimensional wind field model of the mother aircraft based on multi-source sensor data from the mother aircraft and target wingmen, and extracting streamlines from the wind field that satisfy flight maneuverability, parameterizing them as aerodynamic trajectories for waiting, approach, and departure. The aerodynamic trajectories are then time-parameterized to generate a time slot sequence composed of spatial position and time stamps. Based on the wingmen's current position, battery level, and mission priority, target trajectories and target time slots are assigned to the wingmen, enabling them to achieve controllable approach, payload recovery, or in-flight takeoff in complex near-field wind fields. This application can achieve high-safety-separation collaborative operation of multiple wingmen, improving the operational efficiency and stability of the airborne homeport.

Description

Technical Field

[0001] This application relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to a UAV collaborative integrated control method and system based on a distributed architecture. Background Technology

[0002] In large unmanned aerial vehicle (UAV) platforms, to improve mission coverage and continuous operation capabilities, a coordinated aerial approach involving a mother aircraft carrying multiple wingmen is often employed. However, in existing technologies, the flight organization of wingmen in the near-field area of ​​the mother aircraft generally relies on fixed tracks, preset routes, or simple obstacle avoidance strategies based on individual aircraft status. This makes it difficult to maintain flight stability in environments with complex near-field disturbances and significant structural airflow changes, easily leading to increased attitude fluctuations, increased control load, or track deviations. On the other hand, existing mother-wingman coordination relies on static task allocation or sequential scheduling methods. When multiple wingmen simultaneously perform approach, recovery, or in-flight release, it can easily cause overlapping execution sequences, trajectory conflicts, or excessively long waiting times, affecting overall operational efficiency. Furthermore, in existing solutions, the approach and takeoff processes of wingmen lack dynamic path planning capabilities that can adapt to changes in the near-field wind field. The controllability of wingmen is insufficient when approaching structures of the mother aircraft, and it is difficult to consistently guarantee safe distances.

[0003] To address the above issues, this application presents a collaborative integrated control method and system for unmanned aerial vehicles (UAVs) based on a distributed architecture. Summary of the Invention

[0004] The technical problem this application aims to solve is to address the shortcomings of existing technologies by providing a collaborative integrated control method and system for unmanned aerial vehicles (UAVs) based on a distributed architecture. This method constructs a near-field three-dimensional wind field model of the mother aircraft based on multi-source sensor data from both the mother aircraft and the target wingman. Streamlines that meet flight maneuverability requirements are extracted from the wind field and parameterized into aerodynamic trajectories for holding, approach, and departure. Furthermore, the aerodynamic trajectories are time-parameterized to generate a time slot sequence composed of spatial position and time stamps. Based on the wingman's current position, battery level, and mission priority, a target trajectory and target time slot are assigned to the wingman, enabling controllable approach, payload recovery, or in-flight takeoff in complex near-field wind fields.

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A distributed architecture-based UAV cooperative integrated control method is applied to an airborne homeport system including a mother aircraft and multiple wingmen operating around the mother aircraft. The method includes:

[0007] In the body coordinate system corresponding to the mother aircraft, a three-dimensional wind field model of the mother aircraft near the field is constructed based on the sensor data of the mother aircraft and each target wingman, wherein the target wingman is the wingman that needs to perform take-off, approach or recovery operations;

[0008] Based on the near-field three-dimensional wind field model of the mother aircraft, multiple streamlines adapted to preset constraints are extracted, and the streamlines are parameterized into aerodynamic tracks for wingman movement, including at least one waiting track, one approach track, and one departure track.

[0009] Each aerodynamic trajectory is time-parameterized to generate a time slot sequence consisting of spatial position and time label;

[0010] Based on the target wingman's current location, battery level, and mission priority, and in conjunction with the time slot sequence, a corresponding target track and target time slot are assigned to the target wingman to generate control commands for the target wingman.

[0011] The sensor data includes wind speed, wind direction, air pressure, acceleration, and attitude information. The construction of a near-field three-dimensional wind field model of the mother aircraft based on the sensor data from the mother aircraft and each target wingman includes:

[0012] Acquire first sensor data of the target wingman in the airspace outside the mother aircraft, perform spatiotemporal alignment and fusion of the first sensor data, and generate supplementary observation data;

[0013] The second sensor data of the target wingman located in the mother machine mounting chamber is obtained to obtain the boundary layer disturbance information of the mother machine body;

[0014] The supplementary observation data and the boundary layer disturbance information are used as external constraints. The sensor data corresponding to the mother machine are input into the preset wind field reconstruction model. Spatial interpolation and flow field calculation are performed on the near-field three-dimensional wind field of the mother machine to obtain the near-field three-dimensional wind field model of the mother machine.

[0015] The step of acquiring second sensor data from the target wingman located in the mothership mounting chamber to obtain boundary layer disturbance information of the mothership includes:

[0016] Denoising and trend analysis are performed on the wind speed, wind direction and air pressure data in the second sensor to obtain the local airflow change characteristics of the mother machine body;

[0017] Perturbation decomposition and feature extraction are performed on the acceleration and attitude information in the second sensor data to obtain the dynamic response characteristics of the mother machine body;

[0018] By associating and mapping the local airflow change characteristics and the dynamic response characteristics, boundary layer disturbance information is obtained.

[0019] Based on the near-field three-dimensional wind field model of the mother aircraft, multiple streamlines adapted to preset constraints are extracted, and these streamlines are parameterized into aerodynamic trajectories for wingman movement, including:

[0020] Based on the near-field three-dimensional wind field model of the mother machine, continuous wind field paths that meet preset constraints in terms of updraft intensity, lateral disturbance amplitude, and turbulence threshold are identified.

[0021] By performing flow direction integration along the continuous wind field path, multiple three-dimensional spatial streamlines for wingman movement are extracted.

[0022] Curvature smoothing and spatial resampling are performed on each of the three-dimensional spatial streamlines to obtain an aerodynamic trajectory point sequence that meets the maneuverability requirements of wingman flight;

[0023] The aerodynamic trajectory point series is parameterized according to the spatial position in the mother machine body coordinate system, and the parameterized aerodynamic trajectory point series is divided into multiple aerodynamic tracks based on the trajectory phase diagram partitioning.

[0024] The trajectory phase diagram-based partitioning divides the parameterized aerodynamic trajectory point series into multiple aerodynamic orbits, including:

[0025] Based on the spatial position of each trajectory point in the aerodynamic trajectory point sequence in the coordinate system of the mother machine, the radial distance sequence and the relative height sequence relative to the mother machine are calculated, and the radial distance sequence and the relative height sequence are used as the spatial coordinates of the trajectory phase diagram;

[0026] The recommended flight speed and relative altitude corresponding to the aerodynamic trajectory point series are determined by a preset trajectory lookup table sequence, the equivalent energy sequence is calculated, and the equivalent energy sequence is used as the state coordinates of the trajectory phase diagram. The trajectory lookup table sequence is determined by matching the flight speed range and altitude range of different spatial positions based on the wingman flight performance parameters and the near-field three-dimensional wind field model of the mother aircraft.

[0027] Based on the changing trends of the radial distance sequence, relative height sequence, and equivalent energy sequence in the trajectory phase diagram, the parameterized aerodynamic trajectory points are partitioned into phase diagrams. Aerodynamic trajectory points that converge in the radial direction and have decaying equivalent energy are classified as approach tracks. Aerodynamic trajectory points that periodically change within a preset range of radial distance and relative height and have constant equivalent energy are classified as waiting tracks. Aerodynamic trajectory points that diverge in the radial direction and have increasing equivalent energy are classified as departure tracks.

[0028] The step of time parameterizing each aerodynamic orbit to generate a time slot sequence consisting of spatial position and time label includes:

[0029] For each aerodynamic track, based on the spatial point array of the aerodynamic track in the coordinate system of the mother machine, the spatial points in the spatial point array are sorted according to the trajectory advancement direction to obtain an ordered trajectory sequence. The trajectory advancement direction corresponding to the waiting track is the local closed loop direction generated along the spatial point array, the trajectory advancement direction corresponding to the approach track is the convergence direction relative to the mother machine's center of mass toward the mother machine's loading area, and the trajectory advancement direction corresponding to the departure track is the divergence direction relative to the mother machine's center of mass toward the mother machine's external airspace.

[0030] Based on the spatial distance between spatial points in the trajectory sequence and the recommended flight speed of the corresponding spatial point, the target arrival time of each spatial point is calculated, and the target arrival time is used as the time parameter of the spatial point.

[0031] Time discretization is performed on spatial points with time parameters, and continuous time parameters are divided into multiple time intervals to obtain time slots composed of time intervals and corresponding spatial positions. The time slots are used to represent the spatial operation window of the target wingman on the aerodynamic track.

[0032] Based on the time sequence of the time slots, a time slot sequence consisting of multiple time slots is constructed on each of the pneumatic tracks.

[0033] Using the arrival time of the target as a time parameter for a spatial point includes:

[0034] Based on the target arrival time, an original arrival time sequence arranged along the trajectory sequence is obtained;

[0035] Based on the time difference between adjacent spatial points in the original arrival time sequence, spatial points with a time difference less than a preset time merging threshold are time-aggregated to obtain merged spatial points and their corresponding updated arrival times. The updated arrival times are used as the time parameters of the merged spatial points, wherein the updated arrival times are determined by calculating the average value of the target arrival times.

[0036] For spatial points that are not time-aggregated, their corresponding target arrival time is retained as a time parameter.

[0037] If the target wingman corresponds to an approach or recovery operation, the method includes:

[0038] Based on the current position of the target wingman and the spatial distribution of the aerodynamic trajectories, candidate aerodynamic trajectories that the target wingman can reach are determined;

[0039] Based on the remaining battery power of the target wingman and the energy consumption estimate corresponding to the candidate aerodynamic trajectories, the candidate aerodynamic trajectories are screened to obtain a set of optional aerodynamic trajectories;

[0040] Based on the occupancy status of each time slot in the time slot sequence, the minimum safe distance requirement with other wingmen, and the mission priority of the target wingmen, a target track that meets the safe distance constraint is selected from the set of optional aerodynamic tracks.

[0041] In the time slot sequence corresponding to the target trajectory, select the target time slot that satisfies the target wingman's current position, flight speed range, and dynamic constraints, and use the target time slot as the target wingman's operating window.

[0042] If the target wingman corresponds to a takeoff operation, the method includes:

[0043] Based on the spatial position of the mounting area in the coordinate system of the mother aircraft, a releaseable takeoff area matching the mounting area is determined, and candidate departure tracks that are spatially continuous with the releaseable takeoff area are selected from the aerodynamic tracks.

[0044] Based on the occupancy status of each time slot in the time slot sequence corresponding to the candidate departure track set, and the minimum safe interval between the time slot and other wingman operation windows, a target departure track that meets the safe distance and release conditions is selected.

[0045] In the time slot sequence of the target departure track, the target time slot is determined from the time slots that satisfy the target wingman's release attitude constraints and acceleration capability constraints, and the target time slot is used as the takeoff window of the target wingman.

[0046] A distributed architecture-based collaborative integrated control system for unmanned aerial vehicles (UAVs), the system comprising:

[0047] The wind field perception and reconstruction module is used to acquire sensor data from the mother aircraft and each target wingman, process the sensor data of the target wingman located in the airspace outside the mother aircraft and in the mother aircraft mounting chamber, and construct a three-dimensional wind field model of the mother aircraft in the near field.

[0048] The aerodynamic trajectory generation module is used to extract continuous wind field paths that meet preset constraints based on the near-field three-dimensional wind field model of the mother machine, perform curvature smoothing and spatial resampling on the extracted three-dimensional spatial streamlines to obtain aerodynamic trajectory point series, and divide the aerodynamic trajectory point series into waiting track, approach track and departure track based on trajectory phase diagram partitioning.

[0049] The time slot management module is used to parameterize the aerodynamic trajectory points for each aerodynamic orbit, calculate the target arrival time of spatial points, and perform time aggregation and time discretization to generate a time slot sequence composed of spatial location and time label.

[0050] The collaborative scheduling and control module is used to determine the target trajectory and target time slot of the target wingman based on the target wingman's current location, remaining battery power, and task priority, combined with the time slot sequence, and to generate control instructions to be issued to the target wingman based on the target trajectory and target time slot.

[0051] Compared with the prior art, the beneficial effects of this application are:

[0052] This application constructs a three-dimensional wind field model of the mother aircraft near the wind field, extracts flyable streamlines from the wind field and parameterizes them into aerodynamic trajectories, and then combines them with time slot sequences to achieve coordinated scheduling of multiple wingmen. This enables the flight behavior of wingmen in the complex aerodynamic environment near the mother aircraft to be predictable and organized. It can form dynamic paths that conform to the wind field structure in key stages such as approach, recovery and departure, so that wingmen can maintain high attitude stability and controllability under wind field changes. Attached Figure Description

[0053] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0054] Figure 1 An exemplary application scenario diagram provided for an embodiment of this application;

[0055] Figure 2 A schematic diagram of a module of a UAV collaborative integrated control system based on a distributed architecture, provided for an embodiment of this application;

[0056] Figure 3 A flowchart illustrating a UAV collaborative integrated control method based on a distributed architecture, provided in an embodiment of this application;

[0057] Figure 4 A schematic diagram illustrating the classification of pneumatic tracks provided in the embodiments of this application;

[0058] Figure 5 This is a schematic diagram illustrating the principle of time aggregation provided in the embodiments of this application. Detailed Implementation

[0059] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments.

[0060] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0061] In current drone swarm applications, the combination of large fixed-wing platforms and multiple small drones is gradually moving from proof-of-concept to engineering deployment.

[0062] Taking large-scale power line inspections, cross-regional oil and gas pipeline inspections, rapid post-disaster assessments, and emergency communication relays as examples, traditional methods often rely on a large number of small multi-rotor or small fixed-wing aircraft taking off and landing in batches on the ground to replenish power. This is constrained by ground support conditions and makes it difficult to maintain continuous coverage in complex terrain or disaster-stricken areas. In engineering practice, the following concept naturally emerged:

[0063] Large fixed-wing UAVs with longer endurance and range serve as aerial homeports, providing takeoff, release, in-flight recovery, and refueling support for multiple small wingmen, thereby extending the operational radius from near the ground takeoff and landing point to any location within a wide airspace.

[0064] It's important to note that the real constraint on the implementation of the airborne homeport solution is not simple mission planning or link scheduling, but rather the complex airspace near the mother aircraft—a small area where aerodynamics and spatiotemporal coupling are highly integrated. During cruise, turns, and takeoffs / landings, large fixed-wing UAVs generate highly non-uniform three-dimensional near-field wind fields around their wings, fuselages, and pylons. Small wingmen, frequently performing takeoffs, approaches, or recoveries near the mother aircraft, must navigate through this complex wind field. Traditional swarm control often employs geometric formation, fixed airspace division, or conflict detection based on preset routes, assuming the wind field around the mother aircraft is uniform and controllable. This approach only imposes interval constraints and altitude stratification on a two-dimensional plane, making it difficult to guarantee that wingmen can maintain sufficient safety margins while efficiently queuing for takeoffs and landings under actual near-field aerodynamic disturbances.

[0065] Taking rapid post-disaster assessment as an example, multiple wingmen need to make frequent trips to the mother aircraft to transmit data and recharge. If management is carried out solely through simple time-sharing or static waiting airspace division, on the one hand, it is easy for some wingmen to circle and wait for a long time with limited power, posing a risk of forced landing; on the other hand, the near-field wind field of the mother aircraft changes in real time with attitude and flight status, and the statically planned waiting area, approach channel and departure channel are very likely to overlap with the surging wake and downwash, resulting in wingmen attitude shaking, tracking difficulties or even near stall.

[0066] It is important to emphasize that once an airborne homeport architecture is adopted, the area around the mother aircraft is no longer a simple controlled airspace, but rather a dynamic aerodynamic field that constantly deforms with the movement of the mother aircraft and changes with the wind field. Existing collaborative control methods based on geometric rules and fixed airspace divisions are difficult to balance between safety and efficiency.

[0067] At the system level, existing solutions typically separate near-field aerodynamics from takeoff and landing scheduling in order to control complexity:

[0068] The flight control side tries to suppress attitude disturbances within a local area, while the scheduling side manages the takeoff and landing order of wingmen based solely on time queues and priorities. This decoupling approach can be applied to single aircraft or small-scale clusters, but when there are multiple approaching wingmen, multiple wingmen waiting to take off, and several wingmen in a waiting state around the mother aircraft, simple time sequencing is no longer sufficient to reflect the feasibility and risk level of each wingman in the actual wind field, nor can it provide precise aerodynamic channels for the wingmen. This makes it difficult for the airborne mothership system to scale up to the scale and load required for the project.

[0069] This application proposes a trajectory construction and timing assignment approach for aerodynamic environments:

[0070] Instead of drawing approach lines and waiting areas around the mother aircraft based on prior experience, the system starts from the reconstructed near-field three-dimensional wind field. Streamlines are extracted along continuous wind field paths that satisfy constraints such as updraft intensity, lateral disturbance amplitude, and turbulence intensity. These streamlines are smoothed and resampled, then treated as naturally tailwind-friendly, steady-state aerodynamic trajectories. Combining this with the radial distance, relative altitude, and equivalent energy changes of the wingman relative to the mother aircraft, the trajectories are assigned roles, spontaneously evolving into different track types more suitable for waiting, approach, or departure. The mother aircraft's near-field is no longer just a geometrically defined airspace, but is reconstructed into a network of aerodynamic tracks that dynamically adapt to the wind field and mission requirements.

[0071] Furthermore, based on the aerodynamic trajectory, this application further discretizes the time dimension of the wingman's movement along the trajectory. It compresses, aggregates, and discretizes the spatial points on the trajectory and the target arrival times corresponding to the recommended speeds, forming a sequence of time slots jointly defined by spatial location and time labels. Each time slot can be understood as a local spatiotemporal operating window on a specific aerodynamic trajectory, constraining both spatial location and implicitly requiring matching with wind field conditions and the wingman's dynamic capabilities. When the mother aircraft performs coordinated scheduling, it no longer determines the launch or recovery order solely based on task priority and simple time queues. Instead, it jointly matches the target wingman's current position, remaining battery power, and dynamic performance with the aforementioned time slots, allocating target trajectories and target time slots that balance aerodynamic safety, energy feasibility, and task priority. This enables precise coordination of multi-wingman takeoff, approach, and recovery processes in an airborne homeport scenario.

[0072] refer to Figure 1 , Figure 1 This is an exemplary application scenario diagram provided for an embodiment of this application.

[0073] Figure 1 The application scenario shown includes a mother aircraft, wingman one, wingman two, wingman three, and wingman four, with wingman four located in the mounting area of ​​the mother aircraft.

[0074] Figure 1 The image shows wingman 1 and wingman 2 sending a refueling request to the mothership as target wingmen.

[0075] Understandably, in practical applications, the mother aircraft can carry multiple wingmen and perform in-flight refueling, release, recovery, and mission redistribution operations; the wingmen can perform tasks such as inspection, mapping, emergency detection, or communication relay in the external airspace. When a wingman requests to return to the vicinity of the mother aircraft due to insufficient power, mission switching, or the need for data transmission, it needs to complete approach and docking maneuvers in the complex wind field environment near the mother aircraft.

[0076] In the specific implementation not shown in the figure, during cruise, turns, or attitude changes, the mother aircraft often experiences a three-dimensional non-uniform wind field on its fuselage surface, under its wings, and near its payload area, formed by the superposition of downwash, wake vortices, and local turbulence. For smaller wingmen with limited power margins, if traditional fixed approach channels or simple distance-keeping strategies are still used, attitude instability, excessive speed fluctuations, or failure to recover within the limited battery capacity can easily occur under the aforementioned wind field disturbances. Therefore, the flight status of the mother aircraft itself and the operational status of multiple wingmen often create a dynamic coupling effect in the near-field airspace, making traditional cooperative control methods based on static airspace division unable to meet the safety and real-time requirements of the airborne homeport scenario.

[0077] It is important to note that Figure 1 The number of wingmen and the number of wingmen that can be mounted in the mounting area are merely illustrative examples to facilitate understanding of the air homeport application scenarios involved in the embodiments of this application.

[0078] Those skilled in the art will understand that the mother aircraft can be configured with one or more mounting areas depending on its own airframe structure, mission payload capacity, and pylon layout. Each mounting area can also accommodate one or more wingmen depending on its spatial dimensions and mechanical structure design. In actual deployment, the number, type, and mission allocation method of wingmen can be flexibly adjusted according to application requirements, operational airspace restrictions, and mission duration.

[0079] refer to Figure 2 , Figure 2 This is a schematic diagram of a distributed architecture-based collaborative integrated control system for unmanned aerial vehicles (UAVs) provided in an embodiment of this application.

[0080] In one example, the system includes:

[0081] The wind field perception and reconstruction module is used to acquire sensor data from the mother aircraft and each target wingman, process the sensor data of the target wingman located in the airspace outside the mother aircraft and in the mother aircraft mounting chamber, and construct a three-dimensional wind field model of the mother aircraft in the near field.

[0082] The aerodynamic trajectory generation module is used to extract continuous wind field paths that meet preset constraints based on the near-field three-dimensional wind field model of the mother machine, perform curvature smoothing and spatial resampling on the extracted three-dimensional spatial streamlines to obtain aerodynamic trajectory point series, and divide the aerodynamic trajectory point series into waiting track, approach track and departure track based on trajectory phase diagram partitioning.

[0083] The time slot management module is used to parameterize the aerodynamic trajectory points for each aerodynamic orbit, calculate the target arrival time of spatial points, and perform time aggregation and time discretization to generate a time slot sequence composed of spatial location and time label.

[0084] The collaborative scheduling and control module is used to determine the target trajectory and target time slot of the target wingman based on the target wingman's current location, remaining battery power, and task priority, combined with the time slot sequence, and to generate control instructions to be issued to the target wingman based on the target trajectory and target time slot.

[0085] Next, with reference to the accompanying drawings, we will further describe the UAV cooperative integrated control method based on a distributed architecture provided in this application. Figure 3 The method shown is applied to an airborne homeport system comprising a mother aircraft and multiple wingmen operating around the mother aircraft, the method comprising:

[0086] S1: In the body coordinate system corresponding to the mother machine, construct a near-field three-dimensional wind field model of the mother machine based on the sensor data of the mother machine and each target wingman;

[0087] The target wingman is the wingman that needs to perform takeoff, approach, or recovery operations;

[0088] In this embodiment, the core of constructing the near-field three-dimensional wind field model of the mother aircraft lies in making full use of the multi-source sensing capabilities of the mother aircraft and the target wingman. The mother aircraft can collect airflow parameters around itself through its onboard devices such as pitot tubes, barometers, wind vanes, and inertial navigation units; while the target wingman at different positions can supplement the local flow field characteristics that the mother aircraft cannot directly perceive.

[0089] For example, a wingman that has not yet entered the mounting area but is about to perform an approach maneuver can collect more fine-grained disturbance structures in the wake region of the mother aircraft, while a wingman that is already in the mounting chamber waiting to be released can reflect the changing trend of boundary layer disturbances on the surface of the mother aircraft.

[0090] S2: Based on the near-field three-dimensional wind field model of the mother machine, extract multiple streamlines that are adapted to the preset constraints, and parameterize the streamlines into aerodynamic trajectories for the movement of the wingman;

[0091] The pneumatic track includes at least one waiting track, one approach track, and one departure track;

[0092] In this embodiment, the key to extracting aerodynamic trajectories lies in making the passable areas in the wind field explicit, rather than having the wingman hook up in areas with significant disturbances. Based on the three-dimensional wind field distribution, continuous flow paths more suitable for wingman flight can be selected along dimensions such as wind direction continuity, disturbance amplitude, and updraft / downdraft gradient.

[0093] For the airborne homeport scenario in this application, streamline extraction not only needs to identify stable paths, but also needs to ensure that the paths have clear spatial meaning in the coordinate system of the mother aircraft.

[0094] Understandably, streamlines that are close to the mounting area and tend to converge are naturally suitable as approach trajectories; local circumferential paths located below the mother aircraft and with less disturbance can serve as holding tracks; and paths that tend to diverge and are far from the wake of the mother aircraft can serve as departure tracks after the wingman takes off.

[0095] Furthermore, through curvature smoothing, resampling, and spatial parameterization, the streamlines are further transformed into a series of aerodynamic trajectory points that are easy for flight control to execute. This embodiment uses phase diagram features as the basis for trajectory classification, so that each trajectory is not just a geometric line, but a dynamic channel that matches the wingman's dynamic capabilities and wind field characteristics in a physical sense. In this way, the wingman no longer directly confronts unfavorable wind areas, but instead follows the trajectory allowed by the wind field to complete approach, holding, or departure, significantly reducing attitude disturbances and control load, and improving mission continuity.

[0096] S3: Perform time parameterization on each aerodynamic trajectory to generate a time slot sequence consisting of spatial position and time label;

[0097] In this embodiment, the purpose of time parameterization is not only to give the orbits temporal meaning, but also to enable each orbit to naturally accommodate the occupancy window of wingmen, thereby providing a calculable basis for multi-wingman scheduling. To this end, this embodiment calculates the corresponding target arrival time for each trajectory point on the aerodynamic orbit based on spatial distance and recommended flight speed. Since the distance between some trajectory points is extremely small, their independent arrival times are meaningless at the scheduling level; therefore, they are aggregated based on a time difference threshold to obtain key times that better represent the macroscopic temporal structure of the trajectory.

[0098] S4: Based on the target wingman's current location, battery level, and mission priority, and in conjunction with the time slot sequence, assign the target wingman a corresponding target track and target time slot to generate control commands for the target wingman;

[0099] In this embodiment, the core idea of ​​the scheduling strategy is to allow each wingman to enter the appropriate wind field structure at the appropriate time, rather than forcibly assigning it to a fixed channel. To this end, the most suitable target trajectory and target time slot are determined by comprehensively considering the target wingman's current position, battery status, flight performance, and mission priority, combined with the occupied time slots of each aerodynamic trajectory. For approach or recovery requests, the focus is on whether the wingman can reach the approach trajectory start point within the available battery power; for takeoff or departure requests, it is necessary to ensure that the trajectory segment corresponding to the takeoff moment has sufficient aerodynamic stability and safety margin.

[0100] Furthermore, once the target trajectory and target time slot are determined, control commands containing waypoints, speed instructions, attitude constraints, and time windows can be generated, enabling wingmen to achieve refined coordinated flight in dynamic wind fields. This avoids the problem of traditional scheduling that only considers time and not aerodynamics, allowing scheduling decisions to incorporate wind field information and improving the reliability of executing multiple wingmen takeoffs and landings in complex near-field disturbances of the mother aircraft.

[0101] When facing multi-winger collaborative missions performed by large fixed-wing platforms, it is necessary to explain the operational characteristics of the air homeport scenario before introducing the specific steps in the embodiments of this application, so as to more clearly understand the design logic of each subsequent step.

[0102] In engineering deployments involving multiple types of drones working together, the process of recovering, releasing, and recharging wingmen is often analogized to takeoff and landing scheduling at a ground airport. This assumes that as long as several flight paths are designated around the mother drone and conventional obstacle avoidance strategies are employed, the controllability of the swarm operation can be maintained. However, extensive flight testing and wind field simulations reveal that the operating environment of an aerial homeport differs fundamentally from traditional airspace:

[0103] The wings, payload cavities, and wake region of the mother aircraft during flight together constitute a local aerodynamic field that deforms over time. Its disturbance structure not only has strong three-dimensionality, but also causes unexpected flow direction shifts when the mother aircraft's attitude changes and external wind field fluctuations are superimposed.

[0104] It is important to note that when wingmen approach or depart from the mother aircraft, their small size and limited power margin make them particularly sensitive to local turbulence, lateral disturbances, or instantaneous downwash. Once approaching or departing along a conventional pre-set route, the UAV frequently experiences speed loss, frequent attitude corrections, and even being sucked into the wake turbulence. To address this issue, existing technologies have attempted to improve stability by employing more aggressive attitude control algorithms or increasing safety intervals. However, the former is prone to causing control oscillations, while the latter significantly reduces the airspace utilization efficiency around the mother aircraft, failing to meet the demands of high-frequency, continuous operations with multiple wingmen.

[0105] Based on this understanding, this embodiment reconstructs the three-dimensional wind field near the mother aircraft using the distributed sensing capabilities of multiple UAVs, making the previously invisible flow direction, lateral vortex field, and disturbance gradient present in a computable form. Subsequently, spatial paths with good continuity and relatively mild disturbance amplitudes are identified in the three-dimensional wind field structure and transformed into aerodynamic tracks that can be directly used by wingmen, enabling them to perform actions with the wind field rather than against the disturbance when completing approach, holding, or departure tasks. In addition, to resolve potential time conflicts that may arise when multiple wingmen operate simultaneously in the near-field airspace, this embodiment further introduces time constraints into the tracks, expressing them in the form of time slots composed of time sequence labels, so that the available airspace and usable time for each wingman are clearly defined.

[0106] Next, we will further elaborate on the technical content of the near-field three-dimensional wind field model of the mother machine in this application.

[0107] It should be noted that the sensor data described in this application includes wind speed, wind direction, air pressure, acceleration, and attitude information. The specific methods for collecting sensor data can be achieved through lidar, micro-meteorological probes, inertial navigation units, wind and optical flow modules, or flight control sensor components integrated into the mother aircraft and wingman aircraft. This application does not limit the specific methods used.

[0108] In one example, the construction of the near-field three-dimensional wind field model of the mother aircraft based on sensor data from the mother aircraft and each target wingman includes:

[0109] S1.1: Acquire the first sensor data of the target wingman in the airspace outside the mother aircraft, perform spatiotemporal alignment and fusion on the first sensor data, and generate supplementary observation data;

[0110] Specifically, target wingmen located in the airspace outside the mother aircraft will collect airflow information at different positions, altitudes, and inflow angles during flight. Since these wingmen have independent trajectories relative to the mother aircraft, their sensor sampling time bases, reference coordinate systems, and attitude references are all different. If these are directly used as input for wind field reconstruction, spatial misalignment or contradictions in flow field directionality will occur due to inconsistent data sources. Therefore, it is necessary to unify the timestamp base, unify the spatial reference coordinates, and correct for attitude differences to give these data a unified structure suitable for spatial fusion.

[0111] In this embodiment, the first sensor data collected by the wingman in the external airspace includes wind speed vector, wind direction angle, local pressure gradient, airframe acceleration, and attitude Euler angles. To make the data usable in the host aircraft's coordinate system, the original sampling timestamps of the wingman are first regressed and corrected using the global time reference broadcast by the host aircraft, ensuring a unified time reference for the data among different wingmen. Secondly, using the inertial navigation attitude calculation results adjusted on the wingman, the wind field data measured by each sensor is transformed from the wingman's own coordinate system to the host aircraft's coordinate system. During the transformation, a quaternion rotation matrix is ​​used to accurately map each directional component. Subsequently, based on the relative position, relative speed, and communication link delay between the host aircraft and the wingman, the spatial sampling points are spatiotemporally reprojected, allowing the sampling points of multiple wingmen at the same time to be mapped to the actual airflow location in the host aircraft's coordinate system. After the above processing, the airflow data collected by each wingman forms a set of supplementary sampling point cloud data covering the host aircraft's wake region, underwing region, and lateral disturbance region—that is, supplementary observation data.

[0112] S1.2: Obtain the second sensor data of the target wingman located in the mother machine mounting chamber to obtain the boundary layer disturbance information of the mother machine body;

[0113] Specifically, the target wingman located within the mother aircraft's mounting chamber is situated near the mother aircraft's boundary layer. The disturbance characteristics in these regions differ fundamentally from those in the external airspace. The boundary layer exhibits significant velocity decay, shear stress accumulation, and localized flow deflection, directly impacting the stability of the wingman during the final stage of recovery. Without modeling the micro-disturbances in these regions, the mother aircraft cannot accurately predict the airflow structure around the mounting area, potentially leading to insufficient control margin for the wingman as it approaches the platform.

[0114] In this embodiment, the second sensor data collected by the wingman in the mounted chamber mainly includes wind speed attenuation distribution, micro-pressure fluctuations, acceleration vibrations caused by low-speed shear flow, and attitude micro-shifts. Because the airflow velocity in this region is low and the frequency of change is high, the original sensor data contains significant high-frequency noise and periodic external disturbances. Therefore, it is first pre-filtered by frequency band boundaries to remove high-frequency random vibration components and retain effective aerodynamic disturbances. Subsequently, the pressure change trend along the aircraft surface is analyzed, and the wind speed attenuation rate is mapped to a position using a pre-defined aircraft geometry model, allowing the disturbance behavior within the boundary layer to be reconstructed. Simultaneously, to reflect the dynamic response within the boundary layer, this embodiment uses the wingman's acceleration change rate and attitude micro-shift trend to calculate the local flow direction change direction, ensuring that the model includes not only absolute disturbance values ​​but also directional information of the disturbance, thereby distinguishing between structural disturbances and random aerodynamic noise.

[0115] In one example, acquiring the second sensor data of the target wingman located in the mothership mounting chamber to obtain boundary layer disturbance information of the mothership includes:

[0116] Denoising and trend analysis are performed on the wind speed, wind direction and air pressure data in the second sensor to obtain the local airflow change characteristics of the mother machine body;

[0117] Perturbation decomposition and feature extraction are performed on the acceleration and attitude information in the second sensor data to obtain the dynamic response characteristics of the mother machine body;

[0118] By associating and mapping the local airflow change characteristics and the dynamic response characteristics, boundary layer disturbance information is obtained.

[0119] S1.3: The supplementary observation data and the boundary layer disturbance information are used as external constraints. The sensor data corresponding to the mother machine is input into the preset wind field reconstruction model. Spatial interpolation and flow field calculation are performed on the near-field three-dimensional wind field of the mother machine to obtain the near-field three-dimensional wind field model of the mother machine.

[0120] Specifically, the supplementary observation data reflects the diffusion disturbances in the external region surrounding the mothership, the boundary layer disturbance information reflects the local behavior of the body-fitted flow field, and the sensor data onboard the mothership reflects the macroscopic trend of the overall aerodynamic environment of the mothership.

[0121] In this embodiment, wind field reconstruction employs a spatial interpolation method with multi-source constraints. The mothership coordinate system is used as a reference, mothership sensor data as the base field, supplementary observation data as external flow direction constraints, and boundary layer perturbation information as velocity and pressure boundary conditions for the body-fitted region. During wind field interpolation, an interpolation kernel function based on local flow direction consistency is used to weighted aggregate perturbations in the same direction, and perturbation gradient constraints ensure that highly variable regions are not averaged out.

[0122] It is understood that the wind field reconstruction model in this application can be understood as a three-dimensional wind field calculation framework used to uniformly map multi-source aerodynamic information to the coordinate system of the mother machine. Its specific implementation can be flexibly configured according to hardware conditions, computing power or task real-time requirements.

[0123] For example, the wind field reconstruction model can be implemented using a three-dimensional spatial discrete representation based on the mother aircraft's coordinate system. A three-dimensional voxel mesh or unstructured point cloud mesh covering the near-field airspace of the mother aircraft is constructed in the mother aircraft's coordinate system to carry wind field state information. The spatial resolution of the mesh can be set according to the mother aircraft's structural dimensions and the scale of near-field disturbances. For example, a higher resolution is used in the close range near the mother aircraft and its mounting areas, while a lower resolution is used in the external airspace far from the mother aircraft, thus forming an adaptive mesh structure that gradually thickens with distance. As an example, when the mothership is a medium to large fixed-wing UAV or a flight platform with wings and a mounting structure, its body size is usually in the range of several meters to tens of meters. In this case, the grid resolution can be calculated and set according to the body size of the mothership, the size of the mounting area, and the spatial variation rate of near-field airflow disturbance. For example, the resolution can be set to sub-meter to several meters near the mothership and the mounting area, and gradually transition to several meters or more in the external airspace. This application does not limit this, as long as it can reflect the main structural characteristics of the near-field wind field of the mothership.

[0124] Based on this spatial representation, each grid node stores the wind field state variables at its corresponding location. These state variables include at least three components of wind speed in the mother machine's coordinate system, the corresponding wind direction information, and a turbulence index characterizing the degree of local airflow instability. The turbulence index can be obtained by statistically analyzing wind speed fluctuations within a certain time window, for example, based on the magnitude of speed change or speed variance. Those skilled in the art will understand that the specific form of the turbulence index used does not affect the basic idea of ​​this embodiment, as long as it reflects the strength of local airflow disturbances.

[0125] During wind field fusion, the sensor data collected by the main unit and all target aircraft are first converted to the main unit's coordinate system and projected onto the corresponding grid nodes or neighboring grid cells according to their spatial positions. For cases where multiple measurement points exist within the same grid cell, a weighted fusion method can be used. In one optional implementation, the fusion weights are determined by several factors, including the spatial distance between the measurement point and the grid center, the time freshness of the measurement point data, and the confidence level of the measurement point's attitude calculation. Measurement points that are spatially closer, have more recent sampling times, and exhibit stable attitude calculations receive higher weights during the fusion process, ensuring that the reconstructed wind field more closely reflects the current real-world situation.

[0126] For example, in scenarios requiring rapid response, a local interpolation-driven approximate flow field reconstruction method can be adopted. By establishing a basic flow field based on the mother machine's sensor data, and then combining the directional constraints of supplementary observation data to correct the local flow trend, the model can obtain a near-field wind field distribution sufficient to support trajectory extraction at a low computational cost. As another example, in application scenarios where the mother machine has sufficient computing resources or where the task has high requirements for wind field accuracy, a hierarchical solution method based on flow continuity can be adopted, using boundary layer perturbation information as a surface boundary condition input.

[0127] Next, the technical content of the method of this application regarding the pneumatic track will be further elaborated.

[0128] refer to Figure 4 , Figure 4 This is a schematic diagram illustrating the classification of pneumatic tracks provided in the embodiments of this application.

[0129] It is understood that the aerodynamic tracks in this application specifically include holding tracks, approach tracks, and departure tracks. These different tracks play different roles in the near-field wind field structure of the mother aircraft, and their geometric distribution, flow characteristics, and available aerodynamic stability also differ significantly. To ensure that the wingman can maintain attitude stability, smooth dynamic response, and controllable path control under complex near-field disturbances of the mother aircraft, this embodiment conducts a detailed analysis of the wind field structure. Based on factors such as wind direction continuity, disturbance gradient, radial variation trend, and altitude distribution, the extracted continuous streamlines are parameterized and classified to construct aerodynamic tracks with different functions.

[0130] In some specific implementations, the holding track is typically located in a region with low disturbance amplitude and mild flow direction changes within the mother aircraft's coordinate system. This region often exhibits localized circulation or slowly rising airflow patterns. This type of wind field environment allows the wingman to maintain a stable hovering state with relatively low control input, while keeping its relative azimuth to the mother aircraft constant and avoiding entering the high-disturbance wake region. The logic for setting the holding track is derived from observations of the near-field wind field of the mother aircraft:

[0131] At certain locations under the mother aircraft wing, local airflow retraction caused by airfoil, pylon, or cavity structure will naturally form flow paths with a closed tendency. Smoothing these streamlines and using them for wingman waiting can significantly reduce the energy consumption of wingmen when their battery is low or when they are queuing for recovery, and improve safety during in-flight waiting.

[0132] In some specific implementations, the approach trajectory is constructed by utilizing paths with a radial convergence tendency in the wind field, allowing the wingman to smoothly approach the mother aircraft's payload area rather than passively combating turbulence in the wake region. By analyzing the radial components and energy attenuation trends at various points in the wind speed vector field, streamlines that naturally converge toward the payload area in space can be identified. These streamlines, after curvature smoothing and spatial resampling, form the approach trajectory, enabling the wingman to avoid relatively strong disturbances in the wind field during the approach while achieving a stable approach within dynamically acceptable limits. Such trajectories typically maintain relatively smooth airflow direction changes, avoiding multiple attitude corrections by the wingman in the final stage, thus maintaining controllability and safety margins when approaching the mother aircraft structure.

[0133] In other specific implementations, the departure track is typically located in a radially diverging region within the mother aircraft's coordinate system, where wind speed gradually increases. The flow field in this region provides the wingman with a relatively natural acceleration direction, allowing it to quickly escape the wake vortex's influence zone after release. By identifying vortex boundaries, downwash edges, and lateral offset trends in the wind field model, it can be determined that the departure track should avoid low-speed shear zones and the core region of the wake vortex, enabling the wingman to enter safe airspace without significant attitude corrections after release. When constructing the departure track, the divergence trend of the wind field is used as the dominant parameter, making the track spatially directional away from the mother aircraft's center of mass. This provides the wingman with a more stable acceleration environment before full power is established, reducing control load during the initial flight phase.

[0134] In one example, based on the near-field three-dimensional wind field model of the mother aircraft, multiple streamlines adapted to preset constraints are extracted, and the streamlines are parameterized into aerodynamic trajectories for wingman motion, including:

[0135] S2.1: Based on the near-field three-dimensional wind field model of the mother machine, identify the continuous wind field path that meets the preset constraints in terms of updraft intensity, lateral disturbance amplitude, and turbulence threshold;

[0136] Specifically, directly growing a trajectory based on a wind field model can easily lead the wingman into areas of intense disturbance or abrupt changes in flow direction, resulting in attitude fluctuations and even affecting dynamic stability during actual flight. Therefore, before constructing the trajectory, the wind field needs to be screened based on factors such as updraft intensity, lateral disturbance amplitude, and local turbulence level to ensure that the path used for growing streamlines is spatially continuous and the disturbance trend is controllable. This step involves performing a mobility analysis of the wind field, ensuring that subsequent streamline extraction is based on a full understanding of local risks, thereby preventing the wingman from growing a trajectory along unsuitable flight directions.

[0137] In this embodiment, three types of parameters are calculated for each sampling point in the wind speed vector field:

[0138] The first is the upward component of the airflow along the Z-axis of the mother aircraft's coordinate system, which is used to determine whether there is a stable airflow that is conducive to the wingman maintaining altitude;

[0139] Second, the perturbation gradient along the lateral direction. If the gradient changes abruptly, it indicates the presence of a lateral vortex street or shear band, which is not suitable as a flight path.

[0140] Thirdly, the local turbulence intensity is obtained by statistically analyzing the velocity fluctuation amplitude and is used to distinguish between the steady-state region and the highly disturbed region.

[0141] In some specific implementations, these parameters are judged using a combination of thresholds and trends, and the set of spatial points that meet the conditions is marked in the machine coordinate system. Regarding continuity, this embodiment employs a neighborhood filtering method based on velocity direction consistency, ensuring that only when the flow direction consistency of adjacent sampling points exceeds a preset proportion are they considered nodes that can form a continuous path. Through this multi-index joint filtering, a set of continuous, clearly oriented, and perturbation-controllable path starting point regions can be delineated in the three-dimensional wind field.

[0142] S2.2: Perform flow direction integration along the continuous wind field path to extract multiple three-dimensional spatial streamlines for wingman movement;

[0143] Specifically, after determining the continuous wind field path suitable for growth, streamlines need to be grown in space according to the evolution trend of the wind direction vector to construct a three-dimensional spatial path that conforms to the wind field's laws. Because the near-field wind field contains multiple regional small-scale vortices, downwash structures, and directional deflections caused by changes in the mother aircraft's attitude, it is difficult to form a sufficiently stable path based solely on directional information from local points. Therefore, it is necessary to gradually advance along the airflow direction in space using an integral method, extending outward from the fixed reference plane of the mother aircraft's coordinate system, so that the trajectory possesses the characteristics of smooth direction and continuous structure.

[0144] In this embodiment, streamlines are grown segment by segment by spatially projecting along the wind speed vector direction with a preset step size. During each integration, the current position is offset according to the airflow direction, and the integration step size is adjusted based on the local wind speed component. This allows the streamline to have a finer sampling density in rapidly changing regions and a longer propagation length in stable regions, thus balancing computational cost and fidelity. During path growth, if a sudden change in wind direction or turbulence intensity exceeds a threshold, the streamline is prevented from entering impassable areas by inserting transition points, changing the propagation direction, or terminating the streamline. The generated streamline contains a large number of subdivided spatial points, whose positions are gradually pieced together to form a three-dimensional curve with continuous wind direction significance.

[0145] In some optional implementations, this embodiment sets the seed point spacing in association with the grid resolution:

[0146] For example, when the near-body region grid resolution is on the order of 1 meter, seed points can be sampled uniformly within the connected region at 2-meter intervals; when the external spatial domain grid resolution is increased to the order of 2 meters, seed points can be sampled at 3-meter intervals. Simultaneously, to ensure sufficient coverage of the channel structure near the mounting area, a set of directional seed points can be added in a pre-defined fan-shaped space in front of the mounting area. For example, 3 to 5 seed points can be deployed at different height levels with the centerline of the mounting area as the axis, so that the subsequently extracted streamlines can cover the critical spatial zone required for approach. Those skilled in the art will understand that seed point deployment can also be achieved by uniformly scattering points based on the boundaries of connected domains or by prioritizing deployment based on the region with the least disturbance; this embodiment only provides one feasible implementation.

[0147] As an example, within a 0-50 meter radius of the mothership, where the wind field changes rapidly, the integration step size is 0.5 meters; within a 50-150 meter radius, the integration step size is 1 meter; and in the relatively stable area beyond 150 meters, the integration step size is 2 meters. In addition to segmented settings, the step size can also be adaptively adjusted.

[0148] When the wind direction change at two consecutive sampling locations is less than the preset angle and the local turbulence index is at a low level (the specific judgment method can be determined by experiments conducted by those skilled in the art), the step size is increased from 0.5 meters to 1 meter to reduce the number of points; when the wind direction change exceeds 10° or the turbulence index rises, the step size is reduced back to 0.5 meters to improve the curve resolution.

[0149] It is important to emphasize that this embodiment sets multiple termination conditions for the flow integration process to prevent streamlines from entering impassable areas or generating paths that have no engineering significance. Termination conditions may include:

[0150] The process terminates when the integral propulsion enters a grid cell where the local turbulence index exceeds a threshold. This threshold can correspond to the determination that the velocity fluctuation amplitude has continuously exceeded a preset proportion in the most recent sampling. The process terminates when the distance from the streamline point to the boundary of the host structure is less than a safe distance. This safe distance can be set randomly according to the model and the size of the mounted structure to avoid high-risk close-body spaces on the surface of the host body and around the mounted area.

[0151] For example, when a streamline starts 30 meters from the mother machine, advances 50 meters along the near-body region and 100 meters along the middle region before terminating, approximately 100 sampling points can be obtained by sampling the near-body section at a step size of 0.5 meters, and approximately 100 sampling points can be obtained by sampling the middle section at a step size of 1 meter. After resampling at a point spacing of 1 meter during the output renormalization stage, a streamline point sequence consisting of approximately 150 to 170 points can be obtained.

[0152] S2.3: Perform curvature smoothing and spatial resampling on each of the three-dimensional spatial streamlines to obtain an aerodynamic trajectory point sequence that meets the maneuverability requirements of wingman flight;

[0153] Specifically, streamlines obtained directly from integration often exhibit structural issues such as high density, local vibrations, and abrupt curvature changes. Without smoothing, the wingman would need to frequently make attitude corrections while following the trajectory, easily increasing the burden on flight control. In regions with strong local turbulence, the integrated trajectory may even exhibit minute serrated structures. While these structures do not affect aerodynamic resolving power, they are highly detrimental to the wingman's flight control. Therefore, before generating a usable trajectory for the aircraft, the streamlines need to be processed through curvature smoothing and resampling to meet the wingman's maneuverability requirements and ensure smooth tracking within the attitude control range.

[0154] S2.4: The aerodynamic trajectory point series is parameterized according to the spatial position in the coordinate system of the mother machine body, and the parameterized aerodynamic trajectory point series is divided into multiple aerodynamic tracks based on the trajectory phase diagram partitioning;

[0155] Specifically, although the smoothed trajectory points are already controllable, the wind field structures corresponding to different trajectory points are not consistent, requiring further classification to form waiting tracks, approach tracks, and departure tracks. The judgment cannot rely solely on geometric location but should comprehensively consider wind field parameters and dynamic characteristics. Therefore, it is necessary to transform the trajectory points into the mother aircraft's coordinate system, extract features such as radial distance, relative height, and equivalent energy, and construct a trajectory phase diagram to partition the trajectories, allowing trajectories in different regions to be used for different mission phases.

[0156] In one example, the trajectory phase diagram-based partitioning divides the parameterized aerodynamic trajectory point sequence into multiple aerodynamic orbits, including:

[0157] Based on the spatial position of each trajectory point in the aerodynamic trajectory point sequence in the coordinate system of the mother machine, the radial distance sequence and the relative height sequence relative to the mother machine are calculated, and the radial distance sequence and the relative height sequence are used as the spatial coordinates of the trajectory phase diagram;

[0158] The recommended flight speed and relative altitude corresponding to the aerodynamic trajectory point series are determined by a preset trajectory lookup table sequence, the equivalent energy sequence is calculated, and the equivalent energy sequence is used as the state coordinates of the trajectory phase diagram. The trajectory lookup table sequence is determined by matching the flight speed range and altitude range of different spatial positions based on the wingman flight performance parameters and the near-field three-dimensional wind field model of the mother aircraft.

[0159] Based on the changing trends of the radial distance sequence, relative height sequence, and equivalent energy sequence in the trajectory phase diagram, the parameterized aerodynamic trajectory points are partitioned into phase diagrams. Aerodynamic trajectory points that converge in the radial direction and have decaying equivalent energy are classified as approach tracks. Aerodynamic trajectory points that periodically change within a preset range of radial distance and relative height and have constant equivalent energy are classified as waiting tracks. Aerodynamic trajectory points that diverge in the radial direction and have increasing equivalent energy are classified as departure tracks.

[0160] It is understandable that the differences in flyability between different trajectory segments in the near-field wind field of the mother aircraft are not determined solely by geometric position, but by a combination of position, altitude, and energy state. Relying solely on trajectory geometry makes it difficult to distinguish which trajectories are more suitable as approach paths and which are more suitable for holding or departure. Therefore, a multi-dimensional representation method is needed to simultaneously express the spatial orientation and dynamic feasibility of trajectories, ensuring that trajectories have sufficient physical labels before entering the classification stage. Based on this, this embodiment projects the position of each trajectory point onto the mother aircraft's coordinate system and combines recommended velocity and altitude to generate equivalent energy parameters, forming a trend-driven three-dimensional sequence for each trajectory in the trajectory phase diagram. By observing these trend changes, the intention of the trajectory in different directions can be identified—whether its natural wind field properties are trending towards the mother aircraft, maintaining a stable cycle around it, or spreading outwards to form a departure trend.

[0161] In some specific implementations, the mapping relationship between the mother aircraft's coordinate system and the trajectory point cloud must first be established, so that each trajectory point has a radial distance and a relative altitude relative to the mother aircraft's center of mass. These two quantities are the core indicators of the trajectory spatial structure. Subsequently, the recommended flight speed and suitable altitude corresponding to each trajectory point are determined according to a preset trajectory lookup table sequence.

[0162] It is important to note that the trajectory lookup table sequence is not a fixed table, but rather a dynamic recommendation data derived from the near-field three-dimensional wind field model of the mother aircraft, combining the wingman's power envelope, lift-drag characteristics, and acceptable rate of attitude change for each spatial location. By simultaneously calculating the recommended speed and altitude in the trajectory point sequence, an equivalent energy sequence can be constructed using speed and altitude, making the trajectory's dynamic difficulty, energy consumption trend, and wind field orientation specifically presented through this sequence. From an engineering perspective, this method is equivalent to assigning an energy curve to each streamline, allowing those skilled in the art to determine whether the trajectory can be completed within the wingman's allowed energy range.

[0163] In some optional implementations, the calculation method for the trajectory lookup table sequence includes the following aspects:

[0164] In the first aspect, the trajectory lookup sequence can be obtained through structured modeling of the wingman's flight performance parameters. In actual engineering deployments, different wingman models differ in maximum climb rate, optimal lift-to-drag ratio, engine response characteristics, and attitude change constraints. If these constraints are not incorporated into the trajectory parameterization process, the wingman may be assigned to a trajectory segment exceeding its power capabilities, resulting in an inability to maintain stable flight. In this embodiment, when constructing the lookup sequence, a velocity envelope interval is first generated based on the wingman's aerodynamic characteristic curve. Combined with the updraft intensity and disturbance stability in the near-field wind field of the mother aircraft, a feasible speed range suitable for the wingman at different spatial altitudes and radial distances is determined. Subsequently, the trajectory points are velocity-mapped based on the velocity envelope, so that the trajectory points not only have geometric position labels but also flight speed labels that conform to the actual capabilities of the wingman.

[0165] Secondly, the trajectory lookup sequence can be dynamically generated based on the airflow gradient information in the near-field wind field model of the mother aircraft. Local disturbances in the near-field wind field of the mother aircraft often exhibit significant gradient differences with changes in attitude, speed, and external environment. If the trajectory is generated directly according to fixed parameters, the wingman may enter a region of sudden disturbance, increasing the flight control load. Therefore, this embodiment extracts the local upwash and downwash gradients, lateral shear gradients, and turbulence intensity variation trends from the wind field model, and correlates this gradient information with altitude and radial distribution. This allows for a wider recommended speed range in relatively smooth wind field areas, while limiting the speed range in areas with greater wind field disturbances, giving the wingman stronger stability margin when passing through these areas.

[0166] Thirdly, the trajectory lookup sequence can also be empirically fitted using historical flight data of the wingman. For wingmen that have already operated in similar missions, the flight trajectory, attitude change rate, throttle response, yaw stability, and other data recorded by the mother aircraft can reflect the wingman's true adaptability range in wind fields. This embodiment can cluster and fit historical data from different spatial locations to determine the optimal flight speed range or safe altitude window for the wingman at specific altitudes, radial distances, and disturbance intensities. When constructing the lookup sequence, these empirical parameters are used as supplementary constraints to further approximate the wingman's actual operational behavior based on reasonable justification.

[0167] Furthermore, once the radial distance sequence, relative altitude sequence, and equivalent energy sequence constitute a trajectory phase diagram, the trajectory can be partitioned by observing their changing trends as the trajectory points advance. In this embodiment, the approach trajectory typically exhibits a gradually decreasing radial distance, a decreasing relative altitude, and a monotonically decaying equivalent energy; the waiting trajectory shows a periodic closing trend in the phase diagram, with its radial and altitude changes fluctuating within a certain range, while the energy change remains stable; the departure trajectory shows a continuously increasing radial distance, a gradually rising altitude, or a sustained upward trend, accompanied by an increase in equivalent energy. Based on these trends, this embodiment can perform continuity analysis and trend verification on the trajectory point sequence to ensure that the partitioned trajectory physically conforms to the wind flow direction and can be tracked by the wingman's flight control module with a low control load.

[0168] It should be further noted that this embodiment uses a sliding window method for trend statistics:

[0169] M consecutive trajectory points are selected as the analysis window along the trajectory advancement direction. M can be set according to the trajectory sampling density. For example, when the point spacing is 1 meter and it is desired to cover a path length of about 30 to 50 meters, M can be set to 30 to 50. For radial trend discrimination, the radial distance change between the start and end points of the window is compared within each window. When the radial distance decreases overall within the window and the decrease percentage reaches a preset threshold, the window is determined to have a convergent trend. The threshold can be set to about 10% to balance sensitivity and stability, and allows adjustment within the range of 5% to 30% under different turbine models and wind field conditions. Correspondingly, when the radial distance increases overall within the window and the increase percentage reaches a preset threshold, the window is determined to have a divergent trend. For periodic discrimination, this embodiment performs peak and valley detection on the radial distance sequence and the relative height sequence:

[0170] If the radial distance shows at least two distinct peak-valley alternations within a candidate interval, and the relative height synchronously exhibits reciprocating changes, then the interval is determined to have periodic characteristics and can be considered as a candidate segment for the waiting track.

[0171] To avoid misjudging brief jitters as periods, this embodiment may further require that the peak-valley spacing reach the minimum spatial span (e.g., not less than 10 to 30 meters) and that the peak-valley amplitude exceed the minimum change (e.g., radial change not less than 5 meters or exceeding a preset proportion of the current radial value), so as to ensure that the periodic characteristics correspond to the real trajectory structure that can be used for waiting and hovering.

[0172] For the determination of equivalent energy trends, this embodiment statistically analyzes the monotonicity percentage of equivalent energy changes within the same window:

[0173] When the equivalent energy change direction of most trajectory points within a window is consistent, and the proportion of points changing in the same direction exceeds a preset ratio, it is considered that the window has a clear energy decay or energy increase trend. By combining radial trend, periodic characteristics, and energy trend, trajectory points can be divided into approach candidate segments, waiting candidate segments, and departure candidate segments in the phase diagram.

[0174] Next, the technical content of the method of this application regarding time slot sequences will be further elaborated.

[0175] It is understandable that the aerodynamic trajectory obtained in the aforementioned steps is essentially a set of three-dimensional paths composed of several smoothed spatial points. These spatial points, referenced to the mother aircraft's coordinate system, already possess a clear geometric order and spatial distribution. However, the trajectory itself is still only a path in a spatial sense, without containing information about when the wingman passes through a certain point on the trajectory. Without a time dimension marker, even if the trajectory itself is physically flyable, it is impossible to achieve orderly coordination among multiple wingmen, let alone dynamic scheduling of approach, departure, and holding queues. Therefore, this application introduces time parameterization, transforming static spatial points into a sequence with time attributes, enabling the trajectory to evolve from merely describing where it flies to describing when it flies to where it goes, thereby supporting the logical closed loop of subsequent trajectory allocation and collision avoidance processes.

[0176] In this embodiment, the time parameter is not simply distributed evenly according to the trajectory length, but is dynamically calculated based on the recommended flight speed, turning radius, attitude change limits, and local airflow acceleration trend in the near-field wind field of the mother aircraft for each spatial point in the aforementioned trajectory point series. In other words, for each trajectory point, its reasonable target arrival time needs to be determined based on different wind field structures such as its location in the updraft region, recirculation region, or wake boundary region, combined with the wingman's own dynamic envelope.

[0177] It should be noted that, unlike spatial planning of trajectories alone, the generation of time slots in this embodiment is not simply adding an arrival time to each spatial point, but rather performing time aggregation, discretization, and safety window expansion on the spatial point sequence to enable it to directly support parallel scheduling and conflict detection of multiple wingmen.

[0178] After obtaining the target arrival time corresponding to each spatial point, this embodiment does not directly use all spatial points as independent time nodes, but introduces time aggregation and discretization. Specifically, the original arrival time sequence arranged along the trajectory is traversed, and when the difference in arrival time between two adjacent spatial points is less than a preset time merging threshold, the two spatial points are considered to belong to the same continuous operating interval in an engineering sense.

[0179] For example, in the approach trajectory, when a wingman flies along the trajectory at a speed of approximately 15 meters per second and the distance between trajectory points is 1 meter, the arrival time difference between adjacent points is approximately 0.07 seconds. If each point is treated as a separate scheduling node, the time dimension would be too dense and unfavorable for conflict management. Therefore, in this embodiment, a time merging threshold of 0.5 seconds can be set to aggregate multiple consecutive spatial points into a merged spatial segment, and the average or weighted average arrival time of each spatial point within this segment can be used as the time parameter of the merged spatial segment.

[0180] In one example, the step of time parameterizing each aerodynamic orbit to generate a time slot sequence consisting of spatial location and time labels includes:

[0181] S3.1: For each aerodynamic track, based on the spatial point array of the aerodynamic track in the coordinate system of the mother machine, sort the spatial points in the spatial point array according to the trajectory advancement direction to obtain an ordered trajectory sequence. The trajectory advancement direction corresponding to the waiting track is the local closed loop direction generated along the spatial point array, the trajectory advancement direction corresponding to the approach track is the convergence direction relative to the mother machine's center of mass toward the mother machine's loading area, and the trajectory advancement direction corresponding to the departure track is the divergence direction relative to the mother machine's center of mass toward the mother machine's external airspace.

[0182] Specifically, an aerodynamic trajectory consists of a series of spatial points, but the natural order in which these points are generated does not always align with the wingman's flight propulsion direction in the physical world. For example, a waiting streamline near the underside of the mother aircraft's wing may exhibit localized small-scale rollback, forming a non-linearly distributed series of points in the wind field reconstruction. If the propulsion direction is not clearly defined, the wingman may experience path reversal or sudden direction reversal while flying along the trajectory.

[0183] In this embodiment, the propulsion direction is determined by the geometric trend of the trajectory itself and the local flow direction of the wind field. For the waiting trajectory, by analyzing the local tangential vector of the trajectory point and the wind speed direction near that point, the direction exhibiting a closing trend within the loop region is selected as the propulsion direction, allowing the wingman to perform low-energy waiting along the natural retraction direction. For the approach trajectory, by fitting the radial distance of the trajectory point relative to the centroid of the mother aircraft, the direction in which the radial distance continuously shortens is selected as the propulsion direction, ensuring that the wingman always faces the loading area during flight. For the departure trajectory, by detecting the radial distance growth trend and the local airflow divergence trend, the wingman naturally leaves the mother aircraft's influence zone along the airflow. After the direction is determined, the spatial point sequence is reordered according to the propulsion direction, and a direction continuity check is performed on the reordered point sequence to ensure that the reordered trajectory sequence is smooth in direction and does not experience reverse jumps.

[0184] S3.2: Based on the spatial distance between spatial points in the trajectory sequence and the recommended flight speed of the corresponding spatial point, calculate the target arrival time of each spatial point and use the target arrival time as the time parameter of the spatial point;

[0185] Specifically, different distances exist between spatial points, and the recommended flight speed of a wingman in different areas is significantly affected by the wind field structure. Therefore, the propulsion time between different trajectory points cannot be simply calculated as a fixed duration. If the influence of the wind field on speed is not considered, the wingman may be allocated too much time in the high-speed wake region, resulting in unstable behaviors such as hovering or sudden deceleration; conversely, allocating too little time in the updraft region may lead to insufficient power response of the wingman.

[0186] In this embodiment, the recommended flight speed for each spatial point is derived from a trajectory lookup table sequence. By combining the recommended speed with the Euclidean distance between spatial points, the propulsion time of the wingman under ideal aerodynamic conditions can be obtained. In actual calculations, the dynamic transitions of the wingman during acceleration and deceleration phases must also be considered, making the time allocation dependent not only on the positional distance but also on parameters such as the wingman's attitude change rate and tilt angle change rate. For example, in regions with abrupt changes in wind direction, the time interval is extended to allow for attitude adjustment space for the wingman; in regions with stable winds, the time is compressed to improve overall scheduling efficiency. The final target arrival time constitutes a time series that gradually accumulates with the trajectory position, transforming the trajectory from a purely spatial structure into a path with a time progression.

[0187] In one example, the arrival time of the target is used as the time parameter of the spatial point, including:

[0188] Based on the target arrival time, an original arrival time sequence arranged along the trajectory sequence is obtained;

[0189] Based on the time difference between adjacent spatial points in the original arrival time sequence, spatial points with a time difference less than a preset time merging threshold are time-aggregated to obtain merged spatial points and their corresponding updated arrival times. The updated arrival times are used as the time parameters of the merged spatial points, wherein the updated arrival times are determined by calculating the average value of the target arrival times.

[0190] For spatial points that are not time-aggregated, their corresponding target arrival time is retained as a time parameter.

[0191] refer to Figure 5 , Figure 5 This is a schematic diagram illustrating the principle of time aggregation provided in the embodiments of this application.

[0192] like Figure 5As shown, the aerodynamic trajectory includes space point one, space point two, and space point three, where space point one corresponds to target arrival time one, space point two corresponds to target arrival time two, and space point three corresponds to target arrival time three.

[0193] Figure 5 The diagram further illustrates that, based on the time series, the process first determines whether the time difference between target arrival time one and target arrival time two is less than a preset time merging threshold. If the result is yes, spatial point one and spatial point two are grouped together for time aggregation. By averaging the target arrival times corresponding to the two, an updated arrival time is obtained. In the diagram, the merged spatial point represents the new spatial point after aggregation. It assumes the functions of the original spatial point one and spatial point two in the trajectory, thereby reducing the number of parameter points with excessively dense time intervals on the trajectory.

[0194] Meanwhile, for spatial point three, since the time difference between its target arrival time three and the merged spatial point is greater than the time merging threshold, it maintains its independence, does not participate in the aggregation operation, and continues to use the original target arrival time three as the time parameter. Under this structure, the trajectory retains the necessary time resolution, and the time merging mechanism effectively reduces the scheduling burden caused by excessively dense spatial points, making the time parameter distribution more uniform and meaningful for flight control.

[0195] It is understandable that if the time difference between the arrival time of the target and the time of merging the spatial points is less than the time merging threshold, then the two spatial points will continue to be merged.

[0196] S3.3: Discretize the spatial points with time parameters in time, divide the continuous time parameters into multiple time intervals, and obtain a time slot consisting of time intervals and corresponding spatial positions, wherein the time slot is used to represent the spatial operation window of the target wingman on the aerodynamic track.

[0197] Specifically, while the target arrival time defines the ideal time for the wingman on the trajectory, forcing the wingman to strictly match this time in actual scheduling would lead to overly rigid scheduling, making it difficult to adapt to changes in battery status, transient wind fields, or differences in the wingman's power performance. Therefore, it is necessary to discretize the continuous time parameters, dividing them into time intervals, so that the wingman can enter the designated spatial segment within a reasonable time window during flight. The essence of time slots is to intervalize the time axis, making the wingman's flight plan scalable rather than precise at a single point.

[0198] S3.4: Construct a time slot sequence consisting of multiple time slots on each of the pneumatic tracks based on the time order of the time slots.

[0199] Next, the technical details of the method in this application regarding the target orbit and the target time slot will be further elaborated.

[0200] Understandably, the target trajectory and target time slot here correspond to the specific scheduling scheme followed by wingmen in the near-field environment of the mother aircraft in actual engineering. That is, in the case of multiple wingmen operating simultaneously and the near-field wind field of the mother aircraft constantly changing, the spatial path and available time window allocated to each wingman. Because the near-field wind field structure exhibits obvious directional, vortex, and dynamic disturbance characteristics, different aerodynamic trajectories differ in terms of flightability, energy consumption, and safety margin. Therefore, if the target trajectory corresponding to each wingman is not determined in advance, it is difficult to ensure that there will be no spatial conflict or attitude instability caused by the superposition of wind field interference between wingmen. The time slot further specifies the order and rhythm of the wingmen's operation on the trajectory, making the trajectory have scheduling significance rather than just a flightable path, thereby achieving orderly operation under the airborne mothership.

[0201] Scheduling is specifically divided into two types:

[0202] If the target wingman corresponds to an approach or recovery operation, the method includes:

[0203] Based on the current position of the target wingman and the spatial distribution of the aerodynamic trajectories, candidate aerodynamic trajectories that the target wingman can reach are determined;

[0204] Based on the remaining battery power of the target wingman and the energy consumption estimate corresponding to the candidate aerodynamic trajectories, the candidate aerodynamic trajectories are screened to obtain a set of optional aerodynamic trajectories;

[0205] Based on the occupancy status of each time slot in the time slot sequence, the minimum safe distance requirement with other wingmen, and the mission priority of the target wingmen, a target track that meets the safe distance constraint is selected from the set of optional aerodynamic tracks.

[0206] In the time slot sequence corresponding to the target trajectory, select the target time slot that satisfies the target wingman's current position, flight speed range, and dynamic constraints, and use the target time slot as the target wingman's operating window.

[0207] If the target wingman corresponds to a takeoff operation, the method includes:

[0208] Based on the spatial position of the mounting area in the coordinate system of the mother aircraft, a releaseable takeoff area matching the mounting area is determined, and candidate departure tracks that are spatially continuous with the releaseable takeoff area are selected from the aerodynamic tracks.

[0209] Based on the occupancy status of each time slot in the time slot sequence corresponding to the candidate departure track set, and the minimum safe interval between the time slot and other wingman operation windows, a target departure track that meets the safe distance and release conditions is selected.

[0210] In the time slot sequence of the target departure track, the target time slot is determined from the time slots that satisfy the target wingman's release attitude constraints and acceleration capability constraints, and the target time slot is used as the takeoff window of the target wingman.

[0211] It is understandable that, since the trajectory, time slot, and wingman status are all clearly defined in the preceding steps, the scheduling process no longer requires additional aerodynamic modeling work. Instead, it relies on condition filtering and constraint judgment based on the generated data structure. Those skilled in the art will understand that this type of scheduling problem can be solved in engineering using various mature methods such as graph search, time window scheduling, constraint satisfaction solving, and real-time priority ranking. It can be accomplished simply by ensuring trajectory reachability, safe distance, and non-conflicting time windows. The following sections describe the two scheduling methods respectively:

[0212] When a target wingman needs to perform an approach or recovery operation, the scheduling process mainly revolves around how to guide the wingman to the target near the mother aircraft's mounting area. In the first step of this application's embodiments, namely determining the reachable trajectory based on the target wingman's current position and the spatial distribution of its aerodynamic trajectory, this can be achieved by geometric distance, directional consistency, or accessibility judgment based on the wind field. Algorithms such as spatial accessibility analysis and path topology filtering can all meet the requirements.

[0213] Further screening of candidate trajectories based on the remaining battery power of the wingman can employ existing energy consumption estimation models or empirical formulas. For example, the minimum battery power required to execute the trajectory can be estimated by considering trajectory length, recommended speed, and wind field boost / drag factors, thereby eliminating trajectories that the wingman cannot safely complete. Subsequently, the most suitable target trajectory is selected based on the occupancy status of time slots, safety intervals, and task priorities. This can be achieved through methods such as time window conflict detection, priority ranking, and minimum interval constraint checks.

[0214] Finally, select the target time slot that meets the wingman's current position, speed range, and dynamic capabilities from the target orbit time slot sequence to obtain the wingman's actual operating window.

[0215] For wingmen within the loading area and preparing for takeoff, the scheduling logic focuses on smoothly releasing them to a suitable area in the near-field wind field of the mother aircraft. In this embodiment, the determination of the releaseable takeoff area is based on the positional relationship of the loading area within the mother aircraft's coordinate system. This determination can be obtained directly through geometric judgment or coordinate transformation methods. Subsequently, candidate departure tracks that are spatially continuous with the release area are selected from the aerodynamic tracks. This can also be achieved through spatial distance between track points, consistency of flow direction, or connectivity based on wind field direction.

[0216] The key to selecting a departure track is to avoid time and space conflicts between wingmen after their release. Therefore, it is necessary to consider the occupancy of time slots and safety interval constraints. Common technologies in this field, such as time window conflict detection, interval constraint adjustment, and stacked queuing models, can all support this implementation.

[0217] Next, a specific and complete example will be used to illustrate the entire process. The example is only to illustrate the feasibility at the computational level and does not represent the actual values. The specific values ​​can be determined by those skilled in the art through simulation experiments or physical experiments.

[0218] For a given approach trajectory, the orbital point sequence is resampled with a point spacing of 1 meter. A trajectory sequence of approximately 180 meters is formed by selecting 180 consecutive spatial points along the advancement direction. A trajectory lookup table provides a recommended speed range and altitude window for each point in this sequence. For example, in the outer section, which is farther from the mothership and has lower turbulence, the recommended speed range is 14–18 m / s; in the inner section, which is closer to the loading area and experiences increased lateral disturbances, the recommended speed range converges to 10–14 m / s. Based on these recommended speed ranges, to ensure scheduling consistency, 16 m / s for the outer section and 12 m / s for the inner section can be used as reference speeds for time series estimation. Therefore, the time increment for each meter of advancement in the outer section is approximately 0.06 seconds, and the time increment for each meter of advancement in the inner section is approximately 0.08 seconds. The original arrival time sequence can be obtained by accumulating the time increments along the trajectory. For example, the 0th point corresponds to 0 seconds, the 80th point corresponds to about 4.8 to 5.2 seconds, and the 160th point corresponds to about 10.5 to 12 seconds, showing a time sequence that monotonically increases as the trajectory progresses.

[0219] Understandably, since the point-level time increment is on the order of 0.06 to 0.08 seconds, directly using each point as a scheduling node would result in excessively dense time nodes. Therefore, a time merging threshold is introduced for aggregation.

[0220] For example, setting the time merging threshold to 0.5 seconds, approximately 8-9 points in the outer segment are aggregated into a merged spatial segment, and approximately 6 points in the inner segment are aggregated into a merged spatial segment. For a certain outer segment aggregation group containing 9 consecutive points with corresponding arrival times covering approximately 0.54 seconds, the updated arrival time of this group is taken as the average of the arrival times within the group and used as the time parameter for the merged spatial segment. Through the aggregation process, the original 180 points can be compressed into approximately 20-30 merged spatial segments, making the time granularity consistent with the engineering scheduling needs, while preserving the time structure caused by the trajectory advancement sequence and speed differences.

[0221] In this embodiment, the time slots are generated based on the time parameters of the merged spatial segments. For example, for the approach track, to balance alignment accuracy and time flexibility near the mounting area, the time margin is set to 1.5 seconds before and after, so each slot forms a time window of approximately 3 seconds. For the waiting track, the time margin is widened to 4 seconds before and after to absorb rhythm fluctuations caused by cyclic waiting. For the departure track, the time margin is set to 3 seconds before and after to cover the transition process of acceleration and attitude establishment after release.

[0222] Furthermore, to ensure the longitudinal safety separation of multiple wingmen, this embodiment introduces a minimum safety separation verification and conversion at the slot level:

[0223] For example, the minimum safe distance during the approach phase is set to 60 meters, and the minimum time interval requirement of 4 to 5 seconds is calculated based on the approach phase reference speed of 12 to 16 meters per second. Correspondingly, when two approach slots overlap on the time axis or the interval is less than 5 seconds, it is determined that the safe distance is not met.

[0224] For example, if a wingman has already occupied an approach slot [10.0 seconds, 13.0 seconds] (indicating that it will enter a critical space segment near the mounting area within this time window), and another wingman's candidate slot is [12.5 seconds, 15.5 seconds], it is directly judged as a conflict because the windows overlap; if the candidate slot is [14.0 seconds, 17.0 seconds], although it does not overlap, the interval between it and the end time of the previous slot is only 1 second, which still does not meet the 5-second safety interval requirement; at this time, the next available slot can be selected, for example [18.0 seconds, 21.0 seconds], and it can be used as the target time slot. Thus, the time slot not only provides an access window, but also serves as a direct carrier for conflict detection and safety verification, making the scheduling results have verifiable timing safety.

[0225] In a specific application scenario, for a recovery wingman, its current position is approximately 250 meters from the mother aircraft, and its remaining battery power corresponds to approximately 6 minutes of flight time. Candidate trajectories include an approach trajectory and a holding trajectory. Trajectory energy consumption can be estimated using an engineering-feasible baseline cruise power consumption and attitude correction-related additional power consumption. For example, using the energy consumption per unit time at a cruise speed of 16 meters per second as a baseline, and considering the lower recommended speed and higher disturbance level within the approach trajectory, the energy consumption per unit distance of the approach trajectory can be set to 1.2 times that of the holding trajectory. With trajectory lengths of approximately 180 meters and 300 meters respectively, it can be concluded that the approach trajectory is more suitable for rapid recovery, while the holding trajectory is more suitable for queuing and buffering.

[0226] For example, if the most recently available window for the approach track in the available slot sequence is [18 seconds, 21 seconds], and the most recently available window for the waiting track is [6 seconds, 14 seconds], then when the wingman's recovery priority is high, the approach track can be directly selected and locked [18 seconds, 21 seconds]. When the recovery priority is low or there are many approach slot conflicts, the waiting track window [6 seconds, 14 seconds] can be allocated first, and the application for approach slots can be rolled over before the end slot of the waiting track expires, realizing a two-stage closed-loop scheduling of waiting and approach. This closed-loop scheduling can be implemented in engineering by real-time updating of the slot occupancy table.

[0227] Once a wingman confirms that it has occupied a slot, its slot status is recorded as occupied, and a 5-second safety interval is superimposed before and after the corresponding time window as a protective zone for other wingmen to check and avoid.

[0228] For takeoff scenarios, the release maneuver in the payload area requires the wingman to maintain a specific attitude and relative airflow conditions before release, and to establish a stable speed within a finite distance after release. This constraint can be transformed into a time window condition:

[0229] For example, if 2 seconds are needed for attitude alignment and release preparation before release, and 4 seconds are needed to accelerate from 0 to 15 meters per second after release, then a takeoff window should continuously cover a conflict-free period of at least 6 seconds. If a departure track has a candidate window [30 seconds, 36 seconds] in the slot sequence, and this window does not overlap with the critical slot protection zone of other wingmen near the loading area, then this window can be identified as the target takeoff window; if the candidate window is [33 seconds, 38 seconds], although the length meets the 6-second requirement, its front section overlaps with the approach protection zone of another wingman [31 seconds, 35 seconds], then it is judged as not meeting the release safety conditions, and the next window must be selected.

[0230] In some optional embodiments, the method further includes:

[0231] After receiving the track identifier and time slot sequence, the target wingman, with the mother aircraft as a reference, generates control instructions to reach the corresponding spatial position within each target time slot time window based on its own relative position and the three-dimensional wind field model. The control instructions include at least a trajectory tracking control quantity based on time slot timing constraints and a flow field following control quantity based on the flow direction of the pneumatic conveyor belt, so as to complete the waiting, approach or departure process along the pneumatic conveyor belt track.

[0232] During the movement of the target wingman along the pneumatic conveyor belt track, the mother machine and each electronic wingman periodically update the three-dimensional wind field model. When it is predicted, based on the updated wind field and the electronic wingman's status, that the target wingman cannot meet the spatiotemporal constraints of the allocated time slot within a predetermined error range, the electronic wingman sends a time slot adjustment request. The mother machine then reallocates the target pneumatic conveyor belt track and time slot according to the current track and time slot occupancy status, or instructs the electronic wingman to return upstream and wait for the track to re-queue.

[0233] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A collaborative integrated control method for unmanned aerial vehicles (UAVs) based on a distributed architecture, characterized in that, Applied to an airborne homeport system comprising a mother aircraft and multiple wingmen operating around the mother aircraft, the method includes: In the body coordinate system corresponding to the mother aircraft, a three-dimensional wind field model of the mother aircraft near the field is constructed based on the sensor data of the mother aircraft and each target wingman, wherein the target wingman is the wingman that needs to perform take-off, approach or recovery operations; Based on the near-field three-dimensional wind field model of the mother aircraft, multiple streamlines adapted to preset constraints are extracted, and these streamlines are parameterized into aerodynamic trajectories for wingman movement, including at least one holding trajectory, one approach trajectory, and one departure trajectory, including: Based on the near-field three-dimensional wind field model of the mother machine, continuous wind field paths that meet preset constraints in terms of updraft intensity, lateral disturbance amplitude, and turbulence threshold are identified. By performing flow direction integration along the continuous wind field path, multiple three-dimensional spatial streamlines for wingman movement are extracted. Curvature smoothing and spatial resampling are performed on each of the three-dimensional spatial streamlines to obtain an aerodynamic trajectory point sequence that meets the maneuverability requirements of wingman flight; The aerodynamic trajectory point series is parameterized according to the spatial position in the mother machine body coordinate system, and the parameterized aerodynamic trajectory point series is divided into multiple aerodynamic tracks based on the trajectory phase diagram partitioning. Each aerodynamic orbit is time-parameterized to generate a time slot sequence consisting of spatial position and time label, including: For each aerodynamic track, based on the spatial point array of the aerodynamic track in the coordinate system of the mother machine, the spatial points in the spatial point array are sorted according to the trajectory advancement direction to obtain an ordered trajectory sequence. The trajectory advancement direction corresponding to the waiting track is the local closed loop direction generated along the spatial point array, the trajectory advancement direction corresponding to the approach track is the convergence direction relative to the mother machine's center of mass toward the mother machine's loading area, and the trajectory advancement direction corresponding to the departure track is the divergence direction relative to the mother machine's center of mass toward the mother machine's external airspace. Based on the spatial distance between spatial points in the trajectory sequence and the recommended flight speed of the corresponding spatial point, the target arrival time of each spatial point is calculated, and the target arrival time is used as the time parameter of the spatial point. Time discretization is performed on spatial points with time parameters, and continuous time parameters are divided into multiple time intervals to obtain time slots composed of time intervals and corresponding spatial positions. The time slots are used to represent the spatial operation window of the target wingman on the aerodynamic track. Based on the time sequence of the time slots on each of the pneumatic tracks, a time slot sequence consisting of multiple time slots is constructed. Based on the target wingman's current location, battery level, and mission priority, and in conjunction with the time slot sequence, a corresponding target track and target time slot are assigned to the target wingman to generate control commands for the target wingman.

2. The UAV cooperative integrated control method based on a distributed architecture according to claim 1, characterized in that, The sensor data includes wind speed, wind direction, air pressure, acceleration, and attitude information. The construction of a near-field three-dimensional wind field model of the mother aircraft based on the sensor data from the mother aircraft and each target wingman includes: Acquire first sensor data of the target wingman in the airspace outside the mother aircraft, perform spatiotemporal alignment and fusion of the first sensor data, and generate supplementary observation data; The second sensor data of the target wingman located in the mother machine mounting chamber is obtained to obtain the boundary layer disturbance information of the mother machine body; The supplementary observation data and the boundary layer disturbance information are used as external constraints. The sensor data corresponding to the mother machine are input into the preset wind field reconstruction model. Spatial interpolation and flow field calculation are performed on the near-field three-dimensional wind field of the mother machine to obtain the near-field three-dimensional wind field model of the mother machine.

3. The UAV collaborative integrated control method based on a distributed architecture according to claim 2, characterized in that, The step of acquiring second sensor data from the target wingman located in the mothership mounting chamber to obtain boundary layer disturbance information of the mothership includes: Denoising and trend analysis are performed on the wind speed, wind direction and air pressure data in the second sensor to obtain the local airflow change characteristics of the mother machine body; Perturbation decomposition and feature extraction are performed on the acceleration and attitude information in the second sensor data to obtain the dynamic response characteristics of the mother machine body; By associating and mapping the local airflow change characteristics and the dynamic response characteristics, boundary layer disturbance information is obtained.

4. The UAV collaborative integrated control method based on a distributed architecture according to claim 1, characterized in that, The trajectory phase diagram-based partitioning divides the parameterized aerodynamic trajectory point sequence into multiple aerodynamic orbits, including: Based on the spatial position of each trajectory point in the aerodynamic trajectory point sequence in the coordinate system of the mother machine, the radial distance sequence and the relative height sequence relative to the mother machine are calculated, and the radial distance sequence and the relative height sequence are used as the spatial coordinates of the trajectory phase diagram; The recommended flight speed and relative altitude corresponding to the aerodynamic trajectory point series are determined by a preset trajectory lookup table sequence, the equivalent energy sequence is calculated, and the equivalent energy sequence is used as the state coordinates of the trajectory phase diagram. The trajectory lookup table sequence is determined by matching the flight speed range and altitude range of different spatial positions based on the wingman flight performance parameters and the near-field three-dimensional wind field model of the mother aircraft. Based on the changing trends of the radial distance sequence, relative height sequence, and equivalent energy sequence in the trajectory phase diagram, the parameterized aerodynamic trajectory points are partitioned into phase diagrams. Aerodynamic trajectory points that converge in the radial direction and have decaying equivalent energy are classified as approach tracks. Aerodynamic trajectory points that periodically change within a preset range of radial distance and relative height and have constant equivalent energy are classified as waiting tracks. Aerodynamic trajectory points that diverge in the radial direction and have increasing equivalent energy are classified as departure tracks.

5. The UAV collaborative integrated control method based on a distributed architecture according to claim 1, characterized in that, Using the arrival time of the target as a time parameter for a spatial point includes: Based on the target arrival time, an original arrival time sequence arranged along the trajectory sequence is obtained; Based on the time difference between adjacent spatial points in the original arrival time sequence, spatial points with a time difference less than a preset time merging threshold are time-aggregated to obtain merged spatial points and their corresponding updated arrival times. The updated arrival times are used as the time parameters of the merged spatial points, wherein the updated arrival times are determined by calculating the average value of the target arrival times. For spatial points that are not time-aggregated, their corresponding target arrival time is retained as a time parameter.

6. The UAV cooperative integrated control method based on a distributed architecture according to claim 1, characterized in that, If the target wingman corresponds to an approach or recovery operation, the method includes: Based on the current position of the target wingman and the spatial distribution of the aerodynamic trajectories, candidate aerodynamic trajectories that the target wingman can reach are determined; Based on the remaining battery power of the target wingman and the energy consumption estimate corresponding to the candidate aerodynamic trajectories, the candidate aerodynamic trajectories are screened to obtain a set of optional aerodynamic trajectories; Based on the occupancy status of each time slot in the time slot sequence, the minimum safe distance requirement with other wingmen, and the mission priority of the target wingmen, a target track that meets the safe distance constraint is selected from the set of optional aerodynamic tracks. In the time slot sequence corresponding to the target trajectory, select the target time slot that satisfies the target wingman's current position, flight speed range, and dynamic constraints, and use the target time slot as the target wingman's operating window.

7. The UAV cooperative integrated control method based on a distributed architecture according to claim 1, characterized in that, If the target wingman corresponds to a takeoff operation, the method includes: Based on the spatial position of the mounting area in the coordinate system of the mother aircraft, a releaseable takeoff area matching the mounting area is determined, and candidate departure tracks that are spatially continuous with the releaseable takeoff area are selected from the aerodynamic tracks. Based on the occupancy status of each time slot in the time slot sequence corresponding to the candidate departure track set, and the minimum safe interval between the time slot and other wingman operation windows, a target departure track that meets the safe distance and release conditions is selected. In the time slot sequence of the target departure track, the target time slot is determined from the time slots that satisfy the target wingman's release attitude constraints and acceleration capability constraints, and the target time slot is used as the takeoff window of the target wingman.

8. A distributed architecture-based UAV collaborative integration control system, used to implement the distributed architecture-based UAV collaborative integration control method as described in any one of claims 1-7, characterized in that, The system includes: The wind field perception and reconstruction module is used to acquire sensor data from the mother aircraft and each target wingman, process the sensor data of the target wingman located in the airspace outside the mother aircraft and in the mother aircraft mounting chamber, and construct a three-dimensional wind field model of the mother aircraft in the near field. The aerodynamic trajectory generation module is used to extract continuous wind field paths that meet preset constraints based on the near-field three-dimensional wind field model of the mother machine, perform curvature smoothing and spatial resampling on the extracted three-dimensional spatial streamlines to obtain aerodynamic trajectory point series, and divide the aerodynamic trajectory point series into waiting track, approach track and departure track based on trajectory phase diagram partitioning. The time slot management module is used to parameterize the aerodynamic trajectory points for each aerodynamic orbit, calculate the target arrival time of spatial points, and perform time aggregation and time discretization to generate a time slot sequence composed of spatial location and time label. The collaborative scheduling and control module is used to determine the target trajectory and target time slot of the target wingman based on the target wingman's current location, remaining battery power, and task priority, combined with the time slot sequence, and to generate control instructions to be issued to the target wingman based on the target trajectory and target time slot.

Images