An aerial advertising system and method for a drone carrying a flexible LED display

Abstract

The application provides an aerial advertising system and method with a flexible LED display screen carried by a UAV, and relates to the technical field of UAV display, which comprises the following steps: generating formation control instructions and task allocation information based on task requirements and preset advertising picture parameters; controlling the UAV cluster to ascend asynchronously in stages according to the formation control instructions and the task allocation information, completing the division of sub-clusters and hierarchical formation calibration in a target airspace, and forming an initial cluster formation; generating attitude adjustment instructions and display splicing calibration instructions based on the real-time spatial pose of each UAV in the initial cluster formation, and constructing a complete advertising display surface; performing perspective adaptation processing and partition pixel mapping on original advertising data according to the spatial topological characteristics of the complete advertising display surface, and generating adapted target video frame data; generating display trigger instructions based on a high-precision synchronous clock reference and a UAV display delay prediction result, and controlling each flexible LED display screen to synchronously play the target video frame data.

Description

Technical Field

[0001] This invention relates to the field of drone display technology, specifically to an aerial advertising system and method for drones carrying flexible LED displays. Background Technology

[0002] In recent years, with the convergence of drone technology and display technology, using drone swarms for dynamic aerial advertising has become an emerging trend in the outdoor advertising and visual media fields. By raising the display carrier into the air to a vast space, it breaks free from the physical limitations of traditional ground displays in terms of location, size, and viewing angle, creating advertising effects with strong visual impact and spatial coverage, showing broad application prospects in urban landmarks, large-scale events, and the nighttime economy.

[0003] Currently, there are two main technical approaches: The first type is drone-based illuminated dot matrix formations. This type of solution creates patterns by changing the positions of drones. Essentially, it is a low-resolution dot display, capable of showing only simple outlines and text. It is completely unable to present complex dynamic video content that requires continuous pixel surfaces, resulting in weak advertising performance.

[0004] The second type involves using a drone to mount a single rigid display screen. While this solution can play videos, the screen area is small and the visual effect is limited due to the drone's payload. Furthermore, rigid screens have poor wind resistance and low stability during flight, posing safety hazards, and the display scale cannot be expanded by splicing multiple screens.

[0005] Therefore, an aerial advertising system and method for drones carrying flexible LED displays is provided, which can effectively solve the core problem in the prior art that high-definition dynamic video cannot be synchronously presented on a large-scale, highly stable aerial display surface, and greatly improve the display resolution, image continuity, cluster expansion capability, and dynamic stable display performance of aerial advertising in complex airspace environments. Summary of the Invention

[0006] To address the aforementioned technical problems, the present invention aims to provide an aerial advertising system and method for drones carrying flexible LED displays. This system effectively solves the core problem in the prior art that high-definition dynamic videos cannot be synchronously presented on a large-scale, highly stable aerial display surface, greatly improving the display resolution, image continuity, cluster expansion capability, and dynamic stability performance of aerial advertising in complex airspace environments.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for aerial advertising using a drone carrying a flexible LED display screen, the method comprising: Based on the requirements of aerial advertising missions and preset advertising screen parameters, and combined with UAV swarm control logic and airspace environment prediction data, formation control instructions and mission allocation information are generated. Based on the formation control instructions and task allocation information, the UAV cluster is controlled to take off asynchronously in stages, complete the sub-cluster division and hierarchical formation calibration in the target airspace, and form the initial cluster formation. Based on the real-time spatial pose of each UAV in the initial cluster formation, attitude adjustment instructions and display splicing calibration instructions are generated. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. Based on the spatial topology characteristics of the complete advertising display surface, the original advertising data is processed for viewpoint adaptation and partitioned pixel mapping to generate target video frame data that adapts to each display partition and meets the time synchronization requirements. Based on a high-precision synchronous clock reference and the UAV display delay prediction results, a display trigger command is generated to control each flexible LED display to synchronously play the target video frame data.

[0008] Preferably, the generation of formation control instructions and task allocation information includes: Based on the virtual cluster model constructed from task requirements and airspace data, initial task allocation information is obtained; the initial task allocation information includes the display partitions of the flexible LED display screen of each UAV and the target airspace location. Based on the wind disturbance data predicted by the airspace hydrodynamics simulation, a pre-compensation offset is added to the target airspace position in the initial task allocation information to generate enhanced airspace coordinate information. Based on the relationship between the enhanced airspace coordinates and the UAVs that execute adjacent display partitions, multiple UAV sub-clusters are determined, and a collaborative reference machine is assigned to each sub-cluster, generating corresponding formation constraint instructions.

[0009] Preferably, based on formation control instructions and task allocation information, the UAV swarm is controlled to take off asynchronously in stages, and sub-swarm division and hierarchical formation calibration are completed in the target airspace, including: Based on the altitude information in the formation control command, plan the takeoff paths in batches and at staggered times, and control the drone swarm to take off in sequence; Based on the target airspace coordinates of each UAV, a virtual navigation rail is generated, and each UAV is controlled to fly independently along its navigation rail to the target airspace. Based on the local relative positioning of each UAV within a UAV sub-swarm relative to its cooperative reference machine, the UAVs within the sub-swarm are controlled to perform progressive pose adjustments, thereby completing the fine formation construction within the UAV sub-swarm. Based on visual observation of the overall formation of the drone swarm, when a macroscopic deviation is detected, the system controls the designated benchmark drone sub-swarm to adjust its overall pose and controls the remaining drone sub-swarms to follow, until the global accuracy requirements are met.

[0010] Preferably, based on the real-time spatial pose of each drone, attitude adjustment commands and display splicing calibration commands are generated. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed, including: A dynamic splicing gap model is established based on the real-time physical relative relationship between adjacent LED displays and the preset observation point perspective. Based on the aforementioned dynamic splicing gap model, with the goal of minimizing physical gaps, the UAV attitude adjustment command is generated. Based on the dynamic splicing gap model and ambient lighting information, when physical gaps cannot be completely eliminated, pixel-level content rendering compensation instructions for optical compensation are generated. Based on the posture adjustment instructions and content rendering compensation instructions, a display screen splicing calibration instruction set is formed; Based on the actual display effect of the collected images, the dynamic splicing gap model and calibration instruction set are iteratively optimized.

[0011] Preferably, considering the spatial topology characteristics of the complete advertising display surface, the original advertising data undergoes viewpoint adaptation processing and partitioned pixel mapping to generate target video frame data that adapts to each display partition and meets time synchronization requirements, including: Based on the spatial topology of each LED display screen and the coordinates of the preset observation points, the original advertising data is subjected to perspective correction and surface fitting pre-deformation processing. Based on display partition information, high-precision synchronization timestamps, and estimation of processing delay, a pixel mapping algorithm is used to convert the pre-deformed data into a time-synchronized target video frame data sequence. Based on the monitoring results of the real-time network and computing load of the drone swarm, the encoding and distribution strategy is dynamically selected, and the target video frame data is distributed to each drone.

[0012] Preferably, based on a high-precision synchronous clock reference and the UAV display delay prediction result, the generated display trigger command includes: Based on a quantum clock source, it provides an ultra-high precision time reference for all drones; Based on historical and real-time data of drone movement, airspace disturbances, and node load, a prediction model is used to obtain the probability distribution of future display delay for each drone. Based on the aforementioned time base and display delay probability distribution, a display trigger command containing a global absolute trigger timestamp and a pre-trigger action command is generated for each UAV. Based on the causal inference assessment of the cluster communication link status, the distribution strategy for the display trigger command is dynamically selected.

[0013] Preferably, the method further includes: During the display of the advertisement, the pixel-level deviation between the actual displayed image and the expected image is obtained based on visual feedback; Based on the pixel-level deviation, each UAV sub-cluster generates real-time fine-tuning amounts for its own airspace coordinates and display parameters through a distributed algorithm; Based on the real-time fine-tuning, the control commands and display data of each UAV are updated to achieve dynamic stabilization and calibration of the image.

[0014] A second aspect of the present invention also provides an aerial advertising system for a drone carrying a flexible LED display screen, comprising: The generation module is used to generate formation control instructions and task allocation information based on the requirements of aerial advertising tasks and preset advertising screen parameters, combined with UAV swarm control logic and airspace environment prediction data. The initial formation control module is used to control the UAV cluster to take off asynchronously in stages according to the formation control instructions and task allocation information, and to complete the sub-cluster division and hierarchical formation calibration in the target airspace to form the initial cluster formation. The docking module is used to generate attitude adjustment commands and display splicing calibration commands based on the real-time spatial pose of each UAV in the initial cluster formation. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. The processing module is used to perform viewpoint adaptation processing and partition pixel mapping on the original advertising data based on the spatial topology characteristics of the complete advertising display surface, and generate target video frame data that adapts to each display partition and meets the time synchronization requirements. The control module is used to generate display trigger commands based on a high-precision synchronous clock reference and the UAV display delay prediction results, and control each flexible LED display to synchronously play target video frame data.

[0015] Compared with the prior art, the beneficial effects of the present invention are: Large-scale drone swarms are divided into multiple sub-swarms for hierarchical and phased formation construction and calibration. First, fine-grained formation adjustments based on relative positioning are performed within each sub-swarm, followed by macroscopic overall calibration at the swarm level. When splicing flexible displays, a novel approach combines physical gap elimination through drone attitude fine-tuning with pixel-based optical compensation based on ambient light perception to dynamically construct a seamless and uniform advertising display surface. Through sub-swarm hierarchical calibration and physical-optical collaborative splicing technology, a large-area display surface spliced ​​from numerous small flexible screens in the air achieves a near-integrated screen viewing effect with small gaps and uniform brightness, solving the visual consistency problem of multi-screen splicing in the air and greatly enhancing the visual impact and professionalism of the advertisement.

[0016] To address the spatial topology of the aerial dynamic splicing display surface and the perspective of specific ground observation points, the original advertising data undergoes perspective correction and surface fitting pre-deformation processing to ensure undistorted images viewed from the ground, achieving "what you see is what you get." Simultaneously, based on a high-precision synchronous clock and a predictive model of processing and communication delays at each node, video frame data with absolute timestamps is generated. Combined with monitoring of the cluster's real-time network and computational load, encoding and distribution strategies are dynamically selected to ensure efficient and synchronous distribution of massive amounts of display data to each UAV node. Phased asynchronous launch and sub-cluster collaboration strategies reduce the complexity and risk of large-scale simultaneous cluster operations. Pre-compensation based on airspace prediction and dynamic distribution strategies based on load awareness enable the system to proactively adapt to dynamic environments such as wind disturbances and communication fluctuations, improving mission robustness and execution success rate. A closed-loop dynamic stabilization mechanism ensures continuous image stability during long-duration flight displays.

[0017] A quantum clock provides an ultra-high-precision time reference, and based on real-time and historical data of UAV motion status, environmental disturbances, and node load, a predictive model is used to obtain the probability distribution of future display latency for each UAV. Based on this, personalized display trigger commands containing pre-triggered actions are generated for each UAV. During command distribution, the optimal path is dynamically selected based on causal inference of the communication link status. Furthermore, during playback, a closed loop is formed through visual feedback, and distributed algorithms are used to generate real-time fine-tuning commands, achieving dynamic stabilization and calibration of the displayed image. By introducing an ultra-high-precision clock reference and personalized, predictive trigger commands based on predictive models, processing and display latency caused by differences in position, motion, and load among different UAVs is effectively compensated, achieving pixel-level synchronized playback at the millisecond or even microsecond level, avoiding image tearing and misalignment, and ensuring the smoothness and integrity of dynamic video content. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a schematic diagram of an aerial advertising system that uses a drone to carry a flexible LED display screen.

[0020] Figure 2 This is a schematic diagram of an aerial advertising method using a drone carrying a flexible LED display screen. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] Example 1 like Figure 1 As shown in the figure, this embodiment discloses an aerial advertising method for a drone carrying a flexible LED display screen, the method comprising: Based on the requirements of aerial advertising missions and preset advertising screen parameters, and combined with UAV swarm control logic and airspace environment prediction data, formation control instructions and task allocation information are generated; specifically, based on the requirements of aerial advertising missions and preset advertising screen parameters, and combined with UAV swarm control preset logic, the formation control instructions of the UAV swarm and the task allocation information of each UAV are determined. The task allocation information includes the display partition and airspace coordinates of the flexible LED display screen corresponding to each UAV. It should be noted that, based on the mission requirements and preset advertising screen parameters of aerial advertising, and combined with the preset logic of drone swarm control, the formation control commands for the drone swarm and the task allocation information for each drone are determined as follows: A virtual cluster model is constructed based on task requirements and real-time airspace data. Multiple candidate cluster partitioning schemes, coordinate mapping parameters, and anti-disturbance control strategies are simulated in parallel. By evaluating the predicted display effects and system stability indicators of each candidate strategy, a virtual strategy library containing the optimal strategy and its corresponding expected task allocation information is generated. Based on the optimal strategy in the virtual strategy library, calculate and generate physically executable initial task allocation information; the initial task allocation information includes initial display partition information and initial spatial coordinate information; Based on the wind disturbance data predicted by the airspace hydrodynamics simulation, a time-forward pre-compensation offset sequence is added to the initial airspace coordinate data to form enhanced airspace coordinate information with strong anti-disturbance; at the same time, a formation constraint instruction with dynamic tolerance threshold is generated for each UAV sub-cluster composed of UAVs performing adjacent display partition tasks and bound to its enhanced airspace coordinates. Meanwhile, a collaborative reference machine is assigned to each drone sub-cluster, and the origin information of the local collaborative coordinate system within the group is given to it, so as to carry out distributed collaborative fine-tuning in the subsequent formation building and display stages.

[0024] In detail, the multi-source real-time airspace data includes at least dynamic no-fly zone boundaries, airspace airflow gradient distribution, three-dimensional topology of surrounding obstacles, and communication link quality mapping data. Based on this virtual cluster model, a multi-objective optimization algorithm is used to simulate multiple candidate cluster partitioning schemes, coordinate mapping parameters, and anti-interference control strategies in parallel. The candidate schemes cover different sub-cluster sizes, coordinate mapping accuracy, and anti-interference intensity levels. By quantitatively evaluating the predicted display effects (including screen integrity and viewing angle adaptability) and system stability indicators (including cluster response latency and energy consumption redundancy rate) of each candidate strategy, a virtual strategy library containing the optimal strategy and its corresponding expected task allocation information is generated.

[0025] Based on the optimal strategy in the virtual strategy library, combined with the UAV power redundancy threshold, the load limit of the flexible LED display screen, and the communication link bandwidth constraint, physically executable initial task allocation information is calculated and generated; the initial task allocation information includes initial display partition information and initial airspace coordinate information, wherein the display partition information is associated with the pixel dot matrix range of each UAV display screen, and the initial airspace coordinate information is based on the global coordinate system for positioning.

[0026] The prediction window length of the offset sequence is adaptively adjusted based on wind disturbance frequency, and the offset amplitude matches the maximum pitch / yaw rate of the UAV, forming enhanced airspace coordinate information with strong anti-disturbance capabilities. Simultaneously, based on the display partition correlation and airspace location clustering results, UAVs performing tasks in adjacent display partitions are divided into UAV sub-clusters. For each UAV sub-cluster, a formation constraint instruction with a dynamic tolerance threshold is generated, bound to its enhanced airspace coordinates. This dynamic tolerance threshold is adjusted in real-time according to the sub-cluster size and display partition complexity; the larger the sub-cluster size and the more complex the partition edges, the smaller the tolerance threshold.

[0027] Based on the formation control instructions and task allocation information, the drone swarm is controlled to take off asynchronously in stages, complete the sub-swarm division and hierarchical formation calibration in the target airspace, and form an initial swarm formation; specifically, based on the formation control instructions and task allocation information, control signals are sent to each drone in the drone swarm to control each drone to take off with a flexible LED display screen and accurately arrive at the designated airspace, and each drone completes the initial formation arrangement according to the formation control instructions; It should be noted that, based on formation control commands and task allocation information, control signals are sent to each drone in the drone swarm to control each drone carrying the flexible LED display screen to take off and accurately arrive at the designated airspace. Furthermore, each drone completes its initial formation arrangement according to the formation control commands, including: Based on the altitude difference threshold in the formation control command and the airspace coordinates in the task allocation information, the UAV cluster is grouped according to a preset altitude layer, and a batch-by-batch, staggered takeoff path is planned. The first control signal is sent to each UAV to control each UAV to take off sequentially and asynchronously to a safe altitude according to the planned path. After each UAV takes off to a safe altitude, each UAV generates a virtual navigation rail pointing from its safe altitude position to the airspace coordinates specified in its task allocation information. A second control signal is sent to each UAV to control each UAV to initially assemble towards the target airspace along its own virtual rail in an independent navigation mode. The virtual rail is equipped with dynamic safety boundaries in both the horizontal and vertical directions. When all drones in the same drone sub-cluster arrive at the preset neighborhood of their designated airspace coordinates, the third control signal is sent to each drone in the group, using the collaborative reference machine in the drone sub-cluster as a temporary reference point. This controls each drone in the drone sub-cluster to adjust precisely to the relative pose specified by the task allocation information based on local relative positioning in a progressively convergent manner, thereby completing the internal fine formation construction of each drone sub-cluster. After all UAV sub-clusters have completed their internal formation construction, the relative positions and overall formation outlines of each UAV sub-cluster are evaluated based on wide-angle visual observation from the ground. If a macroscopic deviation is detected between the overall formation and the preset formation, a fourth control signal is sent to the designated reference drone sub-cluster to control it to perform overall translation or rotation. The remaining drone sub-clusters follow and adjust based on the established cooperative relationship until the initial formation formed by the entire drone cluster meets the global display accuracy requirements.

[0028] In detail, the drone swarm is grouped according to preset altitude layers using an improved K-means clustering algorithm. The core of the algorithm is to construct a three-dimensional clustering model with core objectives, constraints, and weight allocation. The core objectives are to ensure the integrity of sub-clusters (ensuring that drones in adjacent display zones belong to the same sub-cluster, avoiding cross-cluster connections in subsequent splicing) and to minimize the total length of the takeoff path. The constraints include airspace taboo rules for areas with high incidence of airflow vortices and obstacle distribution. Priorities are balanced through fixed weight allocation (display zone correlation, path efficiency, and airspace safety). During clustering iteration, an abnormal sample dynamic removal strategy is added to identify samples that deviate from the cluster center due to airflow and obstacle influences in real time and redistribute them to ensure that the number of sub-clusters and airspace distribution accurately match the subsequent splicing and takeoff requirements. Based on the clustering results, a hierarchical staggered takeoff path planning strategy is adopted. That is, combining the obstacle topology, communication quality, and airflow stability of each altitude layer, a B-spline smooth path is planned for each batch of drones, and a first control signal is sent to each drone to control them to take off asynchronously to a safe altitude.

[0029] After each UAV ascends to a safe altitude, a positioning accuracy enhancement strategy is adopted, which integrates GPS / IMU combined positioning data with differential correction signals from ground base stations to optimize positioning accuracy to the centimeter level. Based on this, a three-dimensional parametric virtual navigation rail is generated and adapted to the improved clustering results. The differentiated design of the rail path ensures that the rails of UAVs in the same sub-swarm do not interfere with each other.

[0030] Once all UAVs in the same sub-cluster arrive at the designated airspace and preset neighborhood, a 100Hz high refresh rate centimeter-level positioning network is constructed within the cluster using a collaborative baseline UAV as a reference point. This enables real-time perception of the relative positions of each UAV within the sub-cluster. A third control signal is sent to each UAV within the cluster, and a progressively converging PID algorithm (avoiding adjustment oscillations) is used to accurately correct the pose based on local positioning data. Infrared positioning marker data at the edge of the display screen is collected to predict the feasibility of splicing adjacent screens in advance. If there is a risk of splicing, the pose is adjusted synchronously. A phased asynchronous takeoff strategy is adopted to reduce the risk of large-scale cluster takeoff. Based on airspace turbulence prediction and airborne load perception, feedforward control compensation and dynamic data distribution scheduling are achieved. Furthermore, a visual feedback closed-loop calibration mechanism is introduced. Through distributed algorithms, the pixel-level deviation between the actual displayed image and the target image is analyzed in real time to generate fine-tuning amounts for pose and display parameters, achieving dynamic stabilization and online calibration of the advertising image during long-term flight. The entire system can proactively adapt to complex airspace interference such as wind disturbance, communication fluctuations, and changes in computing load, significantly improving mission success rate and continuous operation capability.

[0031] Based on the real-time spatial pose of each UAV in the initial cluster formation, attitude adjustment commands and display splicing calibration commands are generated. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. Specifically, based on a preset display splicing adjustment strategy and combined with the actual airspace coordinates of each UAV, UAV attitude adjustment commands and display splicing calibration commands are generated. The attitude adjustment commands control each UAV to adjust its spatial attitude, and the splicing calibration commands control each flexible LED display to complete edge alignment splicing to form a complete advertising display surface. It should be noted that, based on the preset display screen splicing adjustment strategy and combined with the actual airspace coordinates of each UAV, the generated UAV attitude adjustment commands and display screen splicing calibration commands include: Based on the actual airspace coordinates and attitudes of each UAV, the physical relative relationships between adjacent flexible LED displays in three-dimensional space are calculated in real time, including spacing, height difference, normal angle and line-of-sight occlusion relationship. Using the forward observation point as a viewpoint reference, the physical relative relationship is mapped to a dynamic splicing gap model on the virtual display plane that is related to the viewpoint. Based on the dynamic splicing gap model, with the goal of minimizing the physical gap width, the drone attitude adjustment command is generated by reverse calculation; the drone attitude adjustment command controls the drone to perform micro-translation and yaw / pitch rotation to directly reduce the physical gap between the screens. When physical gaps cannot be completely eliminated, a pixel-level content rendering compensation instruction is generated based on the dynamic splicing gap model. The content rendering compensation instruction controls the edge pixels of adjacent displays to perform brightness gradient superposition and color coordinate interpolation, optically simulating a seamless transition area. The system can sense ambient light intensity and color temperature in real time, and dynamically adjust the pixel compensation parameters in the optical layer calibration command based on the ambient light intensity and color temperature. Based on the aforementioned UAV attitude adjustment instructions and the adjusted content rendering compensation instructions, a complete set of display screen splicing calibration instructions is formed. After the display splicing calibration command is executed, the actual display effect is collected by the image sensor deployed at the forward observation point; the texture continuity, brightness uniformity and color consistency of the splicing area in the collected image are analyzed by algorithm; if residual visible seams are detected, the information is used as feedback to iteratively optimize the parameters of the dynamic splicing gap model and the weights of each command in the display splicing calibration command set.

[0032] A detailed, dynamically stitched gap model is constructed: High-precision RTK positioning data from various UAVs, inertial measurement unit (IMU) attitude data, and vision-based relative pose estimation data from neighboring UAVs are integrated and fused using a Kalman filter to obtain real-time physical relative relationships (spacing, height difference, and normal angle) between adjacent screens, superior to those from a single positioning source. The establishment of the virtual display plane is not a simple projection. Based on the position of the forward observation point (which can be dynamically specified, such as GPS coordinates of a ground control station), the system calculates a unique homography transformation matrix for each pair of adjacent screens. This matrix precisely maps the screen edge segments in three-dimensional space onto a virtual sphere centered on the observation point, and then unfolds it into a two-dimensional plane. The model quantifies the gaps into multiple parameters: the basic physical gap width, the width of the V-shaped optical dark area caused by the screen normal angle, and the length of the stepped shadow caused by the height difference. Based on these parameters, the system classifies gaps into physically eliminateable and optically compensated types, providing a decision-making basis for subsequent command generation.

[0033] Multiple physical constraints are strictly applied during the solution process: the attitude adjustment of a single UAV must be within its stable flight envelope to ensure a safe and smooth adjustment process; after adjustment, the three-dimensional distance between any two UAVs must be greater than a dynamic safety threshold, which is adjusted in real time according to wind speed and UAV maneuverability. Under the premise of meeting the gap reduction objective, the adjustment scheme with the minimum total energy consumption or the most balanced energy consumption of each UAV is prioritized. Sequential quadratic programming or model predictive control algorithms are used to solve this optimization problem online, outputting a set of feasible and optimized micro-attitude adjustment commands in real time.

[0034] In this embodiment, when the physical adjustment reaches its limit, intelligent optical compensation is activated. Based on the dynamic splicing gap model, a pixel band of a certain width (e.g., 3-10 pixels) is drawn at the edges of adjacent screens as a fusion transition zone. The width of this zone is proportional to the physical width and angle of the gap. For the pixels within the fusion transition zone, a bidirectional alpha mixing and color space interpolation algorithm is used.

[0035] For brightness, based on the pixel's distance from the screen edge, its original brightness value is reduced using a Gaussian decay curve, while simultaneously superimposed with brightness information from corresponding pixels on adjacent screens, which is decayed using an inverse curve. For color, in Color space interpolation is used to maintain a smooth transition in hue and saturation, avoiding color banding at the seams.

[0036] Equipped with an ambient light sensor, it acquires real-time ambient light intensity and color temperature. A compensation parameter-ambient light lookup table is established, automatically increasing the overall brightness and contrast enhancement coefficient of pixels in the fusion area under strong light to combat detail loss caused by sunlight glare. When there is a large color temperature deviation, subtle color temperature shift compensation is applied to pixels in the fusion area to make them more harmonious with the surrounding ambient light and reduce visual jarring.

[0037] Based on the spatial topology characteristics of the complete advertising display surface, the original advertising data is processed for viewpoint adaptation and partitioned pixel mapping to generate target video frame data that is adapted to each display partition and meets the time synchronization requirements. Specifically, the original advertising data is obtained, and the original advertising data is segmented based on the display partition information. The target video frame data that is adapted to the flexible LED display screen carried by each UAV is obtained through a pixel mapping algorithm. The target video frame data meets the resolution and splicing coordination requirements of each display partition. It should be noted that the original advertising data is segmented based on display partition information, and the target video frame data adapted to the flexible LED displays carried by various drones is obtained through a pixel mapping algorithm, including: Based on the actual airspace coordinates and real-time attitude of each UAV after the flexible LED display screen completes edge alignment and splicing to form a complete advertising display surface, as well as the preset forward observation point coordinates, the three-dimensional spatial arrangement topology of the flexible LED display screens within each UAV sub-cluster is analyzed, and the original advertising data is pre-deformed based on this topology. The pre-deformation processing includes perspective correction and surface fitting of the images corresponding to each display screen area in the original advertising data according to the perspective of the forward observation point, so as to offset the visual geometric distortion caused by the fluctuation of the aerial formation and the non-coplanarity of the screens. Based on display partition information, a spatiotemporal joint pixel mapping algorithm is used to convert the pre-deformed original advertising data into a target video frame data sequence. The pixel mapping algorithm, based on display partition information, divides and resamples each global video frame into sub-frames that correspond one-to-one with the physical pixel matrix of each flexible LED display screen, and embeds a high-precision synchronization timestamp and frame sequence number into each sub-frame. At the same time, based on the estimated processing delay of each node in the UAV cluster from receiving the command to screen refresh, the generation time of the sub-frame is forward-lookingly offset to ensure that all displays refresh the corresponding frame at the same absolute time point. Based on the monitoring results of the real-time network link quality of the UAV cluster and the computational load status of each airborne processing unit, the encoding strategy and distribution granularity of the target video frame data are dynamically selected, and the target video frame data is transmitted based on the selected encoding strategy and distribution granularity. For UAV sub-clusters with stable links and low load, uncompressed or low-compressed high-fidelity sub-frame data streams are distributed. For UAV sub-clusters with poor links or high load, compressed data packets using inter-frame differential coding and keyframe indexing are distributed, or the sub-frame data is split into multiple data packets and reliably transmitted through multiple paths.

[0038] In detail, a display screen topology graph is constructed, where nodes represent each flexible LED display screen and edges represent adjacency relationships. Each edge stores 3D transformation parameters calculated from the actual spatial coordinates and real-time attitude, including: relative rotation matrix, translation vector, and screen normal vector. A virtual camera coordinate system is defined with the forward observation point coordinates as the camera optical center and the vector pointing to the geometric center of the display surface as the optical axis direction. Based on this, a projection transformation matrix from the world coordinate system to the virtual camera screen coordinate system is calculated for each display screen.

[0039] For the region corresponding to the i-th display screen in the original advertising data, reverse texture mapping is performed using its corresponding projection transformation matrix. That is, each physical pixel of the target display screen is reverse-projected onto the corresponding region of the original advertising image for sampling, thereby directly obtaining the corrected sub-image.

[0040] To address nonlinear distortion caused by screen non-coplanarity, the system employs triangulation and barycentric coordinate interpolation. The screen area is triangulated into a mesh, and the precise texture coordinates of each vertex in the original ad data are calculated. For pixels within the mesh, the texture coordinates of the three vertices are interpolated using the barycentric coordinates of their respective triangles, achieving smooth surface texture mapping and perfectly fitting non-planar displays.

[0041] The spatiotemporal joint pixel mapping algorithm is as follows: Based on a dynamic delay database, an end-to-end delay model is established for each UAV (or a group of UAVs of the same model) to determine the total end-to-end delay of each UAV. In the formula, To perform dynamic prediction based on real-time network round-trip time and packet loss rate, a Kalman filter is used. This is an estimation based on the current CPU load and the historical processing time of the frame type to be decoded (I / P / B frames); This is due to inherent hardware latency, factory calibration. Let the absolute refresh time of the global target be... For UAV i, the absolute time of its subframe generation and transmission , For safety margin, based on the above formula, the generation time is calculated independently for each subframe of each frame and written into the scheduling queue.

[0042] Based on a high-precision synchronous clock reference and UAV display delay prediction results, a display trigger command is generated to control each flexible LED display screen to synchronously play target video frame data. Specifically, based on formation control commands and task allocation information, control signals are sent to each UAV in the UAV cluster to control each UAV to take off with the flexible LED display screen and accurately arrive at the designated airspace, and each UAV completes the initial formation arrangement according to the formation control commands; Based on the completed display splicing calibration results and target video frame data, display trigger commands are sent synchronously to each drone to control each flexible LED display to synchronously display the corresponding target video frame data, thereby achieving stable display of the complete aerial advertising image.

[0043] It should be noted that, based on the completed display splicing calibration results and target video frame data, the display trigger commands sent synchronously to each drone include: Based on the ground-based master quantum clock source, a pair of photon clock signals with quantum entanglement characteristics are generated. One of the photon clock signals is distributed to the quantum receivers carried by each UAV through a free-space optical communication link, so that all UAV nodes can share an ultra-high precision and interference-resistant absolute time reference based on quantum entanglement. Before playback, the historical and real-time collected 3D motion trajectory of the drone, the airspace turbulence spectrum, the network delay matrix between each node, and the load fluctuation data of the airborne processor are input into a trained spatiotemporal prediction neural network model to predict in advance the probability distribution of display delay caused by motion, disturbance and computing load of each drone in the next few frames. Based on the display latency probability distribution, two types of instructions are generated for each frame of data of each drone: a global absolute trigger timestamp instruction based on a quantum clock and a pre-trigger action instruction. The pre-trigger action instruction includes a preload buffer and a specific advance amount for starting the rendering pipeline. The global absolute trigger timestamp and the personalized pre-trigger action instruction are jointly encapsulated into a display trigger instruction; Before distributing the trigger command, a lightweight causal inference engine is started to evaluate the health status of the current cluster communication link and potential interference sources in real time; based on the evaluation results, the optimal distribution strategy is dynamically selected. Each UAV receives the display trigger command through a dedicated high-priority channel and immediately parses out the global absolute trigger timestamp and the local pre-trigger action command.

[0044] In detail, the master quantum clock source is deployed at a ground control center, employing an entangled photon pair generator based on a spontaneous parametric down-conversion (SPDC) process to continuously generate polarization- or energy-time entangled photon pairs at a fixed repetition frequency (e.g., 100 MHz). One beam of reference photons is retained locally for time reference locking, while the other beam of signal photons, after phase distortion caused by atmospheric turbulence is corrected by an adaptive optics compensation module (including a wavefront sensor and deformable mirror), is directionally broadcast to each UAV via a free-space optical communication link. Each UAV carries a quantum receiver containing a single-photon detector array and a time-to-digital converter (TDC), achieving sub-nanosecond time synchronization with the ground master clock using the Bell state measurement principle, and dynamically assessing the reliability of the clock link by real-time monitoring of entanglement fidelity.

[0045] The spatiotemporal prediction neural network model is a multimodal fusion architecture, including: Trajectory encoding branch: Input the 3D position / velocity / acceleration sequence of 10 historical frames, and use a graph convolutional network to model the relative motion constraints between UAVs; Environmental disturbance branch: Input is the spatial turbulence spectrum characteristics retrieved from the airborne MEMS anemometer and lidar. (frequency band power spectral density) Communication and computing load branch: input the end-to-end network delay matrix (sampling rate 1 kHz) measured between nodes and the sliding window of GPU / CPU utilization; The model output is a Gaussian mixture distribution parameter (mean) of the display delay of each drone in several future frames. Standard deviation ), used to quantify uncertainty.

[0046] For the target video data in frame t, the system performs the following operations: First, calculate the global absolute trigger timestamp. ; The absolute start time of playback, obtained through quantum clock synchronization. This represents the video frame interval.

[0047] Based on the predicted Gaussian mixture distribution parameters Generate pre-triggered action commands for the i-th drone, including: The amount of time required for the i-th drone to start data loading in advance for frame t. ; This is for the safety factor.

[0048] The i-th drone should initiate the rendering process at that absolute moment. ; This is a fixed measured value, representing the average time required from the start of the rendering pipeline to the readiness of frame data. and It is encapsulated as a structured instruction packet, with added digital signatures and frame sequence numbers to prevent replay attacks.

[0049] The method further includes: During task execution, the pixel-level deviation between the actual displayed image and the expected image is obtained through visual feedback; Each UAV sub-cluster dynamically generates fine-tuning amounts for the current airspace coordinates and display parameters based on its collaborative baseline machine and distributed algorithm, according to the pixel-level deviation. The real-time control commands and display data of each UAV are updated according to the fine-tuning amount, thereby achieving dynamic stabilization and display calibration of the aerial advertising screen.

[0050] In detail, high frame rate global vision sensors (such as 200 fps panoramic cameras or distributed multi-view camera arrays) deployed on ground or air observation platforms, using a quantum clock synchronized with the LED display as a reference, at each frame t display time... Capture the current over-the-air advertisement image within a preset delay; compare the captured image with the expected target frame. After performing subpixel-level optical flow alignment and color space normalization, the pixel-level residual map is calculated: ; To normalize the image coordinates, This represents the luminance / chromaticity deviation field.

[0051] Each sub-cluster dynamically elects a cooperative baseline machine as a local reference node for its sub-cluster through a distributed consensus protocol (such as the maximum remaining power or minimum position jitter criterion).

[0052] This embodiment also discloses an aerial advertising system for drones carrying flexible LED displays, including: The generation module is used to generate formation control instructions and task allocation information based on the requirements of aerial advertising tasks and preset advertising screen parameters, combined with UAV swarm control logic and airspace environment prediction data. The initial formation control module is used to control the UAV cluster to take off asynchronously in stages according to the formation control instructions and task allocation information, and to complete the sub-cluster division and hierarchical formation calibration in the target airspace to form the initial cluster formation. The docking module is used to generate attitude adjustment commands and display splicing calibration commands based on the real-time spatial pose of each UAV in the initial cluster formation. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. The processing module is used to perform viewpoint adaptation processing and partition pixel mapping on the original advertising data based on the spatial topology characteristics of the complete advertising display surface, and generate target video frame data that adapts to each display partition and meets the time synchronization requirements. The control module is used to generate display trigger commands based on a high-precision synchronous clock reference and the UAV display delay prediction results, and control each flexible LED display to synchronously play target video frame data.

[0053] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0054] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0055] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.

[0056] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0057] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0058] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0059] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0060] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for aerial advertising using a drone carrying a flexible LED display screen, characterized in that, The method includes: Based on the requirements of aerial advertising missions and preset advertising screen parameters, and combined with UAV swarm control logic and airspace environment prediction data, formation control instructions and mission allocation information are generated. Based on the formation control instructions and task allocation information, the UAV cluster is controlled to take off asynchronously in stages, complete the sub-cluster division and hierarchical formation calibration in the target airspace, and form the initial cluster formation. Based on the real-time spatial pose of each UAV in the initial cluster formation, attitude adjustment instructions and display splicing calibration instructions are generated. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. Based on the spatial topology characteristics of the complete advertising display surface, the original advertising data is processed for viewpoint adaptation and partitioned pixel mapping to generate target video frame data that adapts to each display partition and meets the time synchronization requirements. Based on a high-precision synchronous clock reference and the UAV display delay prediction results, a display trigger command is generated to control each flexible LED display to synchronously play the target video frame data.

2. The aerial advertising method for a drone carrying a flexible LED display screen according to claim 1, characterized in that, The generated formation control instructions and task allocation information include: The initial task allocation information is obtained by constructing a virtual cluster model based on task requirements and spatial data. Based on the wind disturbance data predicted by the airspace hydrodynamics simulation, a pre-compensation offset is added to the airspace coordinates in the initial task allocation information to generate enhanced airspace coordinate information. Based on the relationship between the enhanced airspace coordinates and the UAVs that execute adjacent display partitions, multiple UAV sub-clusters are determined, and a collaborative reference machine is assigned to each sub-cluster, generating corresponding formation constraint instructions.

3. A method for aerial advertising using a drone carrying a flexible LED display screen according to claim 2, characterized in that, Based on formation control instructions and task allocation information, the drone swarm is controlled to take off asynchronously in stages, completing sub-swarm division and hierarchical formation calibration in the target airspace, including: Based on the altitude information in the formation control command, plan the takeoff paths in batches and at staggered times, and control the drone swarm to take off in sequence; Based on the target airspace coordinates of each UAV, a virtual navigation rail is generated, and each UAV is controlled to fly independently along its navigation rail to the target airspace. Based on the local relative positioning of each UAV within a UAV sub-swarm relative to its cooperative reference machine, the UAVs within the sub-swarm are controlled to perform progressive pose adjustments, thereby completing the fine formation construction within the UAV sub-swarm. Based on visual observation of the overall formation of the drone swarm, when a macroscopic deviation is detected, the system controls the designated benchmark drone sub-swarm to adjust its overall pose and controls the remaining drone sub-swarms to follow, until the global accuracy requirements are met.

4. A method for aerial advertising using a drone carrying a flexible LED display screen according to claim 3, characterized in that, Based on the real-time spatial pose of each drone, attitude adjustment commands and display splicing calibration commands are generated. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed, including: A dynamic splicing gap model is established based on the real-time physical relative relationship between adjacent LED displays and the preset observation point perspective. Based on the aforementioned dynamic splicing gap model, with the goal of minimizing physical gaps, the UAV attitude adjustment command is generated. Based on the dynamic splicing gap model and ambient lighting information, when physical gaps cannot be completely eliminated, pixel-level content rendering compensation instructions for optical compensation are generated. Based on the posture adjustment instructions and content rendering compensation instructions, a display screen splicing calibration instruction set is formed; Based on the actual display effect of the collected images, the dynamic splicing gap model and calibration instruction set are iteratively optimized.

5. A method for aerial advertising using a drone carrying a flexible LED display screen according to claim 4, characterized in that, Based on the spatial topology characteristics of the complete advertising display surface, the original advertising data undergoes viewpoint adaptation processing and partitioned pixel mapping to generate target video frame data that adapts to each display partition and meets time synchronization requirements, including: Based on the spatial topology of each LED display screen and the coordinates of the preset observation points, the original advertising data is subjected to perspective correction and surface fitting pre-deformation processing. Based on display partition information, high-precision synchronization timestamps, and estimation of processing delay, a pixel mapping algorithm is used to convert the pre-deformed data into a time-synchronized target video frame data sequence. Based on the monitoring results of the real-time network and computing load of the drone swarm, the encoding and distribution strategy is dynamically selected, and the target video frame data is distributed to each drone.

6. A method for aerial advertising using a drone carrying a flexible LED display screen according to claim 5, characterized in that, Based on a high-precision synchronous clock reference and UAV display latency prediction results, the generated display trigger commands include: Based on a quantum clock source, it provides an ultra-high precision time reference for all drones; Based on historical and real-time data of drone movement, airspace disturbances, and node load, a prediction model is used to obtain the probability distribution of future display delay for each drone. Based on the aforementioned time base and display delay probability distribution, a display trigger command containing a global absolute trigger timestamp and a pre-trigger action command is generated for each UAV. Based on the causal inference assessment of the cluster communication link status, the distribution strategy for the display trigger command is dynamically selected.

7. A method for aerial advertising using a drone carrying a flexible LED display screen according to claim 6, characterized in that, The method further includes: During the display of the advertisement, the pixel-level deviation between the actual displayed image and the expected image is obtained based on visual feedback; Based on the pixel-level deviation, each UAV sub-cluster generates real-time fine-tuning amounts for its own airspace coordinates and display parameters through a distributed algorithm; Based on the real-time fine-tuning, the control commands and display data of each UAV are updated to achieve dynamic stabilization and calibration of the image.

8. An aerial advertising system for a drone carrying a flexible LED display screen, implementing the aerial advertising method for a drone carrying a flexible LED display screen as described in any one of claims 1 to 7, characterized in that, include: The generation module is used to generate formation control instructions and task allocation information based on the requirements of aerial advertising tasks and preset advertising screen parameters, combined with UAV swarm control logic and airspace environment prediction data. The initial formation control module is used to control the UAV cluster to take off asynchronously in stages according to the formation control instructions and task allocation information, and to complete the sub-cluster division and hierarchical formation calibration in the target airspace to form the initial cluster formation. The docking module is used to generate attitude adjustment commands and display splicing calibration commands based on the real-time spatial pose of each UAV in the initial cluster formation. Through attitude fine-tuning and display edge calibration, combined with optical compensation strategies, a complete advertising display surface is constructed. The processing module is used to perform viewpoint adaptation processing and partition pixel mapping on the original advertising data based on the spatial topology characteristics of the complete advertising display surface, and generate target video frame data that adapts to each display partition and meets the time synchronization requirements. The control module is used to generate display trigger commands based on a high-precision synchronous clock reference and the UAV display delay prediction results, and control each flexible LED display to synchronously play target video frame data.

Images