Low-altitude aircraft cooperative scheduling method, system, device and storage medium

Abstract

This invention relates to the field of flight scheduling technology, specifically providing a method, system, device, and storage medium for collaborative scheduling of low-altitude aircraft. The method includes: edge nodes receiving real-time data from a swarm of drones and generating corresponding scheduling data using a pre-deployed scheduling model; if the edge nodes confirm that the real-time data is environmental mutation data, they upload the environmental mutation data to a regional cloud; the regional cloud matches the environmental mutation data with historical scenario data; if the match is successful, scheduling data is generated; if the match fails, the environmental mutation data is uploaded to a central cloud; the central cloud generates scheduling data based on the received environmental mutation data and returns the scheduling data to the regional cloud, which then distributes the scheduling data to the edge nodes. This invention constructs a low-altitude traffic management system that combines real-time performance, reliability, and self-evolution capabilities.

Description

Technical Field

[0001] This invention belongs to the field of flight scheduling technology, specifically relating to a method, system, device and storage medium for collaborative scheduling of low-altitude aircraft. Background Technology

[0002] The dynamic changes in the low-altitude traffic environment pose a severe challenge to real-time response. Temporary buildings and construction facilities in urban airspace may alter the airspace structure within hours, and sudden micro-scale meteorological phenomena such as strong winds and fog can often form within 10 minutes, far exceeding the response capabilities of traditional cloud-based models.

[0003] Existing systems rely on periodic model updates, requiring a complete process of data collection, cleaning, training, and validation, typically iterating weekly. This leads to the inability to synchronize newly emerging obstacle layers and meteorological parameters in a timely manner. This problem is particularly prominent in drone swarm operations. When an aircraft encounters a temporary communication tower not recorded in the system, the traditional navigation model may fail to avoid obstacles due to its lack of real-time reconstruction capabilities. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the prior art, the present invention provides a method, system, device and storage medium for coordinated scheduling of low-altitude aircraft to solve the above-mentioned technical problems.

[0005] In a first aspect, the present invention provides a method for cooperative scheduling of low-altitude aircraft, comprising:

[0006] Edge nodes receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model;

[0007] If the edge node confirms that the real-time data is environmental mutation data, it will upload the environmental mutation data to the regional cloud.

[0008] The regional cloud matches the environmental mutation data with historical scenario data. If the match is successful, scheduling data is generated. If the match is unsuccessful, the environmental mutation data is uploaded to the central cloud.

[0009] The central cloud generates scheduling data based on the received environmental mutation data, returns the scheduling data to the regional cloud, and then the regional cloud distributes the scheduling data to the edge nodes.

[0010] In an optional implementation, the method further includes:

[0011] Building a three-tier federation architecture:

[0012] At the edge layer, deploy KubeEdge edge nodes to run real-time data processing and lightweight inference;

[0013] Regional cloud, Kubernetes cluster, responsible for regional data aggregation and model training;

[0014] The central cloud, acting as the federal control plane, manages multiple regional clouds through KubeFed and performs global scheduling.

[0015] In one optional implementation, the edge node receives real-time data from the drone swarm and generates corresponding scheduling data using a pre-deployed scheduling model, including:

[0016] Edge nodes receive real-time data from the drone swarm, including motion status data, location data, and image data.

[0017] Edge nodes utilize locally deployed scheduling models to generate scheduling data and confidence levels based on the real-time data;

[0018] Once the confidence level is confirmed to have reached the set confidence threshold, the scheduling data is sent to the drone swarm.

[0019] In one optional implementation, if the edge node confirms that the real-time data is environmental mutation data, it uploads the environmental mutation data to the regional cloud, including:

[0020] If the confidence level does not reach the confidence threshold, the edge node will upload the real-time data to the regional cloud as environmental mutation data.

[0021] In an optional implementation, the method further includes:

[0022] For data on sudden environmental changes, edge nodes will receive scheduling data from the regional cloud and distribute it to the drone swarm.

[0023] The edge node obtains updated data generated after the drone swarm executes the scheduling data;

[0024] The edge node integrates the environmental mutation data, scheduling data, and update data into new scenario data, and uploads the new scenario data to the regional cloud.

[0025] In an optional implementation, the method further includes:

[0026] The regional cloud guides the scheduling models of multiple edge nodes to perform federated training based on the newly generated scenario data of each node.

[0027] In an optional implementation, the method further includes:

[0028] The central cloud determines flight restriction rules for multiple regions and distributes these rules to the corresponding regional clouds. The regional clouds generate regional-level scheduling instructions for multiple drone swarms based on the received flight restriction rules and send these instructions to the corresponding edge nodes. The edge nodes generate scheduling parameters for the drone swarms based on the received regional-level scheduling instructions and send these parameters to the drone swarms.

[0029] Secondly, the present invention provides a low-altitude aircraft cooperative scheduling system, comprising:

[0030] The first scheduling module is used by edge nodes to receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model.

[0031] The second scheduling module is used by the edge node to upload the environmental mutation data to the regional cloud if the edge node confirms that the real-time data is environmental mutation data.

[0032] The third scheduling module is used by the regional cloud to match the environmental mutation data with historical scene data. If the match is successful, scheduling data is generated; if the match is unsuccessful, the environmental mutation data is uploaded to the central cloud.

[0033] The fourth scheduling module is used by the central cloud to generate scheduling data based on the received environmental change data, and return the scheduling data to the regional cloud, which then distributes the scheduling data to the edge nodes.

[0034] Thirdly, a device is provided, comprising:

[0035] Memory used to store the low-altitude aircraft coordination scheduling program;

[0036] A processor is configured to implement the steps of the low-altitude aircraft cooperative scheduling method as provided in the first aspect when executing the low-altitude aircraft cooperative scheduling program.

[0037] Fourthly, a computer-readable storage medium is provided, on which a low-altitude aircraft cooperative scheduling program is stored, wherein when the low-altitude aircraft cooperative scheduling program is executed by a processor, the steps of the low-altitude aircraft cooperative scheduling method provided in the first aspect are implemented.

[0038] The beneficial effects of this invention are as follows: the low-altitude aircraft collaborative scheduling method, system, equipment, and storage medium provided by this invention achieve real-time control of UAV swarms through edge computing, intelligently filter invalid data to reduce data transmission; regional cloud scene matching reuses verified solutions to reduce decision-making risks; and the central cloud supports the processing of unknown events and drives the continuous evolution of the model. Therefore, this invention constructs a low-altitude traffic management system that combines real-time performance, reliability, and self-evolutionary capabilities.

[0039] Furthermore, the design principle of this invention is reliable, the structure is simple, and it has a very wide range of application prospects. Attached Figure Description

[0040] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a schematic flowchart of a method according to an embodiment of the present invention.

[0042] Figure 2 This is a schematic block diagram of a system according to an embodiment of the present invention.

[0043] Figure 3 This is a schematic diagram of the structure of a device provided in an embodiment of the present invention. Detailed Implementation

[0044] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0045] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0046] The low-altitude aircraft collaborative scheduling method provided in this embodiment of the invention is executed by computer equipment, and correspondingly, the low-altitude aircraft collaborative scheduling system runs on the computer equipment.

[0047] Figure 1 This is a schematic flowchart illustrating a method according to an embodiment of the present invention. Wherein, Figure 1 The implementing entity can be a low-altitude aircraft collaborative scheduling system. Depending on different requirements, the order of steps in this flowchart can be changed, and some steps can be omitted.

[0048] like Figure 1 As shown, the method includes:

[0049] S1. Edge nodes receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model;

[0050] S2. If the edge node confirms that the real-time data is environmental mutation data, it will upload the environmental mutation data to the regional cloud;

[0051] S3. The regional cloud matches the environmental mutation data with historical scenario data. If the match is successful, scheduling data is generated. If the match is unsuccessful, the environmental mutation data is uploaded to the central cloud.

[0052] S4. The central cloud generates scheduling data based on the received environmental mutation data, and returns the scheduling data to the regional cloud, which then distributes the scheduling data to the edge nodes.

[0053] First, construct a three-tier federated architecture:

[0054] By using the Kubeedge and Senda frameworks, a three-tier federated structure of "edge node - regional cloud - central cloud" is constructed to realize edge-cloud federated learning and collaborative reasoning.

[0055] The edge layer (drones / edge base stations, edge devices) schedules lightweight algorithm containers locally to designated edge devices via KubeEdge; the regional cloud (regional KubeEdge cluster intelligent computing center) is responsible for jointly training models and scheduling clusters with edge devices, and generating regional optimization models by aggregating edge model parameters within its jurisdiction; the central cloud (national data center) is responsible for cross-regional cluster docking and unified national data management, ensuring that aircraft can be safely scheduled by clusters nationwide.

[0056] The low-altitude aircraft terminal prioritizes real-time tasks (obstacle avoidance, basic identification). When encountering complex scenarios (such as poor visibility in rainy weather or unfamiliar obstacles), it transmits key data to the regional cloud computing center. The regional cloud integrates information from multiple devices to generate an optimization solution (multi-view analysis, dynamic model switching). If the problem still cannot be solved (such as when historical big data or high-precision models are needed), it calls upon the center's cloud resources for in-depth processing.

[0057] On the edge, differential privacy is employed, and Laplace noise is added during gradient updates to ensure that individual data is untraceable. At the transport layer, homomorphic encryption is used to protect model parameters, which cannot be cracked even if intermediate nodes are hijacked. In the cloud, a Trusted Execution Environment (TEE) is deployed to complete global model aggregation within a secure enclave.

[0058] In one embodiment of the present invention, based on step S1, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0059] 1. Technical details of real-time data reception.

[0060] Communication Protocol: Low-Power Wide-Area Network (LPWAN) protocols (such as LoRaWAN and NB-IoT) are employed to achieve long-distance, low-power communication between the drone swarm and edge nodes, suitable for complex terrain and wide-area coverage scenarios. For image data transmission, an optimized H.265 video encoding protocol is used, and a stable connection is established through the TCP / IP protocol stack to reduce data transmission latency and packet loss rate. Simultaneously, to ensure the reliability and security of data transmission, the DTLS (Datagram Transport Layer Security) protocol is used to encrypt data transmission, preventing data leakage and tampering.

[0061] Data Acquisition and Preprocessing: The UAV is equipped with high-precision sensors (such as an inertial measurement unit (IMU), a GPS / BeiDou dual-mode positioning module, and a vision sensor) to collect motion state data (velocity, acceleration, attitude angle), position data (latitude and longitude, altitude), and image data in real time. Before data transmission, the raw data is preprocessed, including noise reduction (using a Kalman filter algorithm to process motion state data and position data to remove noise interference), data compression (using a JPEGXR compression algorithm for image data to reduce data volume while ensuring image quality), and format conversion (converting different formats of data collected by the sensors into JSON format for easy parsing by edge nodes).

[0062] Data Buffering and Synchronization: A circular data buffer is established at the edge node, with a reasonable buffer size set (dynamically adjusted according to the drone's data transmission frequency and data volume) to prevent data overflow and loss. A timestamp synchronization mechanism is adopted, adding a high-precision timestamp (accurate to the microsecond level) to each data packet. Time synchronization between the drone and the edge node is achieved through the NTP (Network Time Protocol), ensuring that the received real-time data has an accurate time sequence and providing a reliable basis for subsequent scheduling decisions.

[0063] 2. Technical details of scheduling model construction and scheduling data generation.

[0064] Scheduling Model Architecture: A reinforcement learning-based scheduling model is constructed using deep learning frameworks (such as TensorFlow and PyTorch), with Deep Q-Network (DQN) or Policy Gradient Algorithm (PPO) as the basic architecture. The model input layer receives preprocessed real-time data (motion state data, position data, and image data), extracts features through multi-layer convolutional neural networks (CNN) and recurrent neural networks (RNN), performs feature fusion and complex relationship modeling in the hidden layers, and generates scheduling data (including UAV flight path adjustment instructions, task allocation schemes, resource scheduling strategies, etc.) and corresponding confidence scores in the output layer.

[0065] Model Training and Optimization: The scheduling model is trained using historical flight data, simulated scenario data, and real-world mission data. A reward function is constructed to guide the model in learning the optimal scheduling strategy. This reward function comprehensively considers factors such as task completion efficiency (e.g., task execution time, coverage), resource consumption (e.g., power consumption, flight distance), and safety (e.g., collision avoidance, maintaining stable communication links). The Mini-Batch Gradient Descent algorithm is used to update model parameters, and the model is periodically fine-tuned using new real-time data to adapt to environmental changes and mission requirements.

[0066] Scheduling data generation: After receiving real-time data, edge nodes input the data into the trained scheduling model. Based on the characteristics of the input data and the current environmental state, the model calculates the optimal scheduling data through forward propagation. Simultaneously, it uses uncertainty estimation methods (such as Monte Carlo dropout or Bayesian neural networks) to calculate the confidence level of each scheduling data point. The confidence level reflects the model's reliability in generating the scheduling data, and its value ranges from 0 to 1.

[0067] 3. Confidence assessment and technical details of scheduling data distribution.

[0068] Confidence threshold setting: The confidence threshold is dynamically set based on actual task requirements and risk tolerance. For example, for high-risk, critical tasks (such as disaster relief and monitoring of important targets), the confidence threshold is set to 0.9 or higher; for general tasks, the threshold can be appropriately lowered to 0.8. The threshold setting is continuously optimized through experiments and practical applications to improve the efficiency of scheduling decisions while ensuring the effectiveness of task execution.

[0069] Confidence Assessment and Decision-Making: Edge nodes compare the confidence level generated by the scheduling model with a set threshold. If the confidence level reaches or exceeds the threshold, the scheduling data is considered reliable, and it is encapsulated into a control command package in a specific format (containing scheduling data, verification information, target UAV identifiers, etc.). If the confidence level does not reach the threshold, the edge nodes can take the following measures: First, re-invoke the scheduling model and regenerate scheduling data and confidence levels based on new real-time data; second, activate a backup scheduling scheme (such as a scheme generated by a rule-based heuristic scheduling algorithm) to ensure that the UAV swarm can continuously execute tasks.

[0070] Scheduling data delivery: Encapsulated scheduling data is delivered to the drone swarm via a reliable communication link. An ACK (Acknowledgment) mechanism ensures reliable data transmission; after receiving the scheduling data, each drone sends an acknowledgment message to the edge node. If the edge node does not receive an acknowledgment message within a specified time, it retransmits the scheduling data. Simultaneously, to avoid network congestion caused by multiple drones simultaneously receiving data, Time Division Multiplexing (TDM) or Frequency Division Multiplexing (FDM) technologies are used to rationally allocate communication channels, ensuring that scheduling data is accurately and timely delivered to each drone.

[0071] In one embodiment of the present invention, based on step S3, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0072] 1. Regional cloud data storage and management.

[0073] Historical scenario data storage architecture: The regional cloud employs a distributed file system (such as Ceph or GlusterFS) to store massive amounts of historical scenario data. This data is categorized and stored according to different scenario types (such as natural disasters, major events, etc.), and each scenario type is further divided according to chronological order. Simultaneously, to improve data retrieval efficiency, NoSQL databases (such as MongoDB or Cassandra) are used to build indexes, recording key information for each historical scenario, such as scenario type, occurrence time, location, and main characteristics.

[0074] Data Updates and Maintenance: The regional cloud regularly synchronizes the latest historical scenario data from the central cloud or other data sources to ensure data timeliness. Simultaneously, expired or invalid data is cleaned up to free up storage space. During data updates, an incremental update approach is used, updating only the changed data portions to reduce data transfer volume and storage overhead.

[0075] 2. Environmental mutation data preprocessing.

[0076] Data cleaning: When the regional cloud receives data on sudden environmental changes, it first cleans the data to remove noise, outliers, and duplicate data. For example, abnormal temperature values ​​in meteorological data (which may be caused by sensor malfunctions) can be identified and removed using statistical methods (such as the 3σ principle).

[0077] Feature extraction: Key features are extracted from the cleaned data, and these features will be used for matching with historical scene data. Different features are extracted for different types of environmental abrupt change data. For example, for earthquake data, features such as magnitude, focal depth, and epicenter location can be extracted; for fire data, features such as fire area, fire spread rate, and wind direction can be extracted.

[0078] 3. Historical scene data matching algorithm.

[0079] Similarity Calculation Methods: Multiple similarity calculation methods are employed to assess the matching degree between environmental mutation data and historical scene data. For numerical features, methods such as Euclidean distance and Manhattan distance can be used to calculate similarity; for textual features, methods such as cosine similarity and edit distance can be used. For example, Euclidean distance is used to calculate similarity for numerical features such as temperature and humidity in meteorological data; cosine similarity is used for descriptive event information (such as the specific details of a disaster).

[0080] Matching rules and threshold settings: Set matching rules and thresholds to determine whether a match is successful. For example, when the similarity between environmental mutation data and historical scene data in multiple key features exceeds a certain threshold (e.g., 80%), a match is considered successful. Matching rules and thresholds can be dynamically adjusted according to different scene types and data characteristics.

[0081] 4. Generation of scheduling data.

[0082] Template-based generation method: If environmental mutation data matches historical scenario data, the regional cloud will generate scheduling data based on the scheduling template corresponding to the matched historical scenario data. The scheduling template is predefined and contains information such as resource allocation and emergency response measures in that scenario. For example, for an earthquake scenario, the scheduling template may include the dispatch plan for rescue teams and the allocation plan for medical supplies.

[0083] Data Customization and Optimization: When generating scheduling data, the scheduling template is customized and optimized based on the specific circumstances of current environmental changes. For example, the number of rescue teams and the supply of medical supplies are adjusted according to the magnitude and affected area of ​​the current earthquake. Simultaneously, current resource availability and real-time conditions are considered to ensure the feasibility and effectiveness of the scheduling data.

[0084] 5. Data is uploaded to the central cloud.

[0085] Data encapsulation and encryption: If environmental mutation data cannot be matched with historical scenario data, the regional cloud will encapsulate and encrypt the environmental mutation data. Secure encryption algorithms (such as AES) are used to encrypt the data to ensure its security during transmission. Simultaneously, metadata such as data type, collection time, and source information is added to the data encapsulation.

[0086] Upload Protocol and Strategy: A reliable upload protocol (such as HTTP / 2 or MQTT) is used to upload encrypted data to the central cloud. To improve upload efficiency and reliability, a strategy of chunked upload and resumable uploads is employed. Simultaneously, the upload rate is dynamically adjusted based on network conditions to avoid excessive network pressure.

[0087] In one embodiment of the present invention, based on step S4, the following will provide a possible embodiment and describe its specific implementation in a non-limiting manner.

[0088] The central cloud generates scheduling data based on received environmental change data, returns the scheduling data to the regional cloud, and then the regional cloud distributes the scheduling data to the edge nodes.

[0089] Based on the above embodiments, in order to further improve scheduling accuracy, a model update method is provided as a feasible approach:

[0090] 1. Dispatch data to the drone swarm.

[0091] The edge nodes and the drone swarm employ low-latency, high-reliability communication protocols, such as ZigBee or 5G NR-U (5G unlicensed frequency band). For control commands requiring real-time interaction (such as flight path adjustments), 5G NR-U is used to ensure low-latency transmission; for general status information synchronization, ZigBee can be used to reduce power consumption.

[0092] The scheduling data received from the regional cloud is encapsulated into data packets of a specific format, with information such as source address (edge ​​node address), destination address (drone address), and data type added. The data packets are encrypted using a symmetric encryption algorithm (such as AES), and the encryption key is updated periodically to ensure the security of data transmission.

[0093] Based on the topology and task allocation of the drone swarm, scheduling data is distributed using broadcast, multicast, or unicast methods. General scheduling commands are broadcast, while personalized commands for specific drones are unicast. A retransmission mechanism is also implemented; if a drone does not acknowledge receipt within a specified time, the edge node will retransmit the data.

[0094] 2. Obtain updated data on the drone swarm.

[0095] The drone collects and updates data at preset time intervals or event-triggered mechanisms, including flight status (speed, altitude, attitude), mission progress, and environmental perception data (such as temperature, humidity, and air quality). The data collection frequency is dynamically adjusted based on mission requirements and drone resource availability.

[0096] To reduce data transmission volume, the UAV preprocesses the collected updated data, including data compression (such as using the lossless compression algorithm LZ77) and feature extraction. Simultaneously, adaptive modulation and coding techniques are employed to dynamically adjust the transmission rate and coding scheme based on channel conditions, thereby improving data transmission efficiency.

[0097] After receiving updated data from multiple drones, the edge nodes aggregate and fuse the data. Multi-sensor fusion algorithms (such as Kalman filtering) are used to fuse different types of updated data, improving the accuracy and reliability of the data.

[0098] 3. Added scene data integration and uploading.

[0099] Edge nodes integrate environmental change data, scheduling data, and update data according to certain rules to form new scene data. Data association algorithms can be used to link data from different sources to construct a complete scene description. Simultaneously, metadata such as timestamps and geographic locations are added to the new scene data to facilitate subsequent analysis and management.

[0100] Before uploading new scenario data, edge nodes assess the data quality. This includes checking the data's completeness, accuracy, and consistency, and repairing or marking missing or abnormal data. Data quality is quantitatively evaluated using data quality assessment metrics such as missing data rate and error rate.

[0101] New scenario data is uploaded to the regional cloud using a reliable upload protocol (such as HTTP / 2 or MQTT). To improve upload efficiency, a strategy of chunked upload and resumable upload is adopted. Simultaneously, the uploaded data is encrypted to ensure data security during transmission.

[0102] 4. Regional cloud-guided federated training.

[0103] Employing a horizontal federated learning architecture, multiple edge nodes exchange and update model parameters through a regional cloud without sharing the original data. The regional cloud acts as a coordinator, responsible for initializing the global model, distributing model parameters, and aggregating local model parameters.

[0104] The regional cloud selects a suitable model architecture for the scheduling task, such as a convolutional neural network (CNN) or a recurrent neural network (RNN). The global model is initialized, and the initial model parameters are distributed to each edge node.

[0105] Edge nodes perform local training using locally generated new scene data based on the received global model parameters. After each training cycle, the edge nodes calculate the gradient of the local model parameters and upload the gradient information to the regional cloud.

[0106] The regional cloud receives gradient information from each edge node, uses an aggregation algorithm (such as FedAvg) to aggregate local model parameters, and updates the global model parameters. The updated global model parameters are then distributed to each edge node for the next round of training.

[0107] Set training termination conditions, such as reaching a preset number of training rounds or convergence of model performance metrics (such as accuracy and loss function value). When the termination conditions are met, the training process ends, and the regional cloud distributes the final global model parameters to each edge node for actual scheduling tasks.

[0108] In a specific example, we can create a real-time incremental update mechanism and edge-triggered model iteration.

[0109] When an edge node detects a sudden change in the local environment (such as discovering a new obstacle through an anomaly detection algorithm), it automatically initiates local model fine-tuning; the fine-tuning gradient is encrypted and uploaded to the regional cloud, triggering local federated learning (only neighboring nodes participate); the regional cloud can complete model iteration and distribution in a short time, which is much faster than traditional global updates. Using a dynamic weight allocation strategy, the federated aggregation weights are dynamically adjusted based on the data freshness of the edge nodes (e.g., real-time data collected by drones = 0.9, historical data = 0.1) and scene coverage (city node weight > desert node weight), making the model more biased towards the latest high-value data.

[0110] Based on the above embodiments, the following method is further provided:

[0111] The central cloud determines flight restriction rules for multiple regions and distributes these rules to the corresponding regional clouds. The regional clouds generate regional-level scheduling instructions for multiple drone swarms based on the received flight restriction rules and send these instructions to the corresponding edge nodes. The edge nodes generate scheduling parameters for the drone swarms based on the received regional-level scheduling instructions and send these parameters to the drone swarms.

[0112] Specifically, it includes the following steps:

[0113] 1. The central cloud determines flight restriction rules.

[0114] Rule data source:

[0115] Laws and regulations: The cloud center regularly obtains information on national and local laws and regulations concerning drone flights from official channels (such as the website of the Civil Aviation Administration), including information on no-fly zones, restricted flight altitudes, and flight time restrictions.

[0116] Real-time data monitoring: Utilizing satellite remote sensing, meteorological monitoring stations, ground-based radar, and other equipment, information such as the geographical environment, meteorological conditions, and airspace activities is monitored in real time. For example, when a large-scale event, natural disaster, or military exercise is detected in a certain area, that area is designated as a temporary no-fly zone.

[0117] Historical data analysis: Analyze past drone flight data to identify areas and time periods where conflicts or dangers are likely to occur, and use this as a reference for formulating flight restriction rules.

[0118] Rule making and updating:

[0119] Rule-making algorithm: Flight restriction rules are formulated using a combination of expert systems and machine learning algorithms. The expert system sets basic rules based on laws, regulations, and industry experience, while the machine learning algorithm optimizes and adjusts the rules through analysis of real-time and historical data.

[0120] Rule update mechanism: The central cloud monitors data changes in real time and updates flight restriction rules promptly when new laws and regulations, environmental changes, or major events occur. Simultaneously, the rules are regularly evaluated and optimized to ensure their effectiveness and rationality.

[0121] Standardized rule format:

[0122] Standardize flight restriction rules into a standardized format, such as JSON or XML, to facilitate subsequent distribution and processing. The rules should include information such as the restricted area (represented by latitude and longitude coordinates), the restricted time range, the restriction type (no-fly zone, altitude restriction, speed restriction, etc.), and the restriction level.

[0123] 2. The central cloud distributes flight restriction rules to regional clouds.

[0124] Secure and reliable communication protocols, such as HTTPS or MQTT over TLS, are employed to ensure the security and integrity of rule data during transmission. Simultaneously, data compression techniques (such as Gzip) are used to compress the rule data to improve transmission efficiency.

[0125] Distribution strategy: Based on the geographical location, business needs, and network conditions of the regional cloud, a hierarchical distribution or broadcast distribution strategy will be adopted. For large regional clouds, a hierarchical distribution strategy can be adopted, first distributing rules to the regional central node, and then the central node distributes them to its subordinate child nodes; for small regional clouds, a broadcast distribution strategy can be adopted, directly sending rules to all regional clouds.

[0126] Distribution confirmation mechanism: After receiving the flight restriction rules, the regional cloud sends a confirmation message to the central cloud. If the central cloud does not receive a confirmation message within the specified time, it will resend the rule data to ensure that each regional cloud accurately receives the latest rules.

[0127] 3. The regional cloud generates regional-level scheduling instructions.

[0128] Data Fusion and Analysis: The regional cloud integrates and analyzes the received flight restriction rules with local drone swarm information (such as location, status, and mission). Geographic Information System (GIS) technology is used to visualize the flight restriction rules, intuitively displaying the locational relationship between restricted areas and drones.

[0129] Scheduling Algorithm Selection: Based on different scenarios and requirements, select appropriate scheduling algorithms, such as genetic algorithms, ant colony algorithms, or simulated annealing algorithms. These algorithms can optimize the flight paths, task allocation, and resource utilization of drone swarms while meeting flight restriction rules, thereby improving scheduling efficiency and safety.

[0130] Command Generation and Optimization: Based on the results of the scheduling algorithm, regional-level scheduling commands are generated. The commands include information such as the UAV's flight path, task allocation, flight speed, and altitude. Simultaneously, the generated commands are optimized, taking into account factors such as the UAV's endurance, communication capabilities, and payload capacity to ensure the commands' feasibility and effectiveness.

[0131] 4. The regional cloud issues regional-level scheduling instructions to edge nodes.

[0132] Communication protocols and data formats: Use the same communication protocols (such as HTTPS or MQTT over TLS) and data formats (such as JSON or XML) as the central cloud distribution rules to ensure data consistency and compatibility.

[0133] Command distribution strategy: Based on the geographical location of the edge nodes and the distribution of the drone swarm, either group distribution or unicast distribution strategies are adopted. For a centrally distributed drone swarm, the edge nodes can be grouped, and commands are sent to the master node within the group, which then distributes them to other nodes within the group. For a dispersed drone swarm, unicast distribution can be used, directly sending commands to each edge node.

[0134] Command Confirmation and Feedback: After receiving a regional-level scheduling command, the edge node sends a confirmation message to the regional cloud and provides feedback on the command's execution status. Based on the feedback, the regional cloud promptly adjusts the scheduling commands to ensure the flight safety of the drone swarm and the smooth execution of its missions.

[0135] 5. Edge nodes generate scheduling parameters and send them to the drone swarm.

[0136] Parameter Refinement and Adjustment: Based on the received regional-level scheduling instructions, and considering the specific circumstances of the local UAV (such as model, performance, and current status), the edge node refines and adjusts the instructions to generate specific scheduling parameters. For example, it adjusts the flight speed and altitude based on the UAV's endurance; and optimizes the communication strategy based on the UAV's communication range.

[0137] Data format conversion: The scheduling parameters are converted into a format that the UAV can recognize and execute, such as binary code or a specific instruction set. Simultaneously, to ensure data accuracy and reliability, verification information (such as CRC checksums) is added.

[0138] Communication Protocol and Transmission Method: A dedicated protocol for communicating with UAVs, such as ZigBee or 5G NR-U, is used to send scheduling parameters to the UAV swarm. Depending on the distribution of the UAVs and communication requirements, parameters are transmitted via broadcast, multicast, or unicast. A retransmission mechanism is also implemented; if a UAV does not acknowledge receipt within a specified time, the edge node will retransmit the parameters.

[0139] In some embodiments, the low-altitude aircraft cooperative scheduling system may include multiple functional modules composed of computer program segments. The computer programs for each program segment in the low-altitude aircraft cooperative scheduling system may be stored in the memory of a computer device and executed by at least one processor to perform (see details). Figure 1 (Description) Function of coordinated scheduling of low-altitude aircraft.

[0140] In this embodiment, the low-altitude aircraft cooperative scheduling system can be divided into multiple functional modules according to the functions it performs, such as... Figure 2 As shown. The module referred to in this invention is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and is stored in memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0141] The first scheduling module is used by edge nodes to receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model.

[0142] The second scheduling module is used by the edge node to upload the environmental mutation data to the regional cloud if the edge node confirms that the real-time data is environmental mutation data.

[0143] The third scheduling module is used by the regional cloud to match the environmental mutation data with historical scene data. If the match is successful, scheduling data is generated; if the match is unsuccessful, the environmental mutation data is uploaded to the central cloud.

[0144] The fourth scheduling module is used by the central cloud to generate scheduling data based on the received environmental change data, and return the scheduling data to the regional cloud, which then distributes the scheduling data to the edge nodes.

[0145] Figure 3 The low-altitude aircraft cooperative scheduling method provided in the embodiments of this application can be applied to devices. Those skilled in the art will understand that the device structure involved in the embodiments of this invention does not constitute a limitation on the device. A device may include more or fewer components than illustrated, or combine certain components, or have different component arrangements. In the embodiments of this invention, the device includes, but is not limited to, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The device may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the embodiments of this application described and / or claimed herein.

[0146] The device 300 may include a processor 310, a memory 320, and a communication unit 330. These components communicate via one or more buses. Those skilled in the art will understand that the server structure shown in the figure does not constitute a limitation of the present invention. It may be a bus topology or a star topology, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0147] The memory 320 can be used to store execution instructions of the processor 310. The memory 320 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. When the execution instructions in the memory 320 are executed by the processor 310, the device 300 is able to perform some or all of the steps in the above method embodiments.

[0148] The processor 310 serves as the control center of the storage device, connecting various parts of the electronic device via various interfaces and lines. It executes software programs and / or modules stored in the memory 320, and calls data stored in the memory to perform various functions of the electronic device and / or process data. The processor can be composed of integrated circuits (ICs), such as a single packaged IC or multiple packaged ICs with the same or different functions connected together. For example, the processor 310 may consist only of a central processing unit (CPU). In this embodiment of the invention, the CPU may have a single processing core or include multiple processing cores.

[0149] The communication unit 330 is used to establish a communication channel, enabling the storage device to communicate with other devices. It can receive user data sent by other devices or send user data to other devices.

[0150] The present invention also provides a computer storage medium, wherein the computer storage medium may store a program, which, when executed, may include some or all of the steps provided in the embodiments of the present invention. The storage medium may be a magnetic disk, an optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0151] Those skilled in the art will clearly understand that the techniques in the embodiments of the present invention can be implemented using software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium such as a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other medium capable of storing program code. It includes several instructions to cause a computer device (which may be a personal computer, a server, or a second device, network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0152] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0153] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or modules may be electrical, mechanical, or other forms.

[0154] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0155] In addition, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0156] Although the present invention has been described in detail with reference to the accompanying drawings and preferred embodiments, the present invention is not limited thereto. Various equivalent modifications or substitutions can be made to the embodiments of the present invention by those skilled in the art without departing from the spirit and essence of the invention, and such modifications or substitutions should all be within the scope of the present invention. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should also be covered within the protection scope of the present invention.

Claims

1. A method for cooperative scheduling of low-altitude aircraft, characterized in that, include: Edge nodes receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model; If the edge node confirms that the real-time data is environmental mutation data, it will upload the environmental mutation data to the regional cloud. The regional cloud matches the environmental mutation data with historical scenario data. If the match is successful, scheduling data is generated. If the match is unsuccessful, the environmental mutation data is uploaded to the central cloud. The central cloud generates scheduling data based on the received environmental change data, returns the scheduling data to the regional cloud, and then the regional cloud distributes the scheduling data to the edge nodes. Edge nodes receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model, including: Edge nodes receive real-time data from the drone swarm, including motion status data, location data, and image data. Edge nodes utilize locally deployed scheduling models to generate scheduling data and corresponding confidence levels based on the real-time data; Once the confidence level is confirmed to have reached the set confidence threshold, the scheduling data is sent to the drone swarm. If the edge node confirms that the real-time data is environmental mutation data, it will upload the environmental mutation data to the regional cloud, including: If the confidence level does not reach the confidence threshold, the edge node will upload the real-time data to the regional cloud as environmental mutation data.

2. The method according to claim 1, characterized in that, The method further includes: Building a three-tier federation architecture: At the edge layer, deploy KubeEdge edge nodes to run real-time data processing and lightweight inference; Regional cloud, Kubernetes cluster, responsible for regional data aggregation and model training; The central cloud, acting as the federal control plane, manages multiple regional clouds through KubeFed and performs global scheduling.

3. The method according to claim 1, characterized in that, The method further includes: For data on sudden environmental changes, edge nodes will receive scheduling data from the regional cloud and distribute it to the drone swarm. The edge node obtains updated data generated after the drone swarm executes the scheduling data; The edge node integrates the environmental mutation data, scheduling data, and update data into new scenario data, and uploads the new scenario data to the regional cloud.

4. The method according to claim 3, characterized in that, The method further includes: The regional cloud guides the scheduling models of multiple edge nodes to perform federated training based on the newly generated scenario data of each node.

5. The method according to claim 1, characterized in that, The method further includes: The central cloud determines flight restriction rules for multiple regions and distributes these rules to the corresponding regional clouds. The regional clouds generate regional-level scheduling instructions for multiple drone swarms based on the received flight restriction rules and send these instructions to the corresponding edge nodes. The edge nodes generate scheduling parameters for the drone swarms based on the received regional-level scheduling instructions and send these parameters to the drone swarms.

6. A low-altitude aircraft cooperative scheduling system, characterized in that, include: The first scheduling module is used by edge nodes to receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model. The second scheduling module is used by the edge node to upload the environmental mutation data to the regional cloud if the edge node confirms that the real-time data is environmental mutation data. The third scheduling module is used by the regional cloud to match the environmental mutation data with historical scene data. If the match is successful, scheduling data is generated; if the match is unsuccessful, the environmental mutation data is uploaded to the central cloud. The fourth scheduling module is used by the central cloud to generate scheduling data based on the received environmental change data, and return the scheduling data to the regional cloud, which then distributes the scheduling data to the edge nodes. Edge nodes receive real-time data from the drone swarm and generate corresponding scheduling data using a pre-deployed scheduling model, including: Edge nodes receive real-time data from the drone swarm, including motion status data, location data, and image data. Edge nodes utilize locally deployed scheduling models to generate scheduling data and corresponding confidence levels based on the real-time data; Once the confidence level is confirmed to have reached the set confidence threshold, the scheduling data is sent to the drone swarm. If the edge node confirms that the real-time data is environmental mutation data, it will upload the environmental mutation data to the regional cloud, including: If the confidence level does not reach the confidence threshold, the edge node will upload the real-time data to the regional cloud as environmental mutation data.

7. A device, characterized in that, include: Memory used to store the low-altitude aircraft coordination scheduling program; A processor, configured to implement the steps of the low-altitude aircraft cooperative scheduling method as described in any one of claims 1-5 when executing the low-altitude aircraft cooperative scheduling program.

8. A computer-readable storage medium storing a computer program, characterized in that, The readable storage medium stores a low-altitude aircraft cooperative scheduling program, which, when executed by a processor, implements the steps of the low-altitude aircraft cooperative scheduling method as described in any one of claims 1-5.

Images