Cooperative control method and system for low-altitude aircraft
Through dynamic networking and distributed decision-making architecture, efficient collaborative control of low-altitude aircraft in dynamic environments is achieved, solving the problems of poor adaptability and insufficient safety in existing technologies, and improving mission completion rate and safety.
Patent Information
- Application Number
- CN202511808964.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-03-20
AI Technical Summary
Existing low-altitude aircraft cooperative control technologies have poor adaptability in dynamic environments, low group cooperation efficiency, and insufficient safety redundancy capabilities. They are difficult to achieve cross-platform cooperation and cope with sudden obstacles, and there are system integration difficulties and safety risks.
It adopts a dynamic networking and distributed decision-making architecture based on a scheduling center. Through status information acquisition, task area division, real-time monitoring and collaborative command generation, it realizes safe obstacle avoidance, path planning and task allocation within and between aircraft groups, and supports collaborative control of heterogeneous systems.
It improves the mission completion rate of low-altitude aircraft in dynamic environments, shortens the time for resolving airspace conflicts, enhances the recovery rate from communication interruptions, reduces the number of collisions, and supports collaborative operations of multiple types of aircraft.
Smart Images
Figure CN121704529A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of aircraft cooperative control, and more particularly to a cooperative control method and system for low-altitude aircraft. BACKGROUND
[0002] With the rapid development of low-altitude economy, low-altitude aircraft are increasingly widely used in fields such as logistics distribution, city inspection, and emergency rescue. However, the existing low-altitude aircraft cooperative control technology still faces multiple technical bottlenecks in actual application, which restricts its large-scale deployment and efficient operation.
[0003] The current mainstream cooperative control scheme mainly relies on preset path planning and static communication network architecture, which is difficult to adapt to the dynamic changing flight environment. When encountering sudden obstacles (such as bird flocks, temporary buildings) or rapid changes in network topology, the system response capability is seriously insufficient, resulting in a significant increase in flight safety risks. At the same time, the traditional centralized control architecture has the risk of single-point failure, while the pure distributed decision mechanism is difficult to achieve global cooperative optimization, especially in the multi-group task interaction scene, the task conflict and resource competition problem is prominent, and the cooperative efficiency is low. In terms of system compatibility, low-altitude aircraft of different manufacturers use heterogeneous communication protocols and control interfaces, and lack of unified technical standards, making cross-platform cooperation difficult to achieve. This fragmented technology ecosystem not only increases the difficulty of system integration, but also hinders the cooperative operation ability of multiple types of aircraft. In addition, the existing systems generally lack robustness design for communication interruption, equipment failure and other abnormal situations, and lack of safety redundancy mechanism, which is easy to cause collision accidents, seriously restricting the reliability and safety of the low-altitude aircraft cooperative system.
[0004] Although the existing technical solutions have explored, such as proposing a 5G-based unmanned aerial vehicle formation control method, they have not effectively solved the real-time cooperation problem under dynamic ad hoc networks; using reinforcement learning to optimize single-group path planning, but it does not cover the complex cooperative needs of the multi-group interaction scene. These technical limitations show that there is an urgent need for an innovative cooperative control architecture that can balance dynamic environmental adaptability, group cooperative efficiency, heterogeneous system compatibility, and safety redundancy capability to promote substantial breakthroughs in low-altitude aircraft cooperative control technology.
[0005] Therefore, how to improve the dynamic environmental adaptability, group cooperative efficiency, and safety redundancy capability of low-altitude aircraft cooperative control is a problem that needs to be solved by those skilled in the art. SUMMARY
[0006] Therefore, the present application provides a cooperative control method and system for low-altitude aircraft, which improves the dynamic environmental adaptability, group cooperative efficiency, and safety redundancy capability of low-altitude aircraft cooperative control.
[0007] To achieve the above object, the present application adopts the following technical solutions: A cooperative control method for low-altitude aircraft, comprising: Obtaining group state information corresponding to a plurality of control nodes based on a dispatch center; Dividing a plurality of task areas based on the group state information; Generating an optimal control scheme based on the task areas and the group state information to control corresponding groups to work; Monitoring emergency situations in the group working process in real time based on the dispatch center and generating a first fly-around scheme to control the group to execute; The control node generates cooperative instructions based on state space data and controls the aircraft in the group to work cooperatively; The aircraft obtains obstacle information in real time during the working process and automatically avoids obstacles when communication is interrupted.
[0008] In one embodiment, a plurality of task areas are obtained, specifically comprising: The group state information includes aircraft position, aircraft speed, aircraft remaining power, aircraft task state, and aircraft load capacity; Based on the aircraft position, the aircraft speed, and meteorological parameters, an airspace dynamic division algorithm is used to divide the low-altitude airspace into a plurality of overlapping task areas; Based on the task area, the task type, priority weight, and meteorological risk coefficient are associated; The priority weight determines the scheduling order, and the meteorological risk coefficient is used for path correction.
[0009] In one embodiment, the control node determination method is: Based on the current power of each aircraft in the group and the average distance from other aircraft; Based on the ratio of the current power of the single aircraft to the maximum power to obtain a power coefficient; Based on the ratio of the average distance to the maximum distance as a distance coefficient; Based on the power coefficient and the distance coefficient to obtain a comprehensive coefficient; Select the aircraft with the maximum comprehensive coefficient in the group as the control node.
[0010] In one embodiment, the optimal control scheme is generated, specifically comprising: Based on the aircraft load capacity and the aircraft remaining power, the ability matching degree is obtained; Based on the ability matching degree, the priority weight, and the meteorological adaptation degree, the task matching degree is obtained; Select the scheme with the highest task matching degree as the optimal control scheme.
[0011] In one embodiment, generating a first detour route specifically includes: The spatial overlap degree is calculated as the ratio of the overlapping area to the total area of the task area. The ratio of the number of lost packets to the total number of data packets in the group's uploaded data is used as the communication health score. When the spatial overlap is greater than a first threshold or the communication health is greater than a second threshold, the arbitration process is triggered: Predict potential conflict trajectories based on the aircraft position and aircraft speed of each aircraft in the swarm; Based on the potential conflict trajectory, the first detour plan is generated using the improved A* algorithm and sent to the group.
[0012] In one embodiment, generating cooperative instructions specifically includes: The state space data is acquired based on the aircraft; Each of the aforementioned aircraft shares the state space data via the millimeter-wave frequency band; The control node inputs the state space data into the distributed MADDPG framework. Each of the aircraft's Actor networks receives the state space data and generates preliminary action commands; The shared Critic network aggregates the group's global state assessment Q value, feeds back the optimized initial action instructions, and generates the collaborative instructions; The state space data includes: the aircraft's own state information, neighbor's state information, and environmental state information.
[0013] In one embodiment, the method further includes: when multiple groups enter the same task area, performing cross-group collaboration to generate an airspace usage plan. Groups that enter the same task area are considered as co-domain groups; The control nodes corresponding to the same domain group are used as the same domain control nodes; A temporary Mesh network is established between the control nodes in the same domain to share status data; The dispatch center calculates the airspace usage cost based on airspace occupancy time and airspace occupancy area. The co-domain control node calculates the cost function based on the airspace usage cost; Based on the cost function, a Pareto optimal solution is generated using a BPNE solver as the spatial usage scheme that minimizes the cost function; Based on the aforementioned airspace usage scheme, the data is sent to the corresponding domain groups for execution. The control node also generates dynamic weights based on obstacle information, and generates a second bypass scheme based on the dynamic weights to control the corresponding group execution.
[0014] In one embodiment, the automatic obstacle avoidance specifically includes: The aircraft generates an obstacle probability distribution map based on multimodal sensors; When communication is interrupted, model prediction control is used for rolling optimization based on the obstacle probability distribution map to generate multiple candidate paths within a set prediction time domain; Based on the multiple candidate paths, a safety corridor constraint is introduced to ensure that the deviation between the candidate path and the original path is less than a preset value, thereby obtaining an obstacle avoidance path; Automatic obstacle avoidance is achieved based on the obstacle avoidance path.
[0015] In one embodiment, the aircraft monitors its own battery level in real time, and when its own battery level is less than a warning threshold, it acts as a low-battery aircraft and sends out a distress signal. After a nearby aircraft responds to the distress signal, it calculates the corresponding priority based on the current distance between itself and the low-battery aircraft and its current battery level. Select the nearest aircraft with the highest priority and current battery level greater than the set value as the power supply aircraft; The low-battery aircraft is powered by the aforementioned power supply aircraft.
[0016] A cooperative control system for low-altitude aircraft includes: an information acquisition module, an area division module, a group control module, an emergency response module, a cooperative control module, and an individual obstacle avoidance module; The information acquisition module is used to acquire group status information corresponding to multiple control nodes based on the scheduling center; The region division module is used to divide multiple task regions based on the group state information; The group control module is used to generate an optimal control scheme based on the task area and the group status information to control the corresponding group operation; The emergency response module is used to generate a first detour plan based on the real-time monitoring of emergency situations during the group's operation by the dispatch center, and to control the group's execution. The collaborative control module is used by the control node to generate collaborative instructions based on state space data and control the aircraft in the group to work collaboratively. The single obstacle avoidance module is used by the aircraft to acquire obstacle information in real time during operation and automatically avoid obstacles when communication is interrupted.
[0017] As can be seen from the above technical solution, compared with the prior art, this invention discloses a cooperative control method and system for low-altitude aircraft. Through dynamic networking, distributed decision-making, and global optimization, it achieves safe obstacle avoidance, path planning, task allocation, and energy efficiency optimization within and between low-altitude aircraft groups. Testing and verification show that compared with traditional methods, this invention improves the task completion rate by 19.7%, reduces airspace conflict resolution time by 57.3%, improves communication interruption recovery rate by 24.1%, and reduces the number of collisions per thousand sorties by 95.2%. It also supports heterogeneous aircraft protocol conversion and cross-group collaboration, adapting to large-scale applications in various low-altitude economic scenarios. Attached Figure Description
[0018] 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, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0019] Figure 1 The present invention provides a flowchart of a cooperative control method for low-altitude aircraft.
[0020] Figure 2 A schematic diagram of the three-level hybrid control architecture provided by the present invention.
[0021] Figure 3 This is a schematic diagram of the MADDPG algorithm training framework provided by the present invention.
[0022] Figure 4 This is a schematic diagram of the MADDPG algorithm training framework provided by the present invention.
[0023] Figure 5 This is a schematic diagram of the layered obstacle avoidance strategy provided by the present invention.
[0024] Figure 6 This invention provides a schematic diagram of a cooperative control system for low-altitude aircraft. Detailed Implementation
[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0026] Example 1 like Figure 1As shown, this embodiment of the invention discloses a cooperative control method for low-altitude aircraft, including the following steps. For ease of description, these steps are numbered S1 to S6, and these numbers are not intended to limit the sequential relationship between the various steps of this invention: S1 obtains group status information corresponding to multiple control nodes based on the scheduling center.
[0027] like Figure 2 As shown, the present invention adopts a three-level hybrid control architecture of "dispatch center-control node-aircraft". The dispatch center includes a global task allocation layer and an emergency arbitration layer, which are responsible for global task allocation and emergency arbitration, respectively, and receive data uploaded by the control node and issue instructions. like Figure 3 As shown, the control node is the core of the group layer and is served by a dynamically elected cluster head node. This cluster head node is a special individual aircraft within the group, with additional computing and communication capabilities, but it is still part of the individual level. It is responsible for dynamic self-organizing network, local decision-making and data aggregation, and interacts with the dispatch center and aircraft. The aircraft is responsible for perception, communication, and autonomous execution, and feeds back real-time data to the corresponding control nodes. All levels interact through unified data frames to form a data closed loop.
[0028] Furthermore, the group status information includes: unique aircraft identifier, aircraft position, aircraft speed, aircraft remaining battery power, aircraft mission status, and aircraft payload capacity; the group status information forms a unified data frame structure, as shown in Table 1: Table 1 Group Status Information
[0029] The aircraft mission status Task_Status uses OneHot encoding and supports parallel tasks (such as inspection + charging); the unified data frame structure supports parsing by multiple vendors' devices, where the 8-bit mask of the Task_Status field corresponds to the execution status of 8 types of tasks (such as bit0=logistics, bit1=inspection).
[0030] Furthermore, the method for determining control nodes is as follows: Based on the current battery level of each aircraft in the group and its average distance from other aircraft; The power coefficient is obtained based on the ratio of the current power to the maximum power of a single aircraft. The distance coefficient is based on the ratio of the average distance to the maximum distance. A comprehensive coefficient is obtained based on the power coefficient and distance coefficient; The aircraft with the highest comprehensive coefficient in the group is selected as the control node.
[0031] Furthermore, the overall score is: Score = 0.7·E c / E max +0.3·(1-d) a / d max ); Among them, E c E represents the current battery level. max Indicates the maximum battery capacity, d a d represents the average distance. max Indicates the maximum distance.
[0032] Furthermore, the selected control node is responsible for aggregating aircraft data, distributing commands, and managing communication resources (using a TDMA+CSMA hybrid mechanism to ensure high-priority data latency ≤20ms).
[0033] Furthermore, the RAFT consensus algorithm is adopted, and the control node, as the master node, receives the status information of the nodes (aircraft) in the group (Table 1). Data is synchronized through the "proposal-vote-commit" mechanism to ensure that data consistency can still be maintained when 30% of the nodes fail.
[0034] S2 divides multiple task regions based on group state information.
[0035] Furthermore, multiple task regions are obtained, specifically including: Based on the aircraft's position, speed, and meteorological parameters, a dynamic airspace partitioning algorithm is used to divide the low-altitude airspace into multiple overlapping mission areas. Based on the task area, associated task type, priority weight, and meteorological risk coefficient; Priority weights determine the scheduling order, while meteorological risk coefficients are used for path correction.
[0036] Furthermore, in this embodiment, the airspace dynamic partitioning algorithm is based on Voronoi diagrams and mixed integer programming (MIP); the task types include logistics, inspection, and rescue; priority weights (levels 1-5), with emergency task groups (such as rescue) marked as priority level 5; and the meteorological risk coefficient (0-1).
[0037] Furthermore, the method for obtaining the meteorological risk coefficient includes: Real-time meteorological data, including wind speed, is obtained through a meteorological service interface. v w Precipitation intensity R p ,visibility V d and turbulence index T i ; Normalize all parameters: ; ; ; ; in, v max , R max , V max , T max The safety threshold for each parameter; Based on the normalization results, the weighted comprehensive meteorological risk value is calculated. R meteor : ; in, α , β γ delta For the weighting coefficients, satisfying ; Will R meteor Mapping to the 0-1 range yields the meteorological risk coefficient.
[0038] S3 generates the optimal control scheme based on task area and group state information to control the corresponding group operation.
[0039] Furthermore, the optimal control scheme is generated, specifically including: Based on the matching degree between the aircraft's payload capacity and its remaining power acquisition capability; The task matching degree is obtained based on capability matching degree, priority weight, and weather adaptability. The scheme with the highest task matching degree is selected as the optimal control scheme.
[0040] This step utilizes aircraft data to generate network topology, supporting dynamic environmental adaptation.
[0041] Furthermore, the task matching degree R is: R=0.4×Yi+0.3×Q+0.3×N; Where Yi represents priority weight, Q represents meteorological adaptability, and N represents capability matching degree.
[0042] Furthermore, the meteorological suitability Q in the task matching degree is calculated based on the meteorological risk coefficient associated with the task area. The conversion relationship is: Q = 1 - meteorological risk coefficient. Therefore, the lower the meteorological risk of an area, the higher its meteorological suitability, and the higher its priority in task allocation.
[0043] S4, based on the real-time monitoring of emergencies during the group's operation by the dispatch center, generates the first detour plan and controls the group's execution.
[0044] Furthermore, a first detour plan is generated, specifically including: The spatial overlap degree is calculated as the ratio of the overlapping area to the total area of the task area. The ratio of packet loss to total data packets in group uploads is used as the measure of communication health. When the spatial overlap exceeds the first threshold or the communication health exceeds the second threshold, the arbitration process is triggered. Predict potential conflict trajectories based on the aircraft position and velocity of each aircraft in the swarm; Based on the potential conflict trajectory, an improved A* algorithm is used to generate a first bypass plan, which is then sent to the group.
[0045] Furthermore, in this implementation, when the airspace overlap (OR) is greater than 30% or the communication health (PLR) is greater than 15%, the arbitration process is triggered, and the arbitration result is written into the blockchain smart contract. The group synchronously verifies consistency through the millimeter wave band. This step uses early uploaded data to resolve the problem of multi-group task conflicts.
[0046] The S5 control node generates collaborative commands based on state-space data and controls the collaborative operation of aircraft within the group.
[0047] Furthermore, generating cooperative instructions specifically includes: State space data is acquired based on the aircraft; The various aircraft share state space data via millimeter-wave frequency band; Control nodes input state space data into the distributed MADDPG framework; Each aircraft's Actor network receives state space data and generates preliminary action commands; The shared Critic network aggregates the group's global state assessment Q-value, feeds back preliminary action instructions for optimization, and generates collaborative instructions; State space data includes: the aircraft's own state information, neighbor's state information, and environmental state information.
[0048] Furthermore, the aircraft's own state information (12-dimensional) includes: position (x, y, z), velocity (v... x ,v y ,v z ), remaining battery power and task progress; Neighbor status information (24 dimensions) includes: relative position, speed and mission status of neighboring aircraft within 3 hops; Environmental status information (8 dimensions) includes: obstacle distance, wind speed, and no-fly zone markings.
[0049] Furthermore, such as Figure 4As shown, the distributed MADDPG framework works as follows: Each individual Actor network receives local data and generates initial action commands (velocity increment, steering angle); a shared Critic network aggregates the group's global state (task progress, airspace occupancy) to evaluate the Q-value and provide feedback for action optimization. Training uses historical data to pre-train the model (PER and TD3 optimization); this step generates group coordination commands, dynamically adjusting formation spacing to address the problem of low control efficiency within the group.
[0050] Furthermore, the action space constraints of the cooperative instructions are as follows: Speed increment limit: △v x ∈[-5,5]m / s, △v z ∈[-2,2]m / s; where, Δv x Δv represents the change in velocity of the aircraft in the horizontal direction (x-axis, along the main flight direction). z This represents the change in velocity in the vertical direction (z-axis, altitude direction). This constraint ensures smooth aircraft movement and avoids drastic attitude adjustments.
[0051] Furthermore, every 30 minutes, nodes within the group upload their local model gradients to the control node. The control node then updates the global model using a weighted average of node data volume (weight = node data volume / total data volume) and distributes it to each node to improve collaborative accuracy.
[0052] Furthermore, it also includes: when multiple groups enter the same mission area, cross-group collaboration is carried out to generate airspace usage plans. Groups that enter the same task area are considered as co-domain groups; The control nodes corresponding to the same domain group are used as the same domain control nodes; A temporary Mesh network is established between control nodes in the same domain to share state data; The dispatch center calculates the airspace usage cost based on airspace occupancy time and area. The cost function is obtained by calculating the cost of airspace usage for control nodes in the same domain. Based on the cost function, the BPNE solver is used to generate the Pareto optimal solution as the spatial usage scheme for minimizing the cost function; Based on the airspace usage plan, the data is sent to the same domain group for execution.
[0053] Furthermore, the cost function J is: J=ω1·t delay +ω2·E consumed +ω3·C; Among them, t delayE represents the delay time in task execution, that is, the difference between the actual time spent executing the task and the expected or planned time. It reflects the delay caused by various factors such as airspace conflicts and obstacle avoidance. consumed This represents the energy consumed during the execution of a task, such as the electrical energy and fuel consumed during aircraft flight and equipment operation, reflecting the energy cost. C represents the airspace usage cost. ω1, ω2, and ω3 all represent weights. In this implementation, ω1, ω2, and ω3 take values of 0.5, 0.3, and 0.2, respectively.
[0054] Furthermore, the airspace usage cost C is: C = k1·t occupy +k2·A area ; Among them, t occupy Indicates the time the airspace is occupied, A area The area occupied by the airspace is represented by k1 and k2, which are both billing coefficients. In this embodiment, k1 is 0.2 yuan / second and k2 is 0.5 yuan / square meter, which are used to calculate the cost of occupying the airspace.
[0055] Furthermore, emergency task groups (such as rescue teams) are marked as priority level 5 and can occupy airspace free of charge. Other groups calculate the bidding amount based on the cost function J and are ranked to obtain the right of passage.
[0056] Furthermore, the control node collects the historical signal strength of each aircraft in the group and predicts the link stability for the next 10 seconds through an LSTM network. Aircraft with a link stability > 0.8 are selected as relay nodes. Link stability = 1 - predicted packet loss rate. When multiple groups enter the same airspace, a temporary mesh network is established between control nodes in the same airspace. The RPL routing protocol is used to achieve cross-group data forwarding, and the maximum number of hops is limited to 5 to control latency.
[0057] Furthermore, aircraft with a link stability greater than 0.8 are selected as relay nodes. Their core functions are: 1) forwarding cross-group state data to reduce packet loss rate over long distances; 2) ensuring the link reliability of the temporary Mesh network, especially in scenarios where multiple groups overlap in the airspace, to avoid communication interruption; 3) assisting control nodes in synchronizing global model gradients to improve federated learning efficiency.
[0058] Furthermore, such as Figure 5 As shown, the control node also generates dynamic weights based on obstacle information, and generates a second bypass scheme based on the dynamic weights to control the corresponding group execution.
[0059] Furthermore, the dynamic weight U total for: ; Among them, vobs d represents the relative velocity of the obstacle. static d represents the straight-line distance between the current group center and a static obstacle (such as a building or the boundary of a no-fly zone). dynamic It represents the real-time straight-line distance between the current group center and dynamic obstacles (such as flocks of birds or other aircraft).
[0060] The S6 aircraft acquires obstacle information in real time during operation and automatically avoids obstacles when communication is interrupted.
[0061] Furthermore, automatic obstacle avoidance specifically includes: The aircraft generates an obstacle probability distribution map based on multimodal sensors; When communication is interrupted, model prediction control is used for rolling optimization based on the obstacle probability distribution map to generate multiple candidate paths within the set prediction time domain; By introducing safety corridor constraints based on multiple candidate paths, the deviation between the candidate path and the original path is less than a preset value, thus obtaining the obstacle avoidance path; Automatic obstacle avoidance is achieved based on obstacle avoidance paths.
[0062] Furthermore, the aircraft uses its onboard multi-modal sensors and millimeter-wave radar to detect dynamic obstacles and binocular vision to identify static obstacles, generating an obstacle probability distribution map. It also receives group commands through a dual-band communication module (Sub-6GHz long-range command and millimeter-wave high-precision synchronization).
[0063] Furthermore, in normal mode, the aircraft receives MADDPG commands sent by the control node, converts them into motor signals for execution via the MAVLink protocol, and supports heterogeneous systems compatible with Pixhawk4, DJI Manifold2, and SDK.
[0064] Furthermore, if the aircraft's communication health PLR is greater than 50% for 3 seconds, it is determined to be a communication interruption, and the PPO model is switched to form an obstacle avoidance path based on the local obstacle probability distribution map.
[0065] Furthermore, the aircraft monitors its own battery level in real time, and when its battery level is lower than the warning threshold, it acts as a low-battery aircraft and sends out a distress signal. After a nearby aircraft responds to a distress signal, the corresponding priority is calculated based on the current distance between the aircraft and the low-battery aircraft and the current battery level. Select the nearest aircraft with the highest priority and current battery level greater than the set value as the power supply aircraft; Power is supplied to low-battery aircraft based on the power supply aircraft.
[0066] Furthermore, when the aircraft's own battery level is less than 20%, it is marked as a low-battery aircraft and broadcasts a distress signal via the LoRa link. After neighboring aircraft respond to the distress signal, they compete based on the improved Contract Net Protocol to calculate the corresponding priority G. G = 0.7 × Ds + 0.3 × L; Where Ds represents the current battery level of the nearby aircraft responding to the distress signal, and L represents the distance between the nearby aircraft responding to the distress signal and the low battery aircraft. Select the nearest aircraft with the highest priority (G) and remaining power >70% to transfer the backup battery via a robotic arm.
[0067] Furthermore, each aircraft is equipped with LoRa+WiFi6 dual backup links. When the primary link (Sub6GHz) is interrupted, i.e. the communication health is 100% for 2 seconds, it automatically switches to the backup link, the maximum speed decreases by 40%, and the coverage radius is increased to 8km, ensuring that critical data (obstacle avoidance commands) are not lost.
[0068] Furthermore, the aircraft feeds back its real-time status (unified data frame) to the control node, which then aggregates and uploads it to the cloud, forming a closed loop. Federated learning updates the model every 30 minutes to improve accuracy, and this cycle utilizes data from all stages to solve overall technical problems.
[0069] Example 2 Application Scenario 1: Logistics and distribution in densely populated urban areas (1) Task allocation: The dispatch center clusters delivery orders by destination and divides them into areas A and B based on the Voronoi diagram. Area A has an area of 5 km² and a priority level of 3, while area B has an area of 4 km² and a priority level of 2. Group 1 (80% remaining power) and Group 2 (75% remaining power) obtain task permissions through bidding, and the dispatch center issues instructions containing task unit boundaries and priorities.
[0070] (2) Path planning: Group 1 adopts a diamond formation with a spacing of 10m. Each aircraft shares position and velocity data through the MADDPG algorithm to generate a global path with a total distance of 12km. Dynamic adjustment: When encountering a sudden no-fly zone, i.e., airspace overlap = 25%, the control node calculates the obstacle avoidance potential field (dynamic weight U) using the potential field method. total =6.2>5), triggering replanning, extending the path to 14km but reducing energy consumption by 18%.
[0071] (3) Precise delivery: The aircraft uses binocular vision to identify AprilTag tags, and the robotic arm executes the delivery according to MAVLink instructions. The delivery success signal is synchronized to the cloud via the self-organizing network and the Task_Status field is updated.
[0072] Application Scenario 2: Simulating Multi-Group Airspace Conflict Simulation parameter input conditions: Inspection group: 10 aircraft, speed 8m / s, mission radius 2km; Logistics group: 15 aircraft, speed 12m / s, mission radius 3km; Airspace overlap area: 1.2 km², estimated duration of conflict: 120 seconds.
[0073] Game outcome: The dispatch center calculates the logistics group cost function J = 0.5 × 120 (delay) + 0.3 × 500 (energy consumption) + 0.2 × 320 (airspace cost) = 274; The number of patrol groups J = 0.5 × 130 + 0.3 × 450 + 0.2 × 280 = 256; The logistics group paid 320 yuan to obtain priority passage. The inspection team was repositioned to detour at a height of 50-100m, which increased the mission delay by 8%, but reduced the overall cost by 15%. The scheduling center generates a conflict resolution certificate (hash value: 0x7a3e...d41f) and writes it to the Ethereum testnet to ensure immutability.
[0074] Example 3 Verification of the technical effects of this invention: Simulation platform: Software: Gazebo + ROS2 + Python (Reinforcement learning library RLlib); Hardware: NVIDIA Jetson AGX Xavier (edge node), Intel Xeon server (cloud).
[0075] The performance indicators are shown in Table 2: Table 2 Comparison of Performance Indicators
[0076] The formula for calculating the improvement is: (Indicator value of this invention - Indicator value of traditional method) / Indicator value of traditional method × 100%, where the number of collisions per thousand aircraft is a negative indicator, and the calculation logic is (number of collisions per thousand aircraft - number of collisions per thousand aircraft) / number of collisions per thousand aircraft × 100%.
[0077] Real-world flight test (a smart logistics park): Test scale: 5 groups, a total of 80 heterogeneous drones; Results: After 24 hours of continuous operation, the task interruption rate was less than 0.5%, the average delivery error was 0.3m, and the communication handover success rate was significantly improved.
[0078] The method of this invention significantly improves the speed of resolving conflicts in multi-group tasks and the efficiency of airspace utilization. The hierarchical obstacle avoidance strategy reduces the probability of collisions, the autonomous operation time under communication interruption is greatly increased, and it supports protocol conversion for multiple types of heterogeneous aircraft.
[0079] Example 4 like Figure 6 As shown, based on the same inventive concept, the present invention also provides a cooperative control system for low-altitude aircraft, including: an information acquisition module, an area division module, a group control module, an emergency plan module, a cooperative control module, and an individual obstacle avoidance module; The information acquisition module is used to acquire group status information corresponding to multiple control nodes based on the scheduling center; The region partitioning module is used to divide multiple task regions based on group status information; The group control module is used to generate the optimal control scheme based on the task area and group status information to control the corresponding group operation; The emergency response module is used to generate an initial detour plan based on real-time monitoring of emergency situations during the group's work process by the dispatch center, and to control the group's execution. The collaborative control module is used to control nodes to generate collaborative commands based on state space data and control the collaborative operation of aircraft within the group; The single obstacle avoidance module is used by the aircraft to acquire obstacle information in real time during operation and automatically avoid obstacles when communication is interrupted.
[0080] Furthermore, in this embodiment, the functional implementation methods of each functional module correspond one-to-one with the methods described above, and will not be repeated here.
[0081] Example 5 Based on the same inventive concept, the present invention also provides an electronic device, which includes a processor and a memory, wherein the memory stores instructions, characterized in that the instructions are loaded and executed by the processor to implement a cooperative control method for a low-altitude aircraft as described in Embodiment 1.
[0082] Based on the same inventive concept, the present invention also provides a computer device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When the processor executes a program stored in the memory, it can implement a cooperative control method for a low-altitude aircraft as shown in Example 1.
[0083] The electronic device may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, communications interface, and memory communicate with each other via the communication bus. The processor can call logical instructions in the memory to execute a cooperative control method for low-altitude aircraft as described in Embodiment 1.
[0084] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a 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 a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0085] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0086] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A cooperative control method for low-altitude aircraft, characterized in that, include: The scheduling center obtains the group status information corresponding to multiple control nodes; Multiple task regions are obtained based on the aforementioned group state information; Based on the task area and the group state information, an optimal control scheme is generated to control the corresponding group operation; Based on the real-time monitoring of emergency situations during the group's operation by the dispatch center, a first detour plan is generated, and the group is controlled to execute it; The control node generates collaborative instructions based on state space data and controls the aircraft within the group to work collaboratively. The aircraft acquires obstacle information in real time during operation and automatically avoids obstacles when communication is interrupted.
2. The cooperative control method for low-altitude aircraft according to claim 1, characterized in that, Multiple task regions were obtained, specifically including: The group status information includes: aircraft position, aircraft speed, aircraft remaining battery power, aircraft mission status, and aircraft payload capacity; Based on the aircraft's position, speed, and meteorological parameters, a dynamic airspace partitioning algorithm is used to divide the low-altitude airspace into multiple overlapping mission areas. Based on the task area, associated task type, priority weight, and meteorological risk coefficient; The priority weight determines the scheduling order, and the meteorological risk coefficient is used for path correction.
3. The cooperative control method for low-altitude aircraft according to claim 1, characterized in that, The method for determining the control node is as follows: Based on the current battery level of each aircraft in the group and its average distance from other aircraft; The power coefficient is obtained based on the ratio of the current power level to the maximum power level of the single-unit aircraft; The distance coefficient is based on the ratio of the average distance to the maximum distance; A comprehensive coefficient is obtained based on the power coefficient and the distance coefficient; The aircraft with the largest comprehensive coefficient in the group is selected as the control node.
4. The cooperative control method for low-altitude aircraft according to claim 2, characterized in that, Generating the optimal control scheme specifically includes: Based on the matching degree between the aircraft's payload capacity and the aircraft's remaining power acquisition capability; The task matching degree is obtained based on the capability matching degree, the priority weight, and the weather adaptability. The scheme with the highest task matching degree is selected as the optimal control scheme.
5. The cooperative control method for low-altitude aircraft according to claim 2, characterized in that, The first detour plan is generated, which includes: The spatial overlap degree is calculated as the ratio of the overlapping area to the total area of the task area. The ratio of the number of lost packets to the total number of data packets in the group's uploaded data is used as the communication health score. When the spatial overlap is greater than a first threshold or the communication health is greater than a second threshold, the arbitration process is triggered: Predict potential conflict trajectories based on the aircraft position and aircraft speed of each aircraft in the swarm; Based on the potential conflict trajectory, the first detour plan is generated using the improved A* algorithm and sent to the group.
6. The cooperative control method for low-altitude aircraft according to claim 2, characterized in that, The generation of cooperative instructions specifically includes: The state space data is acquired based on the aircraft; Each of the aforementioned aircraft shares the state space data via the millimeter-wave frequency band; The control node inputs the state space data into the distributed MADDPG framework. Each of the aircraft's Actor networks receives the state space data and generates preliminary action commands; The shared Critic network aggregates the group's global state assessment Q value, feeds back the optimized initial action instructions, and generates the collaborative instructions; The state space data includes: the aircraft's own state information, neighbor's state information, and environmental state information.
7. The cooperative control method for low-altitude aircraft according to claim 2, characterized in that, Also includes: When multiple groups enter the same task area, cross-group collaboration is performed to generate an airspace usage plan: Groups that enter the same task area are considered as co-domain groups; The control nodes corresponding to the same domain group are used as the same domain control nodes; A temporary Mesh network is established between the control nodes in the same domain to share status data; The dispatch center calculates the airspace usage cost based on airspace occupancy time and airspace occupancy area. The co-domain control node calculates the cost function based on the airspace usage cost; Based on the cost function, a Pareto optimal solution is generated using a BPNE solver as the spatial usage scheme that minimizes the cost function; Based on the aforementioned airspace usage scheme, the data is sent to the corresponding domain groups for execution. The control node also generates dynamic weights based on obstacle information, and generates a second bypass scheme based on the dynamic weights to control the corresponding group execution.
8. The cooperative control method for low-altitude aircraft according to claim 1, characterized in that, The automatic obstacle avoidance specifically includes: The aircraft generates an obstacle probability distribution map based on multimodal sensors; When communication is interrupted, model prediction control is used for rolling optimization based on the obstacle probability distribution map to generate multiple candidate paths within a set prediction time domain; Based on the multiple candidate paths, a safety corridor constraint is introduced to ensure that the deviation between the candidate path and the original path is less than a preset value, thereby obtaining an obstacle avoidance path; Automatic obstacle avoidance is achieved based on the described obstacle avoidance path.
9. A cooperative control method for low-altitude aircraft according to claim 2, characterized in that, The aircraft monitors its own battery level in real time. When its own battery level is less than the warning threshold, it will act as a low-battery aircraft and send out a distress signal. After a nearby aircraft responds to the distress signal, it calculates the corresponding priority based on the current distance between itself and the low-battery aircraft and its current battery level. Select the nearest aircraft with the highest priority and current battery level greater than the set value as the power supply aircraft; The low-battery aircraft is powered by the aforementioned power supply aircraft.
10. A cooperative control system for a low-altitude aircraft, used to execute a cooperative control method for a low-altitude aircraft as described in any one of claims 1-9, characterized in that, include: The module includes an information acquisition module, a region division module, a group control module, an emergency response module, a collaborative control module, and an individual obstacle avoidance module. The information acquisition module is used to acquire group status information corresponding to multiple control nodes based on the scheduling center; The region division module is used to divide multiple task regions based on the group state information; The group control module is used to generate an optimal control scheme based on the task area and the group status information to control the corresponding group operation; The emergency response module is used to generate a first detour plan based on the real-time monitoring of emergency situations during the group's operation by the dispatch center, and to control the group to execute the plan. The collaborative control module is used by the control node to generate collaborative instructions based on state space data and control the aircraft in the group to work collaboratively. The single obstacle avoidance module is used by the aircraft to acquire obstacle information in real time during operation and automatically avoid obstacles when communication is interrupted.