A method and system for dynamic task allocation and collaborative control of drone clusters in disaster areas

Abstract

The present invention proposes a method and system for dynamic task allocation and collaborative control of a swarm of drones in disaster areas, including: Step S1: Real-time collection of raw data of disaster area scenes through ground-based IoT devices; Step S2: Dynamically marking task priorities on the raw data and integrating the multi-dimensional information of the drones; Step S3: Assigning tasks to the drones based on a hybrid allocation strategy and generating initial obstacle avoidance paths; Step S4: When the drones perform tasks, they update their flight paths in real time and avoid temporary obstacles, coordinate path conflicts through a conflict detection algorithm, and reallocate or downgrade tasks when necessary; Step S5: Periodically and globally adjust task priorities or correct paths; Step S6: Build an adaptive closed-loop control mechanism to track the multi-dimensional information and task attributes of the drones in real time. While ensuring a response time of seconds for high-priority tasks, the present invention maximizes global resource utilization, ultimately achieving the robustness and adaptability of drone swarms in complex scenarios.

Description

Technical Field

[0001] The present invention belongs to the field of wireless communication technology, and in particular relates to a method and system for dynamic task allocation and collaborative control of a cluster of unmanned aerial vehicles (UAVs) in a disaster area. Background Art

[0002] In recent years, with the rapid development of drone technology, its application in disaster relief has become increasingly widespread. Drones, with their high maneuverability, rapid deployment capabilities, and independence from ground infrastructure, have become a core tool for life-saving, material delivery, and environmental monitoring in disaster areas. In particular, in disaster scenarios like earthquakes and floods, drones can quickly access areas with disrupted communications or traffic, collaborating with ground-based Internet of Things (IoT) devices to collect disaster data in real time, locate trapped individuals, and deliver emergency supplies, significantly improving rescue efficiency.

[0003] However, existing drone disaster area task allocation technologies still face the following key challenges:

[0004] 1. Insufficient static priority processing: Traditional methods rely on fixed task priorities, making it difficult to dynamically respond to sudden high-priority demands in disaster areas (such as new life-saving tasks), resulting in delayed allocation of critical tasks.

[0005] 2. Inefficient multi-drone coordination: A single cost function or auction mechanism is prone to falling into local optimality, lacks global resource coordination capabilities, and leads to frequent conflicts among multiple drone missions, limiting resource utilization.

[0006] 3. Limited Path Planning Methods: Disaster-affected environments are complex and constantly changing, with temporary obstacles and communication blind spots potentially emerging at any time. Existing path planning algorithms are mostly based on static maps, lacking real-time obstacle avoidance and flight energy optimization capabilities. The dynamic nature of disaster-affected scenarios makes it difficult for existing path planning methods to meet timeliness requirements.

[0007] 4. Weak real-time adaptability: During disaster response missions, mission locations and requirements often change dynamically (for example, due to the movement of trapped personnel or the addition of new temporary supply demand points). Traditional dynamic programming methods typically require recalculating the entire global solution, which is inefficient and difficult to meet the real-time requirements of disaster response.

[0008] To address these issues, existing methods (such as heuristic rules and continuous convex approximations) rely on convexity assumptions or closed-form expressions, making it difficult to effectively handle the non-convexity, dynamic nature, and conflicting multi-objective nature of disaster-affected task allocation. An efficient and robust dynamic task allocation framework is urgently needed to support real-time decision-making and resource coordination in complex disaster scenarios. Summary of the Invention

[0009] The purpose of this invention is to provide an efficient and robust system for dynamic drone task allocation and collaborative control for disaster relief operations. This system aims to address the timeliness, flexibility, and collaborative efficiency issues inherent in existing drone task allocation methods. Specifically, this invention employs an innovative task allocation strategy to ensure rapid response to high-priority tasks within the complex environment of a disaster area, optimizing resource utilization and allocation efficiency, thereby enhancing the overall performance of drones in disaster relief operations.

[0010] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0011] A method for dynamic task allocation and collaborative control of a swarm of UAVs in a disaster area, comprising the following steps:

[0012] Step S1: collecting raw data of the disaster area scene in real time through ground IoT devices; and preliminarily classifying the raw data;

[0013] Step S2: dynamically marking the task priority of the raw data according to the urgency and rescue value of the task, and integrating the multi-dimensional information of the drone to form a global decision-making data center;

[0014] Step S3: calling the hybrid allocation strategy based on the task priority, assigning tasks to the UAVs based on the hybrid allocation strategy and generating an initial obstacle avoidance path;

[0015] Step S4: When the UAV performs the mission, based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded;

[0016] Step S5: Periodically and dynamically monitor the global decision-making data center, and dynamically adjust task priorities or correct paths globally;

[0017] Step S6: Build an adaptive closed-loop control mechanism to track the multi-dimensional information and mission attributes of the UAV in real time to form a closed-loop control.

[0018] Furthermore, the original data in step S1 includes: the location of trapped persons, material demand distribution, and obstacle information.

[0019] Furthermore, the dynamic task priority in step S2 includes:

[0020] The highest priority, P-3: emergency rescue missions requiring a response within seconds;

[0021] Medium priority P-2: Critical support tasks requiring a minute-level response;

[0022] Low Priority P-1: General support tasks with hourly response times.

[0023] Furthermore, the multi-dimensional information of the drone in step S2 includes:

[0024] Drone status: location coordinates, remaining battery power, and mission queue;

[0025] Environmental parameters: obstacle coordinates, communication blind area range.

[0026] Furthermore, calling a hybrid allocation strategy based on task priority includes:

[0027] For P-3 missions, an allocation method combining an auction mechanism with game theory collaborative optimization is used. By verifying the UAV's time constraints, load constraints, and mission quantity constraints, the highest bid is calculated and the mission is assigned to the optimal UAV.

[0028] For P-2 tasks, a task sequence is generated based on the Monte Carlo tree search and greedy algorithm fusion strategy, and task allocation is optimized by constructing a search tree, simulating benefit evaluation, and backpropagation.

[0029] For P-1 tasks, they are arranged in descending order of benefit density. When energy consumption exceeds the limit, they are split into subtasks and executed by multiple machines in relay.

[0030] Furthermore, the conflict detection algorithm in step S4 is a space-time cube conflict detection algorithm, which includes: mapping the UAV path to a four-dimensional space-time coordinate system, calculating the space-time cube conflict risk, and triggering game theory collaborative optimization when the risk value exceeds a threshold;

[0031] The space-time cube conflict risk is:

[0032] in, is the maximum value, is the minimum value, is the risk value of space-time cube conflict risk, , the closer to the upper limit, the higher the risk, and when it is equal to 1, it is a direct conflict; is the conflict penalty coefficient, For drones At the moment The location coordinates of For drones At the moment The location coordinates of For a safe distance, The start time is , the end time is time period.

[0033] Furthermore, the local reallocation or priority degradation in step S4 includes:

[0034] The conflicting missions are re-added to the auction pool and re-bid based on the updated drone parameters;

[0035] If the reallocation fails, the task priority is forcibly downgraded to release resources for other high-priority tasks.

[0036] Furthermore, the global dynamic adjustment of task priorities or correction of paths in step S5 includes:

[0037] When a task priority upgrade is detected, low-priority tasks are interrupted and resources are reallocated;

[0038] When the task coordinate offset exceeds the threshold, task continuity is ensured through path grafting or subtask splitting;

[0039] When a drone anomaly is detected, the resource recovery protocol is triggered to screen replacement drones and correct conflicting paths.

[0040] Furthermore, the resource recovery protocol includes:

[0041] Calculate the handover time between a failed drone and its replacement:

[0042] If there is a conflict in the handover path, a feasible solution is generated by locally correcting the route or downgrading the task priority;

[0043] in, For the handover time, is the distance between the former drone and its successor, For transfer speed, The data transfer rate for the task.

[0044] On the other hand, the present invention also provides a disaster area UAV cluster dynamic task allocation and collaborative control system, comprising:

[0045] A data acquisition module is used to collect raw data of disaster-stricken areas in real time through ground IoT devices; the raw data is preliminarily classified through a local pre-processing module and transmitted to a collaborative decision-making database;

[0046] A data integration module is used to dynamically prioritize the raw data based on the urgency and rescue value of the mission, and integrate the multi-dimensional information of the drone to form a global decision-making data hub;

[0047] A hybrid task allocation module is used to call a hybrid allocation strategy based on task priority, assign tasks to UAVs based on the hybrid allocation strategy, and generate an initial obstacle avoidance path. The module includes: a hybrid high-priority task allocation submodule, a medium- and low-priority task hybrid algorithm allocation submodule, and a dynamic path planning module.

[0048] Conflict detection and collaborative optimization module: It is used when the UAV performs tasks based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded;

[0049] Global monitoring module: This module is used to periodically and dynamically monitor the global decision-making data center, dynamically adjust task priorities or correct paths globally, and includes an exception handling submodule;

[0050] Closed-loop control module: It is used to build an adaptive closed-loop control mechanism, track the multi-dimensional information and mission attributes of the drone in real time, and form closed-loop control.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. Compared with the traditional static priority allocation mechanism, the present invention can respond to sudden high-priority task demands in real time through dynamic priority adjustment and hierarchical task scheduling strategy, significantly improving the allocation efficiency and execution success rate of emergency tasks (such as life-saving).

[0053] 2. Compared with collaborative control methods that rely on a single algorithm, this invention integrates an auction mechanism, Monte Carlo Tree Search (MCTS), and a greedy algorithm. While ensuring a response time of seconds for high-priority tasks, it also takes into account the global optimization and real-time performance of medium- and low-priority tasks, effectively avoiding local optimal traps.

[0054] 3. Compared with the path planning technology based on static maps, the present invention adopts The algorithm dynamically updates the obstacle avoidance path and, combined with space-time cube conflict detection technology, can adapt to temporary obstacles (such as collapsed buildings and moving vehicles) and environmental changes in real time, significantly improving flight safety and path optimality in complex scenarios.

[0055] 4. Compared with the single design of traditional energy consumption models, this invention maximizes the endurance and resource utilization of drone clusters while ensuring the reliability of task execution through layered energy consumption constraints (a detailed model for high-priority tasks and a simplified model for medium and low-priority tasks) and a dynamic resource recovery mechanism.

[0056] 5. Compared with systems that lack standardized exception handling, the present invention supports priority preemption, task grafting, subtask splitting and multi-UAV relay execution through a dynamic monitoring module and an adaptive closed-loop control strategy, ensuring task continuity and system robustness in abnormal scenarios such as communication interruption and hardware failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0058] Figure 1 This is an overall flow chart of an embodiment of the present invention;

[0059] Figure 2 This is a flow chart of the Monte Carlo Tree Search (MCTS) and greedy algorithm fusion strategy in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0061] Example 1

[0062] like Figure 1 As shown, this embodiment provides a method for dynamic task allocation and collaborative control of a disaster area drone cluster, including the following steps:

[0063] Step S1: collecting raw data of the disaster area scene in real time through ground IoT devices; and preliminarily classifying the raw data;

[0064] In this embodiment, the disaster area scenario parameters (disaster area range: D × D km, including multiple collapsed buildings, moving obstacles and communication blind spots); drone cluster size: 20 aircraft, equipped with multi-functional modules such as material transportation, life detection, and communication relay; ground IoT devices: 50 vital sign sensors, 30 environmental monitors, and 20 emergency material demand terminals are deployed.

[0065] In this embodiment, collecting raw data includes:

[0066] 1. Load the disaster area map: Import the coordinates of collapsed buildings , a digital map of the communication blind spot boundary (radius 200m).

[0067] 2. Initialize the database: enter drone parameters and ground IoT device data.

[0068] 3. Start data collection: The sensor uploads dynamic information of obstacles in real time.

[0069] Ground IoT devices collect raw data from the disaster area in real time (including the location of trapped people, distribution of material needs, obstacle information, etc.) and perform preliminary classification of the raw data.

[0070] Step S2: Dynamically mark the task priority of the raw data according to the urgency and rescue value of the task, and integrate the multi-dimensional information of the drone to form a global decision-making data center; as shown in Table 1:

[0071] Table 1: Global decision-making data hub

[0072]

[0073] Based on the urgency of the task, resource requirements and the degree of direct impact on disaster relief, the task priorities of the present invention are divided into three levels (P-3, P-2, P-1). P-3 (highest priority) P-2 (medium priority) is a mission that is directly related to the safety of personnel or the overall progress of the rescue and requires a response within seconds, such as precise positioning of trapped personnel, emergency delivery of medical supplies, and restoration of key communication nodes. : Critical support tasks) ensure the continuity of rescue and the restoration of basic functions in the disaster area, requiring minute-level response, such as non-emergency material delivery, local environmental monitoring (investigation of risk factors for secondary disasters), infrastructure status inspection, etc.; P-1 (lowest priority) : Conventional support tasks) serve medium- and long-term rescue planning or global resource optimization, allowing hourly response, such as large-scale environmental data collection (air quality monitoring), long-term communication signal coverage expansion, etc.

[0074] After the priority marking is completed, all task attributes (including priority, execution progress) and the real-time status of the drone (position coordinates) are , remaining power , Task Queue ) and environmental parameters (obstacle coordinates, communication blind spot range) and other multi-dimensional information integration to form a global decision-making data center.

[0075] Dynamic marking and priority division of tasks in this embodiment

[0076] P-3 mission mark: 3 life signals detected, marked as the highest priority, requiring a response within seconds.

[0077] P-2 mission marking: The environmental monitoring instrument detected two aftershock risk areas, marked as medium priority, requiring a minute-level response.

[0078] P-1 Task Mark: 10 air quality monitoring needs, marked as the lowest priority.

[0079] Step S3: calling the hybrid allocation strategy based on the task priority, assigning tasks to the UAVs based on the hybrid allocation strategy and generating an initial obstacle avoidance path;

[0080] After the global decision data center is initialized, the high-priority task real-time response module is immediately started. The auction mechanism first initiates bidding for the 𝑃-3 task. First, the drone must meet the following conditions to participate in the P-3 task auction:

[0081]

[0082] in, for Algorithm generates data from drones Highest priority task The initial obstacle avoidance path distance; For drones Flight speed; For the task Estimated execution time; For drones Remaining power; For the dynamic flight power consumption of UAV, a semi-empirical flight power model can be used ; For drones Current load, For the task Required load, For drones The maximum load that can be carried; For drones Currently assigned tasks, Ensure that each drone can only perform one P-3 mission at the same time.

[0083] The drone bids based on multi-dimensional parameters, and the bidding function is designed as follows:

[0084]

[0085] in, It's a drone Complete the task The revenue can be defined in different revenue models according to the mission type and goal. If it is a rescue mission, the revenue can be defined as the number of people rescued: , It's a task Number of people waiting to be rescued, It's a drone If it is a material distribution task, the benefit can be defined as the weighted value of the urgency and quantity of the materials: , It's a task The quantity of supplies, is the urgency value coefficient of the materials; if it is a communication relay task, the benefit can be defined as the area of ​​the disaster area covered: , It's a task The area that needs to be covered, It's a drone Communication radius coverage (such as coverage efficiency per unit area).

[0086] drones The current load (kg) of the drone is used. The heavier the load, the worse the maneuverability of the drone. The penalty term is used to suppress the bidding of high-load drones.

[0087] Indicates drone Execute the task The resource consumption required usually includes time cost and energy cost. The total time for a drone to perform a mission includes flight time and mission execution time: , where flight time , To pass Algorithm-generated drones To the task Initial obstacle avoidance path distance (m), For drones Flight speed (m / s); Mission execution time , indicating the task The total energy consumption of the UAV adopts a detailed energy consumption model, including flight energy consumption and mission execution energy consumption: , of which flight energy consumption , flight power Using the empirical formula, it is defined as an expression proportional to the square of the velocity: ;Task execution energy consumption , The time and energy cost are normalized and then weighted summed: , is the drone's flight time, is the drone battery capacity, is the weight coefficient of time and energy (satisfying ).

[0088] is the global weight coefficient of revenue, load and cost, and the weight coefficient satisfies , ensuring the normalization of utility values.

[0089] After the drone with the highest bid obtains the mission, it generates an initial obstacle avoidance path and locks the mission.

[0090] In this embodiment, high priority tasks (P-3) are allocated through auction.

[0091] Constraint Verification: Detecting Drones Whether the time constraints, load constraints and task quantity constraints are met.

[0092] Bidding and allocation: Bidding and allocation are carried out for drones that meet the constraints and are eligible to bid, and the task is won by the drone with the highest bid. , generate the initial obstacle avoidance path and lock the task.

[0093] Execute the mission and update the path in real time.

[0094] For the dynamic allocation of medium and low priority tasks (P-2 / P-1), this paper adopts the fusion strategy of MCTS (Monte Carlo TreeSearch, an algorithm for optimal selection in the decision-making process) and greedy algorithm (see Appendix). Figure 2 (Figure 2 is a flowchart of this fusion strategy). While strictly adhering to priority rules (P-2 > P-1) and energy constraints, the system significantly improves the efficiency and real-time performance of task sequence generation. The system first retrieves all pending P-2 and P-1 tasks from the collaborative decision-making database and, based on the remaining UAV battery, payload capacity, and environmental parameters, initiates the task allocation engine.

[0095] For P-2 tasks, the MCTS algorithm is used to generate an approximately optimal task sequence. MCTS dynamically balances exploration and exploitation by constructing a search tree with task states as nodes. Each tree node represents the current allocation state and contains the set of selected tasks. , remaining power , cumulative income , and the number of visits The root node is initialized to an empty task set. , remaining power , represents the current remaining power of the drone 𝑖. Starting from the root node, child nodes are selected based on the upper confidence bound (UCB) formula: .in, The default weight is 1.5, which gives priority to branches with high returns or those that have not been fully explored. The total benefit value accumulated by the current node during the search process, including the benefits contributed by all paths passing through the node, Record the total number of times the current node has been visited, The total number of times the parent node of the current node has been visited. If an incompletely expanded node is selected, a random unassigned task is selected. , is the set of tasks that are not currently assigned to drones, i.e., the remaining assignable tasks. Check its energy consumption constraint: , energy consumption constraints Using a simplified energy consumption model, the load-related energy consumption is calculated as ( ) and flight-related energy consumption ( ), where For the task Required load weight (unit: kg); For the task The average energy consumption rate (unit: mAh / kg), which indicates the energy consumption per kilogram of load; For the drone to go from the current position to the mission Flight distance (unit: km); For drones Flight speed (unit: km / h); It is the dynamic power consumption of the drone during flight (unit: mAh / h), which is related to the flight speed.

[0096] like , then expand the new node and update 、 . Perform a fast random simulation on the expanded nodes and select the remaining tasks according to the greedy strategy (preferring the reward density High mission, mission value ,in For drones The maximum load that can be carried, for drones The maximum allowable flight distance; mission execution energy consumption , energy consumption and The formula is consistent) fills the sequence until the battery is exhausted, generating temporary income . The simulated income Back propagate to all nodes on the path and update the cumulative benefits of each node and number of visits After a preset time threshold (e.g., 200ms) or number of iterations (e.g., 1000 times), the branch with the most visits under the root node is selected as the final P-2 task sequence.

[0097] For P-1 tasks, the system uses a greedy algorithm to ensure real-time performance. Sort in descending order, select The highest task is added to the queue until the remaining power is insufficient. If a task cannot be executed due to excessive energy consumption, it is split into standardized subtasks: based on the remaining power Total energy consumption of the task The execution scope is allocated proportionally (e.g. when the remaining power supports 60% energy consumption, the subtask covers 60% of the area or performs 60% of the data collection), and marked as "partially completed". The remaining part will be handled by the subsequent drones. When the number of P-1 tasks exceeds the dynamic threshold (e.g. 1.5 times the size of the drone cluster) or the remaining decision time is insufficient (adaptive calculation is ,in is the decision time threshold, is the number of real-time tasks), the system automatically switches from MCTS to the greedy algorithm; at the same time, a pruning rule is introduced. If the cumulative benefit of the current path is less than 90% of the historical optimal value or the remaining power is less than 20%, the branch expansion is immediately terminated. All generated task sequences need to call the conflict detection algorithm to verify whether they overlap with the space-time cube of the high-priority task path. Conflict detection uses a dynamic safety distance ( = 2 seconds is the system response time, = 10 meters is the physical safety distance), if a conflict is detected (such as the space-time cube of the P-2 path and the P-3 rescue route The distance within the time period is less than ), the game theory collaborative optimization module is triggered to eliminate conflicts by adjusting the flight altitude, delaying the departure time or replanning the local path. The corrected path is synchronized to the collaborative decision database in real time to ensure the dynamic consistency of the global task sequence.

[0098] Step S4: When the UAV performs the mission, based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded;

[0099] During the mission execution phase, the drone uses The algorithm updates the flight path in real time. The algorithm uses onboard sensors to detect dynamic obstacles (such as moving vehicles and newly collapsed buildings) and updates the path distance every 5 seconds. , which is mathematically described as:

[0100]

[0101] pass Algorithm, based on time The current position of the drone ( ), target position ( ) and time Obstacle map ( ) to calculate the updated path.

[0102] After the updated path data is synchronized, the background continuously calculates the path risk between multiple drones. The conflict risk calculation uses space-time cube detection (global conflict prediction): the drone flight path is mapped to the four-dimensional space-time coordinate system , the path of each drone is represented as a space-time cube: , and They represent the start and end time of the drone path in the space-time cube. and The space-time cube in the time period [ Internal satisfaction: , For drones At the moment The location coordinates of For drones At the moment The location coordinates of For a safe distance.

[0103] When detected When all UAVs involved in the conflict submit their current path parameters (such as planned departure time and flight altitude) to the collaborative decision database, each UAV calculates its own optimal adjustment plan based on the current strategies of other UAVs. Each UAV acts as an independent decision-making subject, and its behavioral goal is defined as maximizing individual utility. , The task benefit item quantifies the immediate benefit of the task at the current time point. If high-priority tasks are completed as soon as possible, they can bring higher utility; if they are delayed ( Increase), even if the priority is high, the utility will be greatly reduced due to time decay. This encourages the system to prioritize high-priority and urgent tasks to maximize the total utility; Consider only UAVs as the conflict cost term The total risk of conflict with other drones, the risk of conflict Defined as a continuous risk value based on space-time cube detection: , risk value The closer the distance, the higher the risk. When it is equal to 1, it is a direct conflict. is the conflict penalty coefficient, balancing the mission benefits and risk costs. Using a collaborative optimization algorithm based on Nash equilibrium, each drone calculates its own optimal adjustment plan based on the current strategy of other drones, and uses the gradient descent method ( ) Iteratively update the drone strategy parameters , and resubmit it to the database. This process repeats until the convergence condition is met:

[0104] a) The strategy change of all drones is less than the preset threshold (for example, the path parameter adjustment is less than 1%);

[0105] b) The utility function value tends to be stable and there is no room for significant improvement.

[0106] If irreconcilable conflicts still exist after game optimization (such as insufficient power or task timeout due to path adjustment), the system triggers a local re-auction mechanism: 1) Task reallocation: Conflicting tasks rejoin the auction pool and re-bid according to updated parameters (such as remaining power and real-time location); 2) Priority preemption: High-priority tasks forcefully interrupt low-priority tasks to ensure that urgent needs are met first.

[0107] Step S5: Periodically and dynamically monitor the global decision-making data center, and dynamically adjust task priorities or correct paths globally;

[0108] S5 dynamically adjusts task priorities or corrects paths globally, including:

[0109] When a task priority upgrade is detected, low-priority tasks are interrupted and resources are reallocated;

[0110] When the task coordinate offset exceeds the threshold, task continuity is ensured through path grafting or subtask splitting;

[0111] When a drone anomaly is detected, the resource recovery protocol is triggered to screen replacement drones and correct conflicting paths.

[0112] When the system detects an abnormal event (such as drone hardware failure, communication interruption, or sudden environmental interference), the exception handling module immediately intervenes to maintain mission continuity. For example, if a drone triggers a fault self-check due to motor overheating, the system will mark it as offline and invoke the resource recovery protocol: calculate the distance between the current drone and the successor , combined with the transfer speed and task data transfer rate , ensuring that the total handover time meets the threshold requirements. If a path conflict is detected during the handover process, the system will locally correct the route or downgrade the task priority until a feasible solution is generated. All adjusted status information is synchronized to the central database in real time, providing input for the next round of iteration, forming a "monitoring-evaluation-optimization" closed-loop control. Through the deep coupling of multiple rounds of iteration and exception handling, the system can maximize global resource utilization while ensuring a response time of seconds for high-priority tasks, ultimately achieving the robustness and adaptability of drone swarms in complex scenarios.

[0113] In this embodiment, specifically:

[0114] .When there is a task priority upgrade, a hierarchical response strategy is initiated: P-3 task preemption and P-2 / P-1 task dynamic adjustment.

[0115] .When the task coordinates shift significantly, they will be reallocated through a quick auction mechanism.

[0116] .When the drone’s battery is running low, the mission handover system is triggered.

[0117] Step S6: Build an adaptive closed-loop control mechanism to track the multi-dimensional information and mission attributes of the UAV in real time to form a closed-loop control.

[0118] Finally, the system dynamically monitors the changes in task attributes and synchronizes with the real-time status to build an adaptive closed-loop control mechanism to ensure the efficient response and fault-tolerant operation of the drone cluster in the complex environment of the disaster area. The system continuously scans the collaborative decision-making database in a 5-second cycle to track changes in task priority, target coordinates, environmental parameters and drone status in real time. When it detects that the task priority has been upgraded (such as aftershocks triggering secondary disaster warnings, or the vital signs of trapped people deteriorating), or the coordinates of a task have been significantly offset due to target movement or environmental changes (such as three-dimensional Euclidean distance), the system will automatically detect the changes in the task priority, target coordinates, environmental parameters and drone status. , (Old coordinates) represents the original position coordinates of the task or target recorded before the transformation, including three-dimensional components , (New coordinates) indicates the latest position coordinates of the task or target after the environment changes or system updates, including three-dimensional components ), the system initiates a hierarchical response strategy: 1) P-3 task preemption: directly interrupt the current task of the low-priority UAV (such as forcing the P-1 task to suspend), release its resources (power, load capacity), and reallocate them through a fast auction mechanism, first screening candidate UAVs: meet the time constraints (Reserve 20% of the battery for emergency) and the drones with matching load will enter the auction pool. Then the principle of proximity priority will be adopted. Algorithm generates obstacle avoidance path distance , prioritize assigning the drone closest to the new target while constraining the handover time , ensuring seamless task connection. If there is no response within 10 seconds, the backup fleet will be activated. 2) Dynamic adjustment of P-2 / P-1 tasks: For non-urgent tasks, a "task grafting" strategy is adopted to merge the offset tasks with other tasks on the original path (such as adjusting the material delivery point A→B to A→C→B). The algorithm generates a detour route. If the route adjustment causes the total energy consumption to exceed the limit, the task is split into subtasks, and multiple drones collaborate to complete the task (for example, Drone A delivers 50% of the supplies to the intermediate node, and Drone B relays to complete the remaining transportation).

[0119] Example 2

[0120] This embodiment provides a dynamic task allocation and coordination control system for a swarm of drones in disaster areas, including:

[0121] A data acquisition module is used to collect raw data of disaster-stricken areas in real time through ground IoT devices; the raw data is preliminarily classified through a local pre-processing module and transmitted to a collaborative decision-making database;

[0122] A data integration module is used to dynamically prioritize the raw data based on the urgency and rescue value of the mission, and integrate the multi-dimensional information of the drone to form a global decision-making data hub;

[0123] A hybrid task allocation module is used to call a hybrid allocation strategy based on task priority, assign tasks to UAVs based on the hybrid allocation strategy, and generate an initial obstacle avoidance path. The module includes: a hybrid high-priority task allocation submodule, a medium- and low-priority task hybrid algorithm allocation submodule, and a dynamic path planning module.

[0124] Conflict detection and collaborative optimization module: It is used when the UAV performs tasks based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded;

[0125] Global monitoring module: This module is used to periodically and dynamically monitor the global decision-making data center, dynamically adjust task priorities or correct paths globally, and includes an exception handling submodule;

[0126] Closed-loop control module: It is used to build an adaptive closed-loop control mechanism, track the multi-dimensional information and mission attributes of the drone in real time, and form closed-loop control.

[0127] It should be understood that parts not elaborated in detail in this specification belong to the prior art.

[0128] It should be understood that the above description of the preferred embodiments is relatively detailed and should not be considered as limiting the scope of protection of the present invention. It is not necessary and impossible to list all embodiments here. Under the guidance of the present invention, those skilled in the art may make substitutions or modifications without departing from the scope of protection of the present invention, and all of these modifications fall within the scope of protection of the present invention.

Claims

1. A method for dynamic task allocation and collaborative control of a swarm of UAVs in disaster areas, characterized by: The following steps are involved: Step S1: collecting raw data of the disaster area scene in real time through ground IoT devices; and preliminarily classifying the raw data; Step S2: dynamically marking the task priority of the raw data according to the urgency and rescue value of the task, and integrating the multi-dimensional information of the drone to form a global decision-making data center; Step S3: calling the hybrid allocation strategy based on the task priority, assigning tasks to the UAVs based on the hybrid allocation strategy and generating an initial obstacle avoidance path; Step S4: When the UAV performs the mission, based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded; Step S5: Periodically and dynamically monitor the global decision-making data center, and dynamically adjust task priorities or correct paths globally; Step S6: Build an adaptive closed-loop control mechanism to track the multi-dimensional information and mission attributes of the UAV in real time to form a closed-loop control.

2. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 1 is characterized in that: The original data in step S1 includes: the location of the trapped persons, the distribution of material requirements, and obstacle information.

3. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 2, characterized in that: The dynamic task priority in step S2 includes: The highest priority, P-3: emergency rescue missions requiring a response within seconds; Medium priority P-2: Critical support tasks requiring a minute-level response; Low Priority P-1: General support tasks with hourly response times.

4. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 1, characterized in that: The multi-dimensional information of the drone in step S2 includes: Drone status: location coordinates, remaining battery power, and mission queue; Environmental parameters: obstacle coordinates, communication blind area range.

5. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 3 is characterized in that: The hybrid allocation strategies based on task priority include: For P-3 missions, an allocation method combining an auction mechanism with game theory collaborative optimization is used. By verifying the UAV's time constraints, load constraints, and mission quantity constraints, the highest bid is calculated and the mission is assigned to the optimal UAV. For P-2 tasks, a task sequence is generated based on the Monte Carlo tree search and greedy algorithm fusion strategy, and task allocation is optimized by constructing a search tree, simulating benefit evaluation, and backpropagation. For P-1 tasks, they are arranged in descending order of benefit density. When energy consumption exceeds the limit, they are split into subtasks and executed by multiple machines in relay.

6. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 1, characterized in that: The conflict detection algorithm in step S4 is a space-time cube conflict detection algorithm, which includes: mapping the UAV path to a four-dimensional space-time coordinate system, calculating the space-time cube conflict risk, and triggering game theory collaborative optimization when the risk value exceeds a threshold; The space-time cube conflict risk is: in, is the maximum value, is the minimum value, is the risk value of space-time cube conflict risk, , the closer to the upper limit, the higher the risk, and when it is equal to 1, it is a direct conflict; is the conflict penalty coefficient, For drones At the moment The location coordinates of For drones At the moment The location coordinates of For a safe distance, The start time is , the end time is time period.

7. The method for dynamic task allocation and collaborative control of a swarm of UAVs in a disaster area according to claim 5, characterized in that: The local reallocation or priority degradation in step S4 includes: The conflicting missions are re-added to the auction pool and re-bid based on the updated drone parameters; If the reallocation fails, the task priority is forcibly downgraded to release resources for other high-priority tasks.

8. The method for dynamic task allocation and collaborative control of a disaster area UAV cluster according to claim 1, characterized in that: The global dynamic adjustment of task priorities or correction of paths in step S5 includes: When a task priority upgrade is detected, low-priority tasks are interrupted and resources are reallocated; When the task coordinate offset exceeds the threshold, task continuity is ensured through path grafting or subtask splitting; When a drone anomaly is detected, the resource recovery protocol is triggered to screen replacement drones and correct conflicting paths.

9. The method for dynamic task allocation and collaborative control of a swarm of UAVs in a disaster area according to claim 8, characterized in that: The resource recovery agreement includes: Calculate the handover time between a failed drone and its replacement: If there is a conflict in the handover path, a feasible solution is generated by locally correcting the route or downgrading the task priority; in, For the handover time, is the distance between the former drone and its successor, For transfer speed, The data transfer rate for the task.

10. A dynamic task allocation and collaborative control system for drone clusters in disaster areas, characterized by: include: A data acquisition module is used to collect raw data of disaster-stricken areas in real time through ground IoT devices; the raw data is preliminarily classified through a local pre-processing module and transmitted to a collaborative decision-making database; A data integration module is used to dynamically prioritize the raw data based on the urgency and rescue value of the mission, and integrate the multi-dimensional information of the drone to form a global decision-making data hub; A hybrid task allocation module is used to call a hybrid allocation strategy based on task priority, assign tasks to UAVs based on the hybrid allocation strategy, and generate an initial obstacle avoidance path. The module includes: a hybrid high-priority task allocation submodule, a medium- and low-priority task hybrid algorithm allocation submodule, and a dynamic path planning module. Conflict detection and collaborative optimization module: It is used when the UAV performs tasks based on The algorithm updates the flight path in real time, avoids temporary obstacles in real time, and uses a conflict detection algorithm to detect whether multiple drone paths conflict. If a path conflict is detected, the conflict is resolved; if the conflict cannot be reconciled, the current task is partially reallocated or the priority is downgraded; Global monitoring module: This module is used to periodically and dynamically monitor the global decision-making data center, dynamically adjust task priorities or correct paths globally, and includes an exception handling submodule; Closed-loop control module: It is used to build an adaptive closed-loop control mechanism, tracking the multi-dimensional information and mission attributes of the UAV in real time to form a closed-loop control; The dynamic task allocation and collaborative control system for a disaster area drone cluster is used to execute the steps in a dynamic task allocation and collaborative control method for a disaster area drone cluster as described in any one of claims 1-9.

Images