Task allocation method based on dynamic dual modes
By using a dynamic dual-mode task allocation method, combined with a distributed auction algorithm and a centralized particle swarm algorithm, the real-time performance and reliability of UAV task allocation are achieved. This solves the problem of insufficient timeliness of centralized and distributed methods in dynamic environments, and improves the task execution efficiency and survivability of UAV swarms.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-03
- Publication Date
- 2026-04-07
AI Technical Summary
Existing centralized and distributed task allocation methods each have their advantages and disadvantages, making it difficult to simultaneously meet the requirements of optimal task allocation, real-time performance, and reliability, especially in dynamic battlefield environments where timeliness is insufficient.
A dynamic dual-mode task allocation method is adopted, which defines the roles of UAVs as swarm, group, and UAV levels. Combining auction and particle swarm algorithms, it realizes distributed and centralized negotiation and optimization of tasks, sets time thresholds and task priorities, and supports dynamic selection and task termination mechanisms.
It improves the real-time performance and reliability of task allocation, reduces system complexity, enhances the survivability and mission success rate of UAV swarms in highly competitive environments, and adapts to various dynamic scenarios.
Smart Images
Figure CN121806993A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] Embodiments of the present disclosure relate to the technical field of cluster UAV task allocation, and particularly relate to a task allocation method based on dynamic dual mode. BACKGROUND
[0002] With the rapid development of UAV technology, the application demand of multi-UAV system in complex task scenarios (such as battlefield reconnaissance, disaster rescue, logistics distribution, etc.) is growing. Task allocation, as the core link of multi-UAV cooperation, needs to consider multiple factors such as UAV performance, dynamic environment constraints, and task target priority to achieve the overall optimal efficiency of the system. The current mainstream solutions are mainly divided into centralized task allocation and distributed task allocation, which have significant differences in architecture design and each has advantages and limitations.
[0003] Centralized task allocation achieves global information integration and optimal decision through a central node, and its advantages are as follows: 1) strong global optimality, a unified optimization model can be constructed based on complete battlefield situation information to avoid local suboptimal solution; 2) high resource coordination efficiency, the central node can directly coordinate multi-UAV resource conflicts to reduce task overlap and redundancy; 3) outstanding complex constraint processing capability, supporting centralized calculation of high-dimensional nonlinear optimization algorithms. However, the centralized architecture has single point failure risk, dynamic response lag, and large-scale cluster communication bandwidth pressure, etc., which is difficult to adapt to high real-time and strong anti-destroying task requirements.
[0004] Distributed task allocation adopts a decentralized architecture and relies on autonomous negotiation and local decision among UAV nodes, and its advantages are as follows: 1) high system robustness, no central node dependency, partial node failure does not affect overall task execution; 2) strong dynamic environment adaptability, nodes can quickly adjust strategies based on local information to reduce communication dependence; 3) excellent scalability, when the cluster size expands, the computing and communication pressure can be alleviated through hierarchical negotiation. However, the distributed architecture is prone to global goal conflict caused by node autonomous decision, and has defects such as negotiation shock, slow convergence speed, etc., which is not time-efficient in sudden threat avoidance or emergency task re-allocation scenarios.
[0005] Existing research shows that a single architecture cannot meet the optimal, real-time and reliability requirements of task allocation. Centralized allocation is difficult to cope with dynamic battlefield environment due to the contradiction between global optimization and local response capability; distributed allocation is limited by local information island and negotiation efficiency bottleneck although it can improve system flexibility.
[0006] Therefore, it is necessary to improve one or more problems in the related technical solutions described above.
[0007] It should be noted that this section is intended to provide a background or context to the technical solution of the disclosure stated in the claims. The description herein is not admitted to be prior art merely because it is included in this section. SUMMARY
[0008] The purpose of the embodiments of the present disclosure is to provide a task allocation method based on dynamic dual mode, thereby at least partially overcoming one or more problems caused by the limitations and defects of the related art.
[0009] According to the embodiments of the present disclosure, a task allocation method based on dynamic dual mode is provided, comprising: defining the roles of all unmanned aerial vehicles, dividing all unmanned aerial vehicles into three levels of group level, team level and machine level, and setting group leaders, team leaders and team members; wherein the group level is composed of multiple unmanned aerial vehicle groups, the team level is composed of several unmanned aerial vehicles, and the single machine level is an individual unmanned aerial vehicle executing specific tasks; uploading tasks to the task pool in real time, each task containing a number, attributes, speed, coordinates, capacity, priority, time, group number, machine number and execution state; After the task is published, a synchronization timer is started, and within the time threshold, the unmanned aerial vehicles in the group level broadcast their own state vector, receive the state vector of the neighboring unmanned aerial vehicles and the task information, and perform autonomous negotiation and allocation of tasks based on the auction algorithm to generate a candidate allocation scheme; When there are still unallocated tasks in the task pool after the time threshold is reached, the group leader or the group leader performs centralized task allocation based on the particle swarm algorithm according to the relationship between the number of idle unmanned aerial vehicles in each team level and the number of unallocated tasks, and generates an optimal allocation scheme; When the team members execute tasks, it is judged in real time whether the remaining range of the unmanned aerial vehicle meets the task demand, if not, the team leader sends a task suspension request to the team leader, the team leader freezes the original task state, and reallocates the task until the team member completes the task.
[0010] Further, the group leader is responsible for coordinating cross-group task allocation and does not participate in specific task allocation; the team leader is responsible for coordinating task allocation of unmanned aerial vehicles within the group; the team member is an individual unmanned aerial vehicle executing specific tasks; wherein, The group leader is pre-subscribed by the ground, and when the group leader crashes or is disconnected, the surviving team leader selects a team leader with the smallest machine number to serve as a new group leader, and the new group leader is responsible for coordinating cross-group task allocation and only communicates with each team leader without participating in task allocation; The team leader is the unmanned aerial vehicle with the smallest machine number in the group, responsible for receiving group leader instructions and coordinating task allocation within the group, and communicating with the group leader and the team member.
[0011] Furthermore, the number is a unique serial number generated by the task; the attributes are divided into static targets and moving targets; the speed is the target's movement speed; the coordinates include the target's longitude, latitude, and altitude; the capacity is the number of drones required to execute the task; the priority is divided into two levels according to the target attributes, with moving targets having a priority of 1 and static targets having a priority of 2; the time is the task release time; the group number indicates that the task is preferentially assigned to a designated group; the execution status is divided into unassigned, in execution, and completed; and the drone number is the number of the drone assigned to the task.
[0012] Furthermore, the step of generating candidate allocation schemes through autonomous negotiation based on the auction algorithm includes: Select tasks from the task pool in descending order of priority, and for tasks of the same priority, in ascending order of time. Each idle drone determines whether it meets the mission requirements based on its remaining available range; if it does, it participates in the auction. The bidding value function for each drone participating in the auction is calculated, and the success of the bid is determined based on the bidding value function. The winning drone updates its own task status, and the team leader updates the execution status of the task pool to the group head. If there are still unassigned tasks in the task pool and there are still idle drones in the group, repeat the above steps.
[0013] Furthermore, the bidding value function is:
[0014] in, For drones The remaining available range, For drones Arrive at the mission point distance, As the first weighting coefficient, This is the second weighting coefficient; and The priority is determined by the task priority. High-priority tasks are prioritized based on proximity, while low-priority tasks are prioritized based on cost-effectiveness.
[0015] Furthermore, after the time threshold is reached, if there are still unassigned tasks in the task pool, and the number of idle drones in the group is greater than or equal to the number of unassigned tasks, the group leader will centrally assign tasks based on the particle swarm algorithm. If the number of idle drones in the group is less than the number of unassigned tasks, the group leader reports to the group leader, who then centrally assigns tasks based on the particle swarm optimization algorithm and distributes the results.
[0016] Furthermore, when there are still unassigned tasks in the task pool after the time threshold is reached, the step of centralized task allocation by the group leader or group head based on the particle swarm optimization algorithm, according to the relationship between the number of idle drones and the number of unassigned tasks within each group, includes: The group leader or group head collects the status information vectors and task information of all drones within the group or cluster in real time; For each task point Calculate all idle drones Distance to this point Select a candidate set of drones that meet the requirements of remaining available flight range. ; Initialization by A swarm of particles composed of vectors, for Each task has a position vector. and velocity vector Initial values are randomly generated within a preset range, and the position vector of each particle represents a task allocation scheme; among them, For the first Particle pairs at the task point The allocation result, whose value is the candidate set. Index; the velocity vector is used to adjust the direction and magnitude of changes in the index value; Calculate the fitness function ,in For drones Complete the task The contribution value, Record the position where each particle has the highest fitness. The position with the highest fitness in the particle swarm ; Update particle velocity and particle position ;in, As the first learning factor, As the second learning factor, The first random number, The second random number; Repeatedly update particle velocity and particle position until the maximum number of iterations is reached or fitness converges.
[0017] Furthermore, the mission abort request includes the current location and remaining fuel.
[0018] The technical solutions provided by the embodiments of this disclosure may include the following beneficial effects: In the embodiments of this disclosure, the above-described task allocation method based on dynamic dual-mode achieves the following: First, by defining a three-level role system (group leader, team leader, and team member) of "group-team-machine," hierarchical management of task coordination is realized, reducing system complexity and improving command efficiency. The group leader supports a dynamic election mechanism (e.g., the lowest-ranking machine number is replaced by the team leader), preventing system paralysis due to central node failure and enhancing the system's survivability in highly confrontational environments such as battlefields. The group leader communicates only with the team leader, who is responsible for coordination within the team. Local broadcasting is used during the distributed negotiation phase, effectively reducing the overall network communication pressure and making it suitable for large-scale UAV swarms.
[0019] Secondly, after a task is released, a distributed auction algorithm is prioritized for rapid negotiation to ensure real-time performance. If a task remains unassigned after a timeout, a centralized particle swarm optimization algorithm is initiated for global optimization, balancing efficiency and optimality. Tasks are prioritized based on their attributes (dynamic / static objectives), and the auction algorithm dynamically adjusts bidding weights according to these priorities, ensuring priority allocation of important tasks and rational resource scheduling. Task information is stored in a structured manner, with real-time status updates (unassigned / in execution / completed), supporting parallel management of multiple tasks and cross-group coordination, thereby improving the overall task throughput of the system.
[0020] Thirdly, setting time thresholds prevents distributed negotiation from proceeding indefinitely, ensuring the timeliness of the task allocation process. During the drone's mission, it continuously assesses whether the remaining range meets the requirements; if not, it terminates the mission in real time and triggers a reallocation, improving the success rate and security of mission execution. The distributed phase supports rapid decision-making based on local information, while the centralized phase supports cross-group resource integration. The system can flexibly respond to dynamic scenarios such as new tasks, target movement, and changes in drone status.
[0021] Fourthly, a bidding mechanism based on distance and remaining range, combined with priority weights, enables rapid and fair task allocation, reduces negotiation oscillations, and improves convergence speed. In the centralized phase, particle swarm optimization is used to globally match and optimize multiple tasks and multiple UAVs, improving overall system performance and avoiding local suboptimal solutions. The distributed phase is suitable for high real-time, small-scale scenarios, while the centralized phase is suitable for complex constraints and cross-group coordination scenarios. The combination of both gives the system broad scenario adaptability. Attached Figure Description
[0022] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.
[0023] Figure 1This diagram illustrates the steps of a task allocation method based on a dynamic dual-mode approach in an exemplary embodiment of this disclosure. Figure 2 A flowchart illustrating the implementation of the dynamic dual-mode task allocation method in an exemplary embodiment of this disclosure is shown. Figure 3 This diagram illustrates the role definition of 20 clustered drones in an exemplary embodiment of this disclosure; Figure 4 A flowchart illustrating the implementation of distributed task allocation based on an auction algorithm in an exemplary embodiment of this disclosure is shown. Figure 5 A flowchart illustrating the implementation of centralized task allocation based on particle swarm optimization algorithm in an exemplary embodiment of this disclosure is shown. Figure 6 A diagram illustrating the allocation process of the task allocation method based on dynamic dual-mode in an exemplary embodiment of this disclosure is shown. Figure 7 The diagram shows the allocation result of the task allocation method based on dynamic dual-mode in an exemplary embodiment of this disclosure.
[0024] Figure 8 This diagram illustrates the priority allocation process of the dynamic dual-mode task allocation method in an exemplary embodiment of this disclosure.
[0025] Figure 9 This diagram illustrates the priority allocation results of the task allocation method based on dynamic dual-mode in an exemplary embodiment of this disclosure. Detailed Implementation
[0026] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
[0027] Furthermore, the accompanying drawings are merely illustrative diagrams of embodiments of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities.
[0028] This example implementation provides a task allocation method based on dynamic dual-mode. (See reference...) Figure 1 As shown, the task allocation method based on dynamic dual-mode can include: Step S101: Define roles for all drones, dividing them into three levels: swarm, group, and drone, and assigning a swarm leader, group leader, and group members. The swarm level consists of multiple drone groups, the group level consists of several drones, and the drone level consists of individual drones performing specific tasks. Step S102: Upload tasks to the task pool in real time. Each task includes number, attributes, speed, coordinates, capacity, priority, time, group number, machine number, and execution status. Step S103: After the task is released, start the synchronization timer. Within the time threshold, the drones in the group broadcast their own state vectors, receive the state vectors of neighboring drones and task information, and autonomously negotiate and allocate tasks based on the auction algorithm to generate candidate allocation schemes. Step S104: When there are still unassigned tasks in the task pool after the time threshold is reached, the group leader or group head will perform centralized task allocation based on the relationship between the number of idle drones and the number of unassigned tasks in each group level, and generate the optimal allocation scheme. Step S105: When a team member is performing a task, they should determine in real time whether their remaining flight range meets the task requirements. If not, they should send a task abort request to the team leader. The team leader should freeze the original task status and reassign the task until the team member completes the task.
[0029] The above-described dynamic dual-mode task allocation method achieves two key benefits. First, by defining a three-tiered "group-machine" role structure (group leader, group leader, and group member), hierarchical management of task coordination is achieved, reducing system complexity and improving command efficiency. The group leader supports a dynamic election mechanism (e.g., the lowest-ranking drone takes over as group leader), preventing system paralysis due to central node failure and enhancing the system's survivability in highly competitive environments such as battlefields. The group leader communicates only with the group leader, who is responsible for coordination within the group. Local broadcasting is used during the distributed negotiation phase, effectively reducing network-wide communication pressure and making it suitable for large-scale drone swarms. Second, after task release, a distributed auction algorithm is prioritized for rapid negotiation to ensure real-time performance. If a task remains unassigned after a timeout, a centralized particle swarm optimization algorithm is initiated for global optimization, balancing efficiency and optimality. Tasks are prioritized based on attributes (dynamic / static targets), and the auction algorithm dynamically adjusts bidding weights according to priority, ensuring priority allocation of important tasks and rational resource scheduling. Task information is stored in a structured manner, with real-time status updates (unassigned / in execution / completed), supporting multi-task parallel management and cross-group coordination, thus improving the overall task throughput of the system. Thirdly, setting time thresholds prevents distributed negotiation from going on indefinitely, ensuring the timeliness of the task allocation process. During the UAV's mission, the remaining range is continuously assessed to determine if it meets the requirements; if not, the mission is terminated in real-time and a reassignment is triggered, improving the success rate and safety of mission execution. The distributed phase supports rapid decision-making based on local information, while the centralized phase supports cross-group resource integration. The system can flexibly handle dynamic scenarios such as new tasks, target movement, and changes in UAV status. Fourthly, a bidding mechanism based on distance and remaining range, combined with priority weights, achieves fast and fair task allocation, reducing negotiation oscillations and improving convergence speed. In the centralized phase, particle swarm optimization is used to globally match and optimize multiple tasks and multiple UAVs, improving overall system performance and avoiding local suboptimal solutions. The distributed phase is suitable for high real-time, small-scale scenarios, while the centralized phase is suitable for complex constraints and cross-group coordination scenarios. The combination of the two gives the system broad scenario adaptability.
[0030] Below, we will refer to Figures 1 to 9 The steps of the dynamic dual-mode task allocation method described above in this example embodiment will be explained in more detail.
[0031] This invention proposes a task allocation method based on dynamic dual-mode, which mainly consists of five implementation stages, as follows: Figure 2 As shown.
[0032] In step S101, roles are defined for all drones, and all drones are divided into three levels: group level, cluster level, and drone level. Group leader, cluster leader, and cluster members are set. The group level consists of multiple drone groups, the cluster level consists of several drones, and the drone level is an individual drone that performs a specific task.
[0033] Specifically, the roles of swarm drones are defined as follows: A swarm is divided into three levels: swarm, group, and drone. The swarm level consists of multiple drone groups, the group level consists of several drones, and the individual drone level consists of a drone performing a specific task. The corresponding roles are swarm leader, group leader, and group member. The swarm leader is pre-selected on the ground. When the swarm leader crashes or loses contact, the surviving group leaders elect the group leader with the drone number smallest as the new swarm leader. This new leader is responsible for coordinating cross-group task allocation but does not participate in task allocation; they only communicate with the group leaders. The group leader is the drone with the smallest drone number within that group. They are responsible for receiving instructions from the swarm leader and coordinating task allocation among the drones in the group. They can communicate with the swarm leader and group members.
[0034] like Figure 3 As shown, roles are defined for the 20 drones in the cluster, with drone #1 being the group leader, drone #101 being the group head, and the remaining drones being group members.
[0035] In step S102, tasks are uploaded to the task pool in real time. Each task includes a number, attributes, speed, coordinates, capacity, priority, time, group number, machine number, and execution status.
[0036] Specifically, task pool management involves manual real-time uploading of tasks to the task pool. Each task includes information such as task number, attributes, speed, coordinates, capacity, priority, time, group number, drone number, and execution status. The task number is the unique serial number generated by the task; attributes are divided into static targets (value S) and moving targets (value D); speed refers to the target's movement speed in m / s; coordinates include the target's longitude, latitude, and altitude; capacity refers to the number of drones required to execute the task; priority is divided into two levels based on target attributes: moving targets have priority 1, and static targets have priority 2; time refers to the time the task was published; the group number indicates that the task is preferentially assigned to a specific group; the execution status is divided into unassigned, in progress, and completed; and the drone number refers to the drone assigned to the task.
[0037] In step S103, after the task is published, a synchronization timer is started. Within the time threshold, UAVs in the group broadcast their own state vectors, receive the state vectors of neighboring UAVs and task information, and autonomously negotiate and allocate tasks based on the auction algorithm to generate candidate allocation schemes.
[0038] Specifically, in the distributed task allocation phase after task release: After task release, a synchronization timer starts (preset time threshold t1), and the cluster of UAVs broadcasts its own state vector and receives task information and state vectors from neighboring UAVs. The UAV state vector includes, but is not limited to, UAV number, group number, payload capacity, current status, current task number, list of pending task numbers, remaining range, and coordinates. Task status is categorized as idle, executing a task, returning to base, and undergoing maintenance. The cluster of UAVs autonomously negotiates and allocates tasks based on an auction algorithm, generating candidate allocation schemes; the process is described below. Figure 4Specifically, it is described as follows: Step S1031: Start with the highest priority task in the task pool, and for tasks of the same priority, the earlier the time, the higher the priority.
[0039] Step S1032: Each idle drone first determines whether it can complete the task. The determination method is as follows: ,in For drones Remaining available range, For dynamic safety factor, For drones Arrive at the mission point If the distance is such that the inequality holds, then the drone will be auctioned.
[0040] In one embodiment, The value depends on the drone's fuel consumption performance, and is usually taken as... .
[0041] Step S1033: Calculate the bidding value function for each drone participating in the auction, with the bidding value being distance and remaining range: ,in and The weighting coefficients are determined based on task priority.
[0042] In one embodiment, for high-priority tasks, the nearest neighbor is preferred, and the task with the highest priority is selected. For low-priority tasks, prioritize cost-effectiveness and select the best option. . Step S1034: The winning drone updates its own task status, and the group leader updates the execution status of the task pool to the group head. If there are still unassigned tasks in the task pool and there are still idle drones in the group, repeat steps S1031 to S1033.
[0043] In step S104, when there are still unassigned tasks in the task pool after the time threshold is reached, the group leader or group head performs centralized task allocation based on the relationship between the number of idle drones and the number of unassigned tasks in each group level, and generates the optimal allocation scheme.
[0044] Specifically, in the centralized task allocation phase: if there are still unassigned tasks in the task pool after the time threshold t1 is reached, and the number of idle drones in the group is... The number of unassigned tasks is centrally allocated by the group leader using the particle swarm optimization (PSO) algorithm. If the number of idle drones in the group exceeds the number of unassigned tasks, the group leader reports this to the group head, who then centrally allocates tasks using the PSO algorithm and distributes the results to all group leaders. The process for centralized task allocation using the PSO algorithm is detailed below. Figure 5 Specifically, it is described as follows: Step S1041: The group leader or group head collects the status information vectors and task information of all drones in the group or cluster in real time.
[0045] Step S1042: For each task Calculate all idle drones Distance to this point Filter out those that meet the remaining flight distance Candidate set of drones .
[0046] In one embodiment, The value depends on the drone's fuel consumption performance, and is usually taken as... ; Step S1043: Initialize by A swarm of particles composed of vectors, for Each particle has a position vector and a velocity vector. Initial values are randomly generated within a specified range, and the position vector of each particle represents a task allocation scheme. Representing the Particle pairs task The allocation result, the value is the candidate set. (Index), the velocity vector is used to adjust the direction and magnitude of changes in these index values.
[0047] Step S1044: Calculate the fitness function ,in For drones Complete the task The contribution value is determined by recording the position with the highest fitness for each particle and the position with the highest fitness in the particle swarm. , .
[0048] In one embodiment, for high-priority tasks... For high-priority tasks, take .
[0049] Step S1045: Update particle velocity Particle position .
[0050] In one embodiment, the learning factor is taken in the experiment. , for Random number.
[0051] Step S1046: Repeat steps S1044 and S1045 until the maximum number of iterations or fitness convergence is reached.
[0052] In step S105, when a team member is performing a task, they determine in real time whether their remaining range meets the task requirements. If not, they send a task abort request to the team leader. The team leader freezes the original task status and reassigns the task until the team member completes the task.
[0053] Specifically, in the real-time monitoring phase: the task allocation results are shown below. Figure 7 Upon receiving the task, team members execute it and continuously assess whether their remaining range meets the task requirements. The calculation method is as follows: ,in For drones Remaining available range, The dynamic safety factor is usually taken as , For drones Arrive at the mission point The distance. If the conditions are not met, a task abort request (including current location and remaining fuel) is sent to the team leader. The team leader quickly freezes the original task status and returns to step S103 for reassignment.
[0054] In one specific embodiment, targets 1, 2, 3, and 4 are each assigned to one drone. The swarm of drones receives the targets and autonomously assigns them within the swarm, transmitting the real-time assignment results. The target points to be executed by each drone are connected by dashed lines to facilitate observation of which drone is executing each target. All four targets are connected by swarm drones. Before the task assignment is completed, a single target may appear to be executed by multiple drones. Figure 6 The drones connected to target 4 are shown as drones 5, 7, and 8. The final allocation result is shown as follows. Figure 7 As shown, Target 1 is Unit 2, Target 2 is Unit 3, Target 3 is Unit 12, and Target 4 is Unit 5.
[0055] In one specific embodiment, target 1 is set to have a higher priority than target 2, and the swarm of drones prioritizes executing target 1 during the task allocation process. Figure 8 The system shows that drone #3 has been assigned target 2, but the assignment completion time has not yet arrived. During flight, drone #3 reaches a position equidistant from targets 1 and 2. When a drone is equidistant from multiple targets, the target value plays a decisive role in mission planning. Drone #3 is assigned to target 1 in real time. Figure 9 As shown.
[0056] The above-described dynamic dual-mode task allocation method achieves several advantages. First, by defining a three-tiered "group-machine" role structure (group leader, team leader, and team member), hierarchical management of task coordination is achieved, reducing system complexity and improving command efficiency. The group leader supports a dynamic election mechanism (e.g., the lowest-ranking drone number takes over as group leader), preventing system paralysis due to central node failure and enhancing the system's survivability in highly competitive environments such as battlefields. The group leader communicates only with the team leader, who is responsible for coordination within the team. The distributed negotiation phase utilizes local broadcasting, effectively reducing network-wide communication pressure and making it suitable for large-scale drone swarms.
[0057] Secondly, after a task is released, a distributed auction algorithm is prioritized for rapid negotiation to ensure real-time performance. If a task remains unassigned after a timeout, a centralized particle swarm optimization algorithm is initiated for global optimization, balancing efficiency and optimality. Tasks are prioritized based on their attributes (dynamic / static objectives), and the auction algorithm dynamically adjusts bidding weights according to these priorities, ensuring priority allocation of important tasks and rational resource scheduling. Task information is stored in a structured manner, with real-time status updates (unassigned / in execution / completed), supporting parallel management of multiple tasks and cross-group coordination, thereby improving the overall task throughput of the system.
[0058] Thirdly, setting time thresholds prevents distributed negotiation from proceeding indefinitely, ensuring the timeliness of the task allocation process. During the drone's mission, it continuously assesses whether the remaining range meets the requirements; if not, it terminates the mission in real time and triggers a reallocation, improving the success rate and security of mission execution. The distributed phase supports rapid decision-making based on local information, while the centralized phase supports cross-group resource integration. The system can flexibly respond to dynamic scenarios such as new tasks, target movement, and changes in drone status.
[0059] Fourthly, a bidding mechanism based on distance and remaining range, combined with priority weights, enables rapid and fair task allocation, reduces negotiation oscillations, and improves convergence speed. In the centralized phase, particle swarm optimization is used to globally match and optimize multiple tasks and multiple UAVs, improving overall system performance and avoiding local suboptimal solutions. The distributed phase is suitable for high real-time, small-scale scenarios, while the centralized phase is suitable for complex constraints and cross-group coordination scenarios. The combination of both gives the system broad scenario adaptability.
[0060] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of embodiments of this disclosure, "a plurality of" means two or more, unless otherwise explicitly specified.
[0061] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.
[0062] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.
Claims
1. A task allocation method based on dynamic dual-mode, characterized in that, include: All drones are assigned roles and divided into three levels: swarm, group, and drone. A swarm leader, group leader, and group members are assigned. The swarm level consists of multiple drone groups, the group level consists of several drones, and the drone level consists of individual drones that perform specific tasks. Tasks are uploaded to the task pool in real time. Each task includes a number, attributes, speed, coordinates, capacity, priority, time, group number, machine number, and execution status. After the task is released, a synchronization timer is started. Within the time threshold, drones in the group broadcast their own state vectors, receive the state vectors of neighboring drones and task information, and autonomously negotiate and allocate tasks based on the auction algorithm to generate candidate allocation schemes. When there are still unassigned tasks in the task pool after the time threshold is reached, the group leader or group head will perform centralized task allocation based on the relationship between the number of idle drones and the number of unassigned tasks in each group level, and generate the optimal allocation scheme. When a crew member is performing a task, they must determine in real time whether their remaining range meets the task requirements. If not, they must send a task abort request to the team leader. The team leader will then freeze the original task status and reassign the task until the crew member completes it.
2. The task allocation method based on dynamic dual-mode according to claim 1, characterized in that, The group leader is responsible for coordinating cross-group task allocation but does not participate in specific task allocation; the group leader is responsible for coordinating drone task allocation within the group; group members are individual drones that perform specific tasks; among them, The group leader is pre-installed on the ground. When the group leader crashes or loses contact, the surviving group leaders elect the group leader with the smallest machine number to serve as the new group leader. The new group leader is responsible for coordinating the cross-group task allocation, only communicating with the group leaders and not participating in task allocation. The group leader is the drone with the smallest serial number in the group. The leader is responsible for receiving instructions from the group leader, coordinating task allocation within the group, and communicating with the group leader and group members.
3. The task allocation method based on dynamic dual-mode according to claim 1, characterized in that, The number is a unique serial number generated by the task; the attributes are divided into static targets and moving targets; the speed is the target's movement speed; the coordinates include the target's longitude, latitude, and altitude; the capacity is the number of drones required to execute the task; the priority is divided into two levels according to the target attributes, with moving targets having a priority of 1 and static targets having a priority of 2; the time is the task release time; the group number indicates that the task is preferentially assigned to a designated group; the execution status is divided into unassigned, in progress, and completed; the drone number is the number of the drone assigned to the task.
4. The task allocation method based on dynamic dual-mode according to claim 1, characterized in that, The steps for generating candidate allocation schemes through autonomous negotiation based on auction algorithms include: Select tasks from the task pool in descending order of priority, and for tasks of the same priority, in ascending order of time. Each idle drone determines whether it meets the mission requirements based on its remaining available range; if it does, it participates in the auction. The bidding value function is calculated for each drone participating in the auction, and the bidder with the highest bidding value function value wins the bid. The winning drone updates its own task status, and the team leader updates the execution status of the task pool to the group head. If there are still unassigned tasks in the task pool and there are still idle drones in the group, repeat the above steps.
5. The task allocation method based on dynamic dual-mode according to claim 4, characterized in that, The bidding value function is: in, For drones The remaining available range, For drones Arrive at the mission point distance, As the first weighting coefficient, This is the second weighting coefficient; and The priority is determined by the task priority. High-priority tasks are prioritized based on proximity, while low-priority tasks are prioritized based on cost-effectiveness.
6. The task allocation method based on dynamic dual-mode according to claim 5, characterized in that, If, after the time threshold is reached, there are still unassigned tasks in the task pool, and the number of idle drones in the group is greater than or equal to the number of unassigned tasks, the group leader will centrally assign tasks based on the particle swarm algorithm. If the number of idle drones in the group is less than the number of unassigned tasks, the group leader reports to the group leader, who then centrally assigns tasks based on the particle swarm optimization algorithm and distributes the results.
7. The task allocation method based on dynamic dual-mode according to claim 6, characterized in that, When there are still unassigned tasks in the task pool after the time threshold is reached, the group leader or group head performs centralized task allocation based on the relationship between the number of idle drones and the number of unassigned tasks within each group level. This includes the following steps: The group leader or group head collects the status information vectors and task information of all drones within the group or cluster in real time; For each task point Calculate all idle drones Distance to this point Select a candidate set of drones that meet the requirements of remaining available flight range. ; Initialization by A swarm of particles composed of vectors, for Each task has a position vector. and velocity vector Initial values are randomly generated within a preset range, and the position vector of each particle represents a task allocation scheme; among them, For the first Particles at the task point The allocation result, whose value is the candidate set. Index; the velocity vector is used to adjust the direction and magnitude of changes in the index value; Calculate the fitness function ,in For drones Complete the task The contribution value, Record the position where each particle has the highest fitness. The position with the highest fitness in the particle swarm ; Update particle velocity and particle position ;in, As the first learning factor, As the second learning factor, The first random number, The second random number; Repeatedly update particle velocity and particle position until the maximum number of iterations is reached or fitness converges.
8. The task allocation method based on dynamic dual-mode according to claim 1, characterized in that, The mission abort request includes the current location and remaining fuel.