Task adaptive allocation method for unmanned aerial vehicle group man-machine system

Abstract

The invention discloses a task adaptive allocation method for an unmanned aerial vehicle group man-machine system. The task adaptive allocation method comprises the following steps: S1, constructing an adaptive man-machine system task allocation module framework; s2, establishing a human model for an operator based on a cognitive load theory and the like, and establishing an unmanned aerial vehicle model for an unmanned aerial vehicle group based on theories such as flight dynamics and the like; s3, forming a man-machine function preliminary distribution strategy of a tetragonal agent based on a top Agent perception and decision-making module of a self-adaptive man-machine system; the states of the unmanned aerial vehicle group and the operators are monitored in real time; and S4, based on an evaluation result and a control right switching rule, performing secondary task allocation on a man-machine function through an adaptive man-machine system task allocation module, and adjusting specific tasks of each Agent. According to the method, the dynamic problem of man-machine function distribution of an unmanned aerial vehicle group man-machine system under complex multiple constraints is solved, the reasonability and safety of man-machine function distribution under multiple influence factors are ensured, and the efficiency of executing complex tasks by an unmanned aerial vehicle group is improved.

Description

Technical Field

[0001] The present invention relates to the field of drone swarm human-machine systems, and in particular to a method for adaptively allocating tasks for drone swarm human-machine systems. The method can dynamically adjust task allocation according to the status of operators and drones during task execution, and belongs to the technical scope of human-machine collaboration and task scheduling. Background Art

[0002] With the rapid development of drone technology, drone swarms have become widely used in complex missions with constantly changing real-time conditions. Operators and drones must maintain a high degree of coordination during mission execution. However, due to the variability of the environment and fluctuations in the mission load, traditional fixed task allocation methods are unable to flexibly adapt to these changes, often resulting in reduced mission efficiency and even potential safety hazards.

[0003] When performing tasks in complex environments, drone swarm human-machine systems often face challenges such as high task loads for both operators and drones, dynamic environmental changes, and difficulties in coordinated operations. Furthermore, the sudden nature of tasks and environmental uncertainty mean that both the operator's workload and the drone's performance status may fluctuate at any time. Traditional task allocation methods lack flexibility and real-time responsiveness, potentially leading to inefficient execution and even mission failure. Therefore, an adaptive system is urgently needed that can adjust task allocation in real time based on the status of the drone swarm human-machine system and environmental information to ensure efficient and safe execution in complex environments. Summary of the Invention

[0004] This paper proposes an adaptive task allocation method for a drone swarm human-machine system. This method, through an adaptive task allocation system framework, achieves intelligent allocation of drone swarm task execution and operator collaboration. This method utilizes a top-level agent for task allocation and scheduling, enabling real-time collection and analysis of operator load status and drone task execution. It dynamically adjusts task allocation strategies based on system feedback, thereby improving the efficiency and safety of task completion.

[0005] To achieve the above objectives, the present invention provides a method for adaptively allocating tasks to a human-machine system of a swarm of drones, comprising the following steps:

[0006] S1 builds the framework of the adaptive human-machine-robot (HMR) task allocation module, clarifying the basic responsibilities of the four intelligent agents including the operator, the top-level agent, the leader drone, and the follower drones;

[0007] S2 builds a human model for the operator based on cognitive load theory and other theories, and builds a machine model for the drone swarm based on flight dynamics theory to evaluate the real-time status of the operator and drones;

[0008] S3, based on HMR's top-level agent perception and decision-making module, forms a preliminary human-machine function allocation strategy for the four-party intelligent agent; monitors the status of the drone swarm and operators in real time; and evaluates the current task execution effect and safety based on environmental information;

[0009] Based on the evaluation results and control switching rules, S4 reassigns tasks to human-machine functions through the adaptive human-machine system task allocation module and adjusts the specific tasks of each agent.

[0010] Specifically, the constructed HMR (Human-Machine-Robot) task allocation module framework includes: an operator, a top-level agent, a leader drone, and follower drones. This framework clarifies the basic responsibilities and relationships of each agent, providing structural support for dynamic task allocation in human-machine collaboration. The operator's responsibilities include providing human oversight and support, reviewing the top-level agent's task planning and allocation, and intervening to adjust and control tasks when necessary. The top-level agent's responsibilities include providing global situational awareness, task planning, and oversight for the entire mission. The leader drone's responsibilities include navigating, commanding, and leading the entire drone swarm. The follower drones' responsibilities include assisting the leader drone and improving the breadth and accuracy of data collection through a distributed approach.

[0011] Specifically, in the adaptive task allocation system, the system introduces a human capability model to evaluate the operator's status in real time to reflect the operator's capability status at different times. By dynamically updating the operator's capability status parameters, the task allocation process can be adjusted according to the operator's actual situation.

[0012] Among them, the indicators involved in the human capability model include operator capability changes, fatigue and other performance parameters, which are used to dynamically evaluate the operator's status during the task allocation process.

[0013] Specifically, the drone performance model is a core module in the adaptive task allocation system. It is used to dynamically evaluate each drone's real-time status and execution capabilities in different mission environments. By monitoring multiple performance indicators in real time, the model provides a basis for task allocation and reallocation, ensuring that the system can optimize and adjust based on the drone's actual conditions, improving the efficiency and stability of task execution.

[0014] The drone swarm consists of a leader drone and follower drones, and a drone model is built based on the drones' performance conditions. The drone model includes parameters such as the drones' flight status and flight performance, which are used to dynamically evaluate the drones' mission execution capabilities during mission execution.

[0015] Specifically, the top-level agent situational awareness module collects status data from the drone swarm and operators. This data is combined with mission information and real-time environmental information, and, with the goal of achieving comprehensive mission execution, outputs a task fitness matrix and a task allocation matrix. These serve as preliminary task allocation results, forming the initial human-machine function allocation strategy for the HMR system. Through information exchange and status feedback, each agent achieves a comprehensive understanding of the mission environment and drone swarm status. Continuous feedback and supervision optimize the task allocation strategy and ensure the rationality of the initial allocation.

[0016] Among them, the situational awareness module of the top-level agent collects information through various sensors carried by the drone swarm to provide data support for preliminary task allocation; the decision-making reasoning module proposes an optimized HMR system human-machine function allocation strategy through continuous iterative optimization.

[0017] The top-level agent analyzes the operator's workload, fatigue, and other performance indicators, as well as the drone's battery level and payload, and then re-evaluates task priorities and allocation strategies. Using the HMR adaptive algorithm, the system optimizes task allocation while satisfying multiple constraints.

[0018] Specifically, the top-level agent-based situational awareness module predicts the mission achievement of the drone swarm's human-machine system and performs real-time safety assessments. When the operator or drone status changes (for example, when the system detects that the operator or drone's status approaches a preset critical value, such as when operator fatigue exceeds a threshold or the drone's battery level falls below a safe level), or when new conditions emerge in the mission environment (such as changes in environmental information), the HMR adaptive algorithm dynamically adjusts human-machine functions. The adaptive module adjusts task allocation based on the constraints of the "human-machine-environment" three state information, aiming to maximize the efficiency of drone swarm mission execution by optimizing the overall task adaptability, comprehensive execution effectiveness, and comprehensive safety.

[0019] The operator and the top-level agent can monitor each other bidirectionally, allowing the operator to oversee the top-level agent's decisions in real time during mission execution. In special circumstances, such as unexpected changes in the drone swarm and the environment (such as a sudden change in the mission execution environment or loss of drone control), or when the top-level agent's decision is inconsistent with the actual mission requirements, the operator can proactively intervene in the top-level agent's task allocation decisions if they discover deviations or potential risks.

[0020] Among them, after the operator intervention, the top-level agent re-evaluates the task allocation, reviews the plan according to the current situation and the status of the drone, and optimizes it based on the feedback information to ensure the maximum utilization of system resources, thereby ensuring the flexibility of the system and the safety of task execution.

[0021] Specifically, the present invention supports the recording of monitoring data and the generation of task reports after task completion. The task report includes the overall completion effect of the task, etc. This report provides a data basis for subsequent task allocation optimization and resource management.

[0022] Specifically, the method for adaptively allocating human-machine functions in a drone swarm system of the present invention can achieve flexible human-machine function allocation in a complex mission environment, effectively improving the efficiency, safety, and resource utilization of mission execution.

[0023] The technical effects of the present invention are:

[0024] This invention utilizes an adaptive task allocation and decision-making feedback mechanism to monitor operator and drone status in real time and flexibly adjust task allocation based on changes in the mission environment. This improves the system's response speed and accuracy in high-risk scenarios, effectively addressing the rigid and unresponsive task allocation issues faced by drone swarms in complex and changing environments. It optimizes resource utilization, enhances the safety and reliability of mission execution, and supports the accumulation of mission data, providing a data foundation for continuous system optimization. This enables the system to demonstrate exceptional adaptability and efficiency in complex tasks such as fire monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 A flowchart of a method for adaptively allocating tasks to a human-machine system of a swarm of drones provided by the present invention;

[0026] Figure 2 A module framework for adaptive HMR task allocation system;

[0027] Figure 3 This is a schematic diagram of the top-level Agent module framework;

[0028] Figure 4 The logic flow chart of the adaptive multi-objective optimization algorithm;

[0029] Figure 5 This is the logical framework diagram of the two-way supervision decision-making between the operator and the top-level agent. DETAILED DESCRIPTION

[0030] In order to more clearly understand the implementation scheme and technical solution of the present invention, the following embodiments are described in detail with reference to the accompanying drawings. The embodiments are some examples of the present invention, and other embodiments that can be implemented by those skilled in the art without creative work are also within the scope of protection of the present invention. In addition, the various embodiments or technical features of the present invention can be combined with each other to form a feasible technical solution, but it is necessary to ensure that no technical contradictions or unfeasible situations arise after the combination. If the combination is not feasible under certain circumstances, the combination is not within the scope of protection of the present invention.

[0031] With the development of drone technology, drone swarms are widely used in complex mission scenarios. Operators and drones must maintain a high degree of coordination during mission execution to cope with changing environments and dynamic loads. However, mission status is often affected by factors such as changes in status information from operators, drone swarms, and the environment. Traditional fixed task allocation methods are difficult to meet the mission requirements in complex mission scenarios, which can easily lead to reduced efficiency and safety risks. Therefore, an adaptive task allocation system based on operator status and drone performance is urgently needed to ensure that drone swarms can respond flexibly in complex environments.

[0032] Based on this, this paper proposes an adaptive task allocation method for a human-machine system in a swarm of drones. First, an adaptive human-machine system task allocation module framework is constructed, clarifying the basic responsibilities of the four intelligent agents (operator, top-level agent, leader drone, and follower drones), laying the foundation for task execution. Then, a human model of the operator is established based on theories such as cognitive load, and a machine model of the drone swarm's performance is constructed based on theories such as flight dynamics. Environmental information parameters are simultaneously acquired to capture real-time state changes in the "human-machine-environment" relationship. Next, the top-level agent's perception and decision-making module analyzes the system's real-time status, forming a preliminary allocation strategy for the four intelligent agents and evaluating the execution performance and safety of the current task. Finally, based on the evaluation results and control authority switching rules, the adaptive human-machine system task allocation module performs a secondary task allocation among human and machine functions, adjusting the specific tasks of each agent to achieve dynamic coordination between the drone swarm and the operator, ensuring efficient task completion and system safety.

[0033] See also Figure 1 The present invention provides a method for adaptively allocating tasks to a human-machine system of a drone swarm, comprising the following steps:

[0034] S1 builds the framework of the adaptive human-machine-robot (HMR) task allocation module, clarifying the basic responsibilities of the four intelligent agents including the operator, the top-level agent, the leader drone, and the follower drones;

[0035] It is worth noting that, see Figure 2The framework constructs a modular framework for adaptive HMR task allocation based on agent conditions and performance, capable of adaptively distributing homogeneous, independent task workloads across diverse scenarios. The framework consists of removable modular functional blocks, with independent modules evaluating human and robot status and performance, ensuring real-time workload allocation based on the health and performance of both the human operator and the autonomous robot. Furthermore, the newly introduced top-level agent module enables smooth task redistribution, mitigating the potential negative impact of sudden changes in workload or tasks on operator or drone performance.

[0036] Among them, see Figure 3 The top-level agent is designed as a situational awareness module consisting of four functional modules: target recognition, state prediction, effect estimation, and task safety, and an independent decision-making module, to achieve efficient task execution and human-machine collaboration. This structural setting is based on situational awareness theory, distributed artificial intelligence theory, and hierarchical control theory: situational awareness theory enables each agent to focus on different information processing tasks, achieving comprehensive perception of the environment; distributed artificial intelligence theory gives agents the ability to analyze information and perform parallel processing, improving the system's response speed and resource utilization efficiency; hierarchical control theory ensures a clear division of labor between the situational awareness module and the decision-making module: the former provides real-time status information, while the latter is responsible for centralized task decision-making. Through this division of labor and collaboration, the system can maintain stability in complex dynamic environments, ensuring the accuracy and efficiency of task execution.

[0037] A drone swarm consists of a leader drone and multiple follower drones. The leader drone performs core functions within the swarm, including data collection and reporting, task allocation, coordinated control, and communication relay. As the swarm's coordination center, the leader drone aggregates the status, location, and mission progress of each drone, reporting them in real time to the top-level agent to support decision-making. Simultaneously, the leader drone transmits task assignments from the top-level agent to each drone, dynamically adjusting them during mission execution. It also acts as a relay in the communication link, ensuring smooth information flow between the swarm, the top-level agent, and the operator. Through its own perception and information fusion, the leader drone provides comprehensive situational awareness, supports efficient collaborative operation within the swarm, and enhances the swarm's ability to perform in complex mission environments.

[0038] The operator performs key functions of monitoring, decision-making, and direct control within a swarm mission. Through the top-level agent, they obtain information on the swarm's status and mission progress to ensure smooth mission execution. In specific situations, such as mission changes or loss of drone control, the operator and the top-level agent can engage in two-way oversight. The operator monitors the top-level agent's decisions in real time during mission execution and proactively intervenes to adjust task allocations if deviations or potential risks are detected. Furthermore, after mission execution, the top-level agent reassesses task allocations based on operator feedback to ensure maximum utilization of system resources. The operator can also optimize the plan based on the current mission situation and drone status, ensuring system flexibility and mission safety.

[0039] S2 builds a human model for the operator based on cognitive load theory and other theories, and builds a machine model for the drone swarm based on flight dynamics theory to evaluate the real-time status of the operator and drones;

[0040] During the development of the human model, parameters were primarily based on cognitive load theory and psychophysiological research. Based on cognitive load theory, the model incorporates a parameter reflecting changes in operator performance to assess the operator's cognitive load during task execution. This parameter is set based on indicators such as task complexity, attention allocation, and task frequency. Furthermore, psychophysiological research has shown that operator fatigue is a significant factor influencing task performance. Therefore, a fatigue parameter was incorporated into the model to characterize the operator's ongoing work status by reflecting declining physical and mental energy.

[0041] Among them, the ability function of a person at time k is defined as c h (k)∈[0,1]. Then the capability change equation can be expressed as:

[0042] c h (k+1)=min{1,max{0,c h (k)+αΔp(k)-βf h (k)}}

[0043] Where: c h (k) the operator's capability status at time k; c h (k+1) is the operator's ability status at time k+1, ranging from [0,1]; α is the positive impact coefficient of performance change on ability, which is used to amplify the impact of positive change on ability status; β is the negative impact coefficient of fatigue on ability, which is used to indicate the inhibitory effect of fatigue on the operator's ability; f h (k) is the fatigue parameter, which indicates the fatigue level of the operator at time k, ranging from [0, 1]; Δp(k) is the change in the performance index at time k.

[0044] The UAV model is based on flight dynamics, resource optimization theory, and multi-constraint control theory. Resource optimization theory ensures mission completion with minimal battery consumption by rationally allocating flight paths, speeds, and payloads within limited battery capacity. Multi-constraint control theory ensures stable and safe operation within multiple constraints, including altitude, speed, battery level, and payload. Core model parameters such as altitude, speed, battery level, and mission payload are set based on these theories, achieving the dual optimization of minimizing battery consumption and minimizing mission completion time.

[0045] Among them, the UAV performance model is expressed by constrained optimization, and the UAV performance change equation can be expressed as:

[0046]

[0047] The following constraints are met:

[0048] in, represents the total battery consumption, and K represents the total time to complete the task.

[0049] Where h(k) represents the flight altitude at time k, subject to the upper and lower limits H min and H max ; v(k) represents the flight speed at time k, which is limited by the upper and lower speed limits V min and V max ; b(k) represents the battery power, which must meet the minimum power requirement B min ; L(k) represents the task load, which must not exceed the maximum allowable load L max .

[0050] S3-1 forms a preliminary human-machine function allocation strategy for the four-party intelligent agent based on the top-level agent perception and decision-making module of HMR; it also monitors the status of the drone group and the operator in real time based on environmental information;

[0051] Among them, see Figure 4 ,Through the top-level Agent situation awareness module, the status data of the UAV swarm and ,operators are collected, and combined with the real-time information of the ,established tasks and task environment, a preliminary human-machine function allocation ,strategy is formed for the system.

[0052] Among them, the types of sensors carried by the drone swarm may include but are not limited to infrared sensors, smoke sensors, gas sensors, optical imagers, multispectral sensors, lidars, anemometers, humidity sensors, pressure sensors, GPS sensors and other types of sensors that can obtain real-time environmental information.

[0053] Wherein, the environment state vector is set to E(k)={e1(k), w2(k), ..., e n (k)}. The environment state vector E(k) will be used as the input of the top-level agent to optimize task allocation and path planning.

[0054] During the execution of the UAV swarm mission, the situational awareness module of the top-level agent integrates the information of the human model and the machine model, and uses the decision-making reasoning module to dynamically formulate and adjust the preliminary allocation strategy of human and machine functions. Based on the top-level agent module, the input is the real-time capability status of the operator c h (k), the performance state vector of the drone group is M(k) = {h(k), v(k), b(k), L(k)}, and the environment state vector is E(k) = {e1(k), e2(k), ..., e n (k)}.

[0055] Use weighted methods to extract features from different types of data and assign different weights according to task requirements:

[0056] F norm (k)=ω1c h，norm (k)+ω2M norm (k)+ω3e norm (k); where F norm (k) represents the normalized comprehensive feature vector, and ω1, ω2, and ω3 are weight factors.

[0057] To address dynamic allocation requirements, we introduce the concept of task priority. This weight quantifies the importance or urgency of tasks, helping the system prioritize more important tasks when allocating them. High-priority tasks are prioritized for allocation to drones with sufficient resources and excellent performance. This ensures that high-priority tasks are completed first even when time and resources are limited. If certain tasks become more urgent due to environmental changes, their priority can be dynamically increased.

[0058] P j =α1F static，j +α2F dynamics，j ; F static，j :Task importance caused by static factors;

[0059] F dynamics，j : Task importance caused by dynamic factors.

[0060] Priority can be adjusted dynamically during task execution

[0061]

[0062] P ij (k) represents task T at time k jPriority of the overall task. j The higher the value, the higher the task priority. j (0): initial priority of the task; γ: dynamic adjustment weight of the time factor; t start : Task generation time; δ: Dynamic adjustment weight of urgency; R j Dynamically changing mission objectives determine their urgency. High-priority missions are assigned to drones with sufficient resources and superior performance, ensuring their completion within time and resource constraints. If certain missions become more urgent due to changing circumstances, their priority can be dynamically increased.

[0063] The eigenvector F norm (k) Input to the decision-making reasoning model module of the top-level agent to calculate the fitness of each task:

[0064]

[0065] Among them, D ij (k) represents task T j With state F norm (k)’s adaptability and its correlation with task priority. The larger the value, the better the UAV M i More suitable for performing task T at time step k j ;T j is the task requirement vector.

[0066] A comprehensive objective function is established using the multi-objective optimization method.

[0067] In order to ensure the rationality of task allocation, it is necessary to combine fitness with constraints. Examples of constraints include:

[0068] (1) Time constraint: Task T j Must be completed within the specified time window:

[0069] (2) Non-repeated assignment: at the same time step k, task T j Can only be executed by the operator or a single drone:

[0070] (3) No over-allocation: UAV M i and operator O's resource usage cannot exceed its capabilities:

[0071] (4) Task requirement constraints: Task requirements must be fully met:

[0072] (5) Leader UAV coordination constraints: The leader UAV must participate in tasks coordinated with the following UAVs:

[0073] (6) Task priority constraint: Tasks need to be completed in order from high priority to low priority:

[0074] Based on D ij (k) and constraints, comprehensive task allocation variables

[0075]

[0076] Dynamically adjust the allocation result to prioritize resources to the adaptability D at each time step k ij (k) the highest priority P ij (k) Higher tasks.

[0077] On this basis, the optimization goal of task allocation can be further defined as:

[0078]

[0079] Furthermore, the comprehensive task adaptation matrix D and the task allocation matrix X are output.

[0080] S3-2 Evaluate the effectiveness and safety of current mission execution;

[0081] The core indicators for evaluating task execution effectiveness include the following: Task completion rate R task ; Task completion time T eff ; Resource utilization: U resource ; Coverage: C coverage

[0082] The core indicators for assessing the safety of mission completion include the following: Mission risk level R j ; Communication stability P loss Drone safety Mission failure rate F task .

[0083] Comprehensive implementation effect: The weight of each indicator is set according to task requirements.

[0084] Comprehensive Security: ρ1, ρ2, ρ3, ρ4: The weights of each indicator, set according to task requirements.

[0085] Combining the overall adaptability, comprehensive execution effect and comprehensive security of the task, while balancing the needs between various goals, the final objective function achieves comprehensive optimization of the system through task allocation.

[0086] The final objective function is as follows:

[0087]

[0088] Based on the evaluation results and control switching rules, S4 reassigns tasks to human-machine functions through the adaptive human-machine system task allocation module and adjusts the specific tasks of each agent.

[0089] Among them, see Figure 5 .

[0090] Abnormal situations that may involve switching of control rights include but are not limited to the following: changes in mission requirements, abnormal drone status, environmental changes, communication status, mission priority adjustments, lack of situational awareness information, human-machine interaction requirements, redundancy strategies, etc.

[0091] Define a collaborative efficiency coefficient η(k) to quantify the degree of cooperation between agents: η(k) = ε1η OM (k)+ε2η LM (k). η(k) will serve as a reference factor for switching decision-making power between humans and machines.

[0092] η OM (k) is the efficiency coefficient of collaboration between the operator and the UAV swarm. The efficiency of collaboration is related to the operator's ability and the performance of the UAV swarm. OM (k) = f(c h (k), M(k)), where f(·) is the mapping function between operator capability, UAV swarm performance and collaborative efficiency. LM (k) is the coordination efficiency coefficient between the leader UAV and the follower UAV. The coordination efficiency is related to the performance of the leader UAV and the follower UAV. LM (k) = f(M L (k), M(k)), where f(·) is the mapping function between the leader UAV performance, the follower UAV performance, and the collaborative efficiency. The weights of the indicators ε1 and ε2 are set according to the mission requirements.

[0093] During the task execution, the operator and the top-level agent continuously monitor the execution status of the task and perform two-way intervention in special circumstances (such as emergencies, changes in task requirements, etc.). When: η(k)>η threshold , give priority to letting the operator participate in decision-making. If the operator is capable enough and determines that there is a risk in the system or the task objectives are not correctly aligned, the operator intervenes. The operator adjusts the execution plan of the current task. After the intervention, the operator feeds back the plan to the top-level agent. The top-level agent re-evaluates and optimizes the task allocation plan based on the current situation awareness data and updates the task allocation matrix X ij (k). When η(k)≤ηthreshold In certain scenarios, the leader drone makes its own decisions, which will not be explained in detail.

[0094] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A task adaptive allocation method for a UAV swarm human-machine system, characterized in that: The following steps are involved: S1 builds the framework of the adaptive human-machine-robot (HMR) task allocation module, clarifying the basic responsibilities of the four intelligent agents including the operator, the top-level agent, the leader drone, and the follower drones; S2 builds a human model for the operator based on cognitive load theory and a machine model for the drone swarm based on flight dynamics theory to evaluate the real-time status of the operator and drones. S3 is based on the top-level Agent perception and decision-making module of HMR to form a preliminary allocation strategy for human-machine functions of the four-party intelligent agent; Real-time monitoring of the status of drone swarms and operators; And combine environmental information to evaluate the performance and safety of the current task; Based on the evaluation results and control switching rules, S4 reassigns tasks to human-machine functions through the adaptive human-machine system task allocation module and adjusts the specific tasks of each agent.

2. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: The operator's responsibilities include providing human supervision and support, reviewing the top-level agent's task planning and allocation, and intervening to adjust and control tasks when necessary. The top-level agent's responsibilities include providing global situational awareness, task planning, and supervision of the entire mission. The leader drone's responsibilities include navigating, commanding, and leading the entire drone swarm. The follower drone's responsibilities include assisting the leader drone and improving the breadth and accuracy of data collection through a distributed approach.

3. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: The human model includes operator capability changes and fatigue performance parameters, which are used to dynamically evaluate the operator's status during task allocation.

4. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: The drone swarm includes a leader drone and a follower drone, which establishes a drone model based on the performance conditions of the drones; the drone model includes the flight status and flight performance parameters of the drones, which are used to dynamically evaluate the drones' mission execution capabilities during mission execution.

5. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: The top-level agent system includes a situational awareness module and a decision-making reasoning module; the situational awareness module includes a target recognition agent, a state prediction agent, an effect estimation agent, and a task safety estimation agent. The agents achieve real-time state awareness of the drone swarm through information exchange; the situational awareness module hardware equipment includes various sensors and detectors such as image recognition and infrared perception, which are used to acquire image data; the decision-making reasoning agent in the decision-making reasoning module receives situation information from the situational awareness module, analyzes the situation in the task area, determines the priority of the task, and executes adaptive task allocation and adjustment strategies based on the real-time state of the drone swarm and task execution feedback to ensure the real-time nature of task allocation and the coverage integrity of monitoring data.

6. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: The top-level agent monitors the status feedback of the operator and the UAV during the task execution through the HMR task allocation module, and dynamically adjusts the task allocation to meet the real-time task requirements; The specific steps include: collecting operator workload and fatigue-related performance data, and evaluating the operator's status through a human model; collecting the UAV's flight status and flight performance parameters, and evaluating its performance status through a UAV model; combining real-time environmental data to form "human-machine-environment" three-state information. The top-level agent adjusts the task allocation strategy in real time, assigning the appropriate task to the operator or UAV in the best condition, thereby improving the overall completion effect of the target task.

7. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: When allocating tasks, the HMR task allocation module performs multi-objective optimization based on multiple constraints such as drone swarm performance, operator performance, environmental information, and task information, with the overall adaptability, comprehensive execution effect, and comprehensive safety of the tasks as the goals, to maximize the efficiency of drone swarm task execution.

8. The method for adaptive task allocation for a drone swarm human-machine system according to claim 1, characterized in that: Throughout the entire mission, the operator and the top-level agent can conduct two-way supervision. The operator can monitor the top-level agent's decision-making and execution process in real time, and proactively intervene when deviations or potential risks are discovered, providing the top-level agent with decision-making adjustment suggestions to ensure the rationality and safety of the mission and the efficient achievement of mission goals.

Images