Heterogeneous unmanned aerial vehicle cooperative task allocation method and system based on genetic algorithm

Abstract

The invention discloses a heterogeneous unmanned aerial vehicle cooperative task allocation method and system based on a genetic algorithm, and belongs to the technical field of heterogeneous unmanned aerial vehicle cooperative task allocation, and the method comprises the steps: dividing a target region into sub-regions through k-means clustering, generating a first execution strategy of a heterogeneous unmanned aerial vehicle in each sub-region through an improved genetic algorithm, generating a second execution strategy of each sub-region heterogeneous unmanned aerial vehicle through a deep reinforcement learning network, determining a third execution strategy of each sub-region heterogeneous unmanned aerial vehicle based on existence information of each sub-region high-frequency task, analyzing a sorting penalty factor of each task, and determining a fourth execution strategy of each sub-region heterogeneous unmanned aerial vehicle; and determining a fifth execution strategy of the heterogeneous unmanned aerial vehicles in the sub-regions. According to the method, the total voyage of the system is effectively reduced, the task scheduling efficiency is improved, the dynamic balance of the task resource load of the unmanned aerial vehicle is realized, meanwhile, the suppression of interference between tasks and the intelligent coordination of resource conflicts are realized, and the task completion stability and the scheduling intelligence are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of collaborative task allocation for heterogeneous unmanned aerial vehicles (UAVs), and in particular to a method and system for collaborative task allocation for heterogeneous UAVs based on a genetic algorithm. Background Art

[0002] With the rapid development of unmanned systems technology, drones are increasingly being used in military reconnaissance, disaster monitoring, logistics and transportation, environmental inspection, and other fields. In particular, in multi-task, long-flight, and complex environment scenarios, a single drone often cannot meet the overall mission requirements, thus driving the research and application of multi-drone collaborative operations. In practical applications, multiple drones often need to collaborate to complete multiple tasks with widely distributed spatial distribution and heterogeneous execution constraints. To improve overall system efficiency and mission completion, task allocation and path planning must be rationally implemented based on factors such as mission requirements, drone performance, geographic information, and time constraints. Compared to homogeneous systems, heterogeneous drone systems are composed of drones of various types and capabilities, with differences in flight speed, sensor type, payload capacity, and battery capacity, making the task scheduling process more complex.

[0003] To address large-scale, heterogeneous, and multi-constrained task allocation, intelligent optimization algorithms have been widely used in UAV task allocation due to their strong adaptability and global search capabilities. Among them, genetic algorithms, a classic evolutionary computation method, simulate natural selection and genetic mechanisms to obtain near-optimal solutions to complex combinatorial optimization problems. They have been widely used in various scenarios, including path optimization, multi-objective scheduling, and resource allocation.

[0004] For example, the invention patent announcement with announcement number: CN114815896B discloses a method for collaborative task allocation for heterogeneous multi-UAVs, including: step one: establishing a hierarchical heterogeneous multi-UAV collaborative task allocation model; step two: heterogeneous multi-UAVs collaboratively executing the assigned tasks; step three: establishing a hierarchical heterogeneous multi-UAV collaborative task allocation cost function; step four: initializing a quantum wasp swarm and setting parameters; step five: defining and calculating the distance between the quantum wasp and the food; step six: sorting all quantum wasps according to the distance between the quantum wasp and the food; step seven: the quantum wasp executes deterministic or random flight motion with equal probability, and uses a simulated quantum revolving gate to evolve the quantum position of the quantum wasp during the flight; step eight: applying a greedy selection strategy to determine the quantum position of the next generation of quantum wasps; step nine: judging the termination of evolution and outputting a task allocation plan.

[0005] For example, the invention patent announcement with announcement number: CN111399533B discloses a method for collaborative task allocation and path optimization for heterogeneous multi-UAVs. The method may include: first determining relevant information about UAVs in the target area, relevant information about UAV sites, and relevant information about target objects that need to be acquired using UAVs, and then calculating the Euclidean distance from the UAV site to all target objects and the Euclidean distance between all target objects. Then, a HU-TAP-VP model for the variable-reward task allocation problem of heterogeneous UAVs is established to obtain an initial set of task allocation plans for executing collaborative tasks. Finally, a hybrid genetic simulated annealing algorithm HGSA with an adaptive switching mechanism is used for optimization to obtain a safe and flyable path for each UAV.

[0006] However, in the process of implementing the technical solutions of the invention in the embodiments of the present application, the present application found that the above technology has at least the following technical problems:

[0007] Current collaborative task allocation methods for heterogeneous UAVs mostly focus on task path optimization, algorithm convergence performance improvement, or static resource matching accuracy. For example, the optimal task-path combination is achieved through evolutionary algorithms, hybrid intelligent agent mechanisms, or multi-objective optimization models. However, in the actual task execution process, certain tasks are frequently selected among multiple UAVs due to factors such as their location, type, or priority, and UAVs usually independently adjust the task execution order based on their respective local strategies. Existing technologies do not analyze the consistency of task priorities between different UAVs, resulting in unaware sorting conflicts. The system may encounter problems such as drones repeatedly visiting the same area, serious path overlap, and task congestion, which can easily lead to severe conflicts in task execution plans between UAVs, resulting in an increase in task execution failure rate and a decrease in resource utilization efficiency. Summary of the Invention

[0008] A first aspect of the present invention provides a method for collaborative task allocation of heterogeneous UAVs based on a genetic algorithm, comprising the following steps:

[0009] The heterogeneous UAV mission execution area is recorded as the target area, which is divided into sub-areas by k-means clustering, and the first execution strategy of heterogeneous UAVs in each sub-area is generated by an improved genetic algorithm.

[0010] Based on the first task execution strategy of the heterogeneous UAVs in each sub-area, the tasks are executed and the task information is shared in real time. The second execution strategy of the heterogeneous UAVs in each sub-area is generated through the deep reinforcement learning network. The existence information of high-frequency tasks in each sub-area is simultaneously analyzed. Based on the existence information of high-frequency tasks in each sub-area, the third execution strategy of the heterogeneous UAVs in each sub-area is determined.

[0011] When the third execution strategy of the sub-region heterogeneous UAV is task reanalysis, the ranking penalty factor of each task of the heterogeneous UAV is analyzed, and the fourth execution strategy of the sub-region heterogeneous UAV is determined in combination with the corresponding second execution strategy of the heterogeneous UAV.

[0012] Based on the fourth execution strategy of sub-region heterogeneous UAVs, sub-region task execution conflicts are determined, and the first execution strategy of task allocation is determined. When the first execution strategy of task allocation is to analyze and adapt UAVs, each hotspot task is extracted, and the heterogeneous UAV adaptation index of each hotspot task is analyzed to determine the fifth execution strategy of sub-region heterogeneous UAVs.

[0013] Furthermore, the first execution strategy of heterogeneous drones in each sub-region is generated. The specific analysis method is as follows:

[0014] Obtain a set of tasks to be assigned, extract spatial features from the set of tasks to be assigned, and divide the tasks into several spatial task sub-regions based on the geographical location information of the tasks using the k-means clustering algorithm, which are recorded as sub-regions.

[0015] Each sub-region is treated as an independent optimization unit, and a sub-task description vector including task number, task location coordinates, and resource requirement information is constructed, thereby obtaining the sub-task description vector of each sub-region statistically.

[0016] For each sub-region, a dual-objective fitness function is constructed with the goals of minimizing the total flight range of UAVs and minimizing the standard deviation of task resource load. The location coordinates, resource requirements and time constraints contained in the description vector of each task are used as input features for the individual construction and fitness calculation of the genetic algorithm. By combining the improved genetic algorithm with elite retention, roulette wheel selection and dynamic crossover mutation, the optimal initial task allocation and path execution plan for the heterogeneous UAVs in each sub-region is generated, and it is recorded as the first execution strategy for the heterogeneous UAVs in each sub-region.

[0017] Furthermore, the second execution strategy of the heterogeneous drones in each sub-region is generated through the deep reinforcement learning network. The specific analysis process is as follows:

[0018] Task information includes task priority, task time window, task location information, and task completion status.

[0019] The task information is input into the pre-trained deep reinforcement learning network to generate task execution sequence adjustment decisions for heterogeneous drones in each sub-area.

[0020] According to the task execution order adjustment decision of the heterogeneous UAVs in each sub-area, the priorities of the currently assigned tasks of the heterogeneous UAVs in each sub-area are dynamically adjusted to form an updated task priority list of the heterogeneous UAVs in each sub-area, which is recorded as the second execution strategy of the heterogeneous UAVs in each sub-area.

[0021] Furthermore, based on the existence information of high-frequency tasks in each sub-area, the third execution strategy of the heterogeneous UAVs in each sub-area is determined. The specific process is as follows:

[0022] Extract the preset task execution count threshold in the database.

[0023] The number of times each task is executed is obtained based on the preset monitoring period.

[0024] The tasks whose execution times are less than the task execution times threshold are recorded as low-frequency tasks.

[0025] Tasks whose execution times are greater than or equal to the task execution times threshold are recorded as high-frequency tasks.

[0026] Traverse each task in the sub-area. If there is a high-frequency task, record the existence of the high-frequency task in the sub-area as the existence of a high-frequency task. Simultaneously, record the third execution strategy of the heterogeneous UAV in the sub-area as a task for re-analysis.

[0027] If all tasks in the sub-area are low-frequency tasks, the existence of high-frequency tasks in the sub-area is recorded as the absence of high-frequency tasks, and the third execution strategy of the heterogeneous drone in the sub-area is simultaneously recorded as directly executing the second execution strategy of the heterogeneous drone in the sub-area.

[0028] Traverse each sub-area, and determine the third execution strategy of the heterogeneous UAVs in each sub-area based on the existence information of high-frequency tasks in each sub-area.

[0029] Furthermore, we analyze the ranking penalty factors of each task of heterogeneous UAVs. The specific analysis process is as follows:

[0030] During the preset monitoring period, the status information of each task is collected, including the path overlap rate, resource cache hit rate and urgency of each task.

[0031] Analyze the sorting penalty factor of each task based on the status information of each task.

[0032] The ranking penalty factor of each task is the quantitative data of the degree of influence of the path overlap rate, resource cache hit rate and urgency of each task on the execution constraints and interference of the task. The specific analysis process is as follows: the reference values ​​of the resource cache hit rate and urgency are compared with the resource cache hit rate and urgency of the corresponding tasks, the reference value of the path overlap rate is compared with the path overlap rate of each task, and the comparison results are coupled with the corresponding allocation factors to obtain the ranking penalty factor of each task.

[0033] Furthermore, the fourth execution strategy for heterogeneous UAVs in the sub-region is determined. The specific analysis process is as follows:

[0034] A dynamic task scheduling optimization network based on deep reinforcement learning is constructed, which takes the second execution strategy of heterogeneous UAVs in the corresponding sub-region and the ranking penalty factor of each task in the corresponding sub-region as joint input, and outputs the task priority adjustment result of the corresponding sub-region.

[0035] The result of adjusting the task priority of the sub-region is recorded as the fourth execution strategy of the heterogeneous UAVs in the sub-region.

[0036] The sorting penalty factor is used to quantify the potential interference risk during the task execution order adjustment process.

[0037] Furthermore, the first execution strategy of task allocation is determined. The specific analysis process is as follows:

[0038] According to the fourth execution strategy of the heterogeneous drones in the sub-region, the heat count of each task is obtained.

[0039] Extract the preset heat count threshold in the database.

[0040] Tasks whose heat count is less than the heat count threshold are recorded as low-heat tasks.

[0041] Tasks whose heat count is greater than or equal to the heat count threshold are recorded as hot tasks.

[0042] Traverse each task in the sub-area. If there is a hot task, the first execution strategy of the task assignment is recorded as the analysis and adaptation drone.

[0043] If all tasks are low-heat tasks, the first execution strategy of task assignment is recorded as the fourth execution strategy of directly executing the heterogeneous drones in the sub-region.

[0044] Furthermore, we analyze the adaptation index of heterogeneous drones for each hotspot task. The specific analysis steps are as follows:

[0045] Obtain heterogeneous drone adaptation data for hotspot tasks, including each drone's distance, remaining resources, and historical success rate.

[0046] The adaptation index of heterogeneous UAVs for hot tasks is analyzed based on the adaptation data of heterogeneous UAVs for hot tasks.

[0047] The heterogeneous UAV adaptation index for hotspot missions is a quantitative indicator of the impact of the distance, remaining resources, and historical success rate of each UAV on the degree of adaptation between the hotspot mission and each UAV. The specific analysis process is as follows: the reference value of the distance is compared with the distance of each UAV, the remaining resources and historical success rate of each UAV are compared with the corresponding reference values, and the comparison results are coupled with the corresponding allocation coefficient to obtain the heterogeneous UAV adaptation index for hotspot missions.

[0048] Traverse each hotspot task to obtain the heterogeneous drone adaptation index of each hotspot task.

[0049] Furthermore, the fifth execution strategy for heterogeneous UAVs in the sub-region is determined. The specific analysis steps are as follows:

[0050] A1, the drone corresponding to the maximum value of the heterogeneous drone adaptation index of the hotspot task is used as the execution drone of the corresponding hotspot task.

[0051] A2, traverses each hotspot task and obtains the executing drone of each hotspot task.

[0052] If the executing drones of each hotspot task do not overlap, the drone execution confirmation information of each hotspot task is generated.

[0053] If there are overlapping execution drones, the hotspot tasks corresponding to the overlapping execution drones are obtained and recorded as the hotspot tasks to be confirmed.

[0054] Obtain the second execution strategy for overlapping drones, and use the hotspot task with the highest priority among all the hotspot tasks to be confirmed as the execution task of the overlapping drone, generate drone execution confirmation information for the hotspot task, and record the remaining tasks in all the hotspot tasks to be confirmed as the second-confirmed hotspot tasks.

[0055] A3, for each secondary confirmation hotspot task, re-obtain the drone corresponding to the maximum value of the heterogeneous drone adaptation index and execute A2 until the corresponding execution drone is determined, and generate the drone execution confirmation information of the corresponding hotspot task.

[0056] A4 receives the drone execution confirmation information of each hotspot task and removes the corresponding hotspot task from the remaining drone task priority list, thereby obtaining the updated task priority list of heterogeneous drones and recording it as the fifth execution strategy of the sub-region heterogeneous drones.

[0057] A second aspect of the present invention provides a heterogeneous UAV collaborative task allocation system based on a genetic algorithm, comprising:

[0058] The first execution strategy generation module is used to record the heterogeneous UAV task execution area as the target area, divide the target area into sub-areas through k-means clustering, and generate the first execution strategy of the heterogeneous UAV in each sub-area through an improved genetic algorithm.

[0059] The third execution strategy determination module is used to execute tasks based on the first task execution strategy of the heterogeneous UAVs in each sub-area, share task information in real time, generate the second execution strategy of the heterogeneous UAVs in each sub-area through the deep reinforcement learning network, and simultaneously analyze the existence information of high-frequency tasks in each sub-area. Based on the existence information of high-frequency tasks in each sub-area, the third execution strategy of the heterogeneous UAVs in each sub-area is determined.

[0060] The fourth execution strategy determination module is used to analyze the sorting penalty factors of each task of the heterogeneous drone in the sub-region when the third execution strategy of the heterogeneous drone in the sub-region is task reanalysis, and determine the fourth execution strategy of the heterogeneous drone in the sub-region in combination with the second execution strategy of the corresponding heterogeneous drone.

[0061] The fifth execution strategy determination module is used to determine the sub-area task execution conflict based on the fourth execution strategy of the sub-area heterogeneous UAV, determine the first execution strategy for task assignment, and when the first execution strategy for task assignment is to analyze and adapt the UAV, extract each hotspot task, analyze the heterogeneous UAV adaptation index of each hotspot task, and determine the fifth execution strategy for the sub-area heterogeneous UAV.

[0062] One or more technical solutions provided in the present invention have at least the following technical effects or advantages:

[0063] 1. The genetic algorithm-based collaborative task allocation method for heterogeneous UAVs provided by the present invention can achieve efficient scheduling of various types of heterogeneous UAVs in large-scale, multi-type mission scenarios. By introducing task area division, dual-objective genetic optimization and reinforcement learning dynamic priority adjustment mechanism, it not only effectively reduces the total system range and improves task scheduling efficiency, but also achieves dynamic balancing of UAV task resource loads; further, based on high-frequency task identification, sorting penalty factor calculation and hot task allocation conflict judgment mechanism during task execution, it achieves the suppression of interference between tasks and intelligent coordination of resource conflicts, thereby improving the stability of task completion and the intelligence of scheduling.

[0064] 2. The present invention determines the third execution strategy of heterogeneous drones in each sub-area based on the existence information of high-frequency tasks in each sub-area. It can dynamically identify highly repetitive tasks that frequently appear during task execution, and optimize the drone task scheduling decision logic based on task heat, resource utilization and path overlap to avoid repeated task execution or resource waste. By introducing high-frequency task identification results to guide scheduling behavior, the system's ability to judge task priority and scheduling response speed are effectively improved, and the system's resource scheduling accuracy and overall task completion efficiency in task-intensive scenarios are enhanced, thereby achieving more efficient and stable collaborative task execution effects.

[0065] 3. The present invention determines the fifth execution strategy for heterogeneous drones in sub-regions by analyzing the adaptation index of heterogeneous drones for each hotspot task. In scenarios where hotspot tasks are concentrated or the risk of resource conflicts is high, it can accurately evaluate the matching degree and execution capability of different types of drones for hotspot tasks, thereby achieving a more reasonable adaptation allocation between tasks and drones. This method effectively avoids the execution conflicts and performance degradation caused by the repeated selection of high-hot tasks by unsuitable drones, improves the coordination of system task scheduling, resource utilization and task completion rate, and has higher scheduling intelligence and task allocation stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 Flowchart of the heterogeneous UAV collaborative task allocation method based on genetic algorithm provided in an embodiment of the present application.

[0067] Figure 2 A schematic diagram of the structure of a heterogeneous UAV collaborative task allocation system based on a genetic algorithm provided in an embodiment of the present application. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0069] Reference Figure 1 As shown, the first aspect of the present invention provides a method for heterogeneous UAV collaborative task allocation based on a genetic algorithm, comprising the following steps:

[0070] It should be noted that "heterogeneous UAVs" refers to the combination of UAVs of multiple types, with different performance and functions, in one system or mission. These UAVs differ in platform structure, functional configuration, flight capability, sensor load, communication method, energy system, etc.

[0071] The heterogeneous UAV mission execution area is recorded as the target area, which is divided into sub-areas by k-means clustering, and the first execution strategy of heterogeneous UAVs in each sub-area is generated by an improved genetic algorithm.

[0072] In this embodiment, the first execution strategy of the heterogeneous drones in each sub-region is generated. The specific analysis method is as follows:

[0073] Obtain a set of tasks to be assigned, extract spatial features from the set of tasks to be assigned, and divide the tasks into several spatial task sub-regions based on the geographical location information of the tasks using the k-means clustering algorithm, which are recorded as sub-regions.

[0074] It should be noted that the geographical location information of the task can be directly extracted from the database.

[0075] Each sub-region is treated as an independent optimization unit, and a sub-task description vector including task number, task location coordinates, and resource requirement information is constructed, thereby obtaining the sub-task description vector of each sub-region statistically.

[0076] For each sub-region, a dual-objective fitness function is constructed with the goals of minimizing the total flight range of UAVs and minimizing the standard deviation of task resource load. The location coordinates, resource requirements and time constraints contained in the description vector of each task are used as input features for the individual construction and fitness calculation of the genetic algorithm. By combining the improved genetic algorithm with elite retention, roulette wheel selection and dynamic crossover mutation, the optimal initial task allocation and path execution plan for the heterogeneous UAVs in each sub-region is generated, and it is recorded as the first execution strategy for the heterogeneous UAVs in each sub-region.

[0077] It's important to note that fitness calculations are based on the fitness function, which is key to evaluating individual performance. The optimization goal is to maximize its fitness. If an individual satisfies all constraints, the corresponding fitness is calculated based on the objective function; otherwise, the fitness is set to its minimum value, and individuals that do not meet the constraints are discarded. Therefore, the initial population base has a significant impact on our optimization results. We strive to maximize the population base to achieve optimal results. To facilitate subsequent calculations, fitness values ​​are typically normalized (so that their sum is 1). Next, we construct a cumulative probability table, which accumulates the selection probabilities of each individual in sequence for subsequent random number sampling. Finally, to select individuals, we perform the following steps: randomly generate a number r in the interval [0, 1], find an individual that meets the conditions, and select it as the selected individual. Repeat the above steps until the desired number of individuals is selected.

[0078] It should also be added that the above-mentioned conditions are individual selection conditions pre-set in the database.

[0079] Based on the first task execution strategy of the heterogeneous UAVs in each sub-area, the tasks are executed and the task information is shared in real time. The second execution strategy of the heterogeneous UAVs in each sub-area is generated through the deep reinforcement learning network. The existence information of high-frequency tasks in each sub-area is simultaneously analyzed. Based on the existence information of high-frequency tasks in each sub-area, the third execution strategy of the heterogeneous UAVs in each sub-area is determined.

[0080] In this embodiment, the second execution strategy of the heterogeneous drones in each sub-region is generated through a deep reinforcement learning network. The specific analysis process is as follows:

[0081] Task information includes task priority, task time window, task location information, and task completion status.

[0082] It should be added that task information can be directly extracted from the database.

[0083] The task information is input into the pre-trained deep reinforcement learning network to generate task execution sequence adjustment decisions for heterogeneous drones in each sub-area.

[0084] According to the task execution order adjustment decision of the heterogeneous UAVs in each sub-area, the priorities of the currently assigned tasks of the heterogeneous UAVs in each sub-area are dynamically adjusted to form an updated task priority list of the heterogeneous UAVs in each sub-area, which is recorded as the second execution strategy of the heterogeneous UAVs in each sub-area.

[0085] In this embodiment, the third execution strategy of the heterogeneous UAVs in each sub-area is determined based on the high-frequency task existence information of each sub-area. The specific process is as follows:

[0086] Extract the preset task execution count threshold in the database.

[0087] The number of times each task is executed is obtained based on the preset monitoring period.

[0088] It should be noted that the number of task executions can be collected through system program logs.

[0089] The tasks whose execution times are less than the task execution times threshold are recorded as low-frequency tasks.

[0090] If the number of executions of a task is less than the threshold, it means that the frequency of the task is low and the demand for the task is low.

[0091] Tasks whose execution times are greater than or equal to the task execution times threshold are recorded as high-frequency tasks.

[0092] If the number of executions of a task is greater than or equal to the threshold, it means that the task occurs frequently and has a high usage frequency. It may be a routine task or a critical task.

[0093] Traverse each task in the sub-area. If there is a high-frequency task, record the existence of the high-frequency task in the sub-area as the existence of a high-frequency task. Simultaneously, record the third execution strategy of the heterogeneous UAV in the sub-area as a task for re-analysis.

[0094] If there are high-frequency tasks, it means that there are important and frequently occurring task types in the area. In order to ensure the accuracy and efficiency of task scheduling, it is necessary to further analyze the tasks in the area and improve the accuracy of task allocation and scheduling.

[0095] If all tasks in the sub-area are low-frequency tasks, the existence of high-frequency tasks in the sub-area is recorded as the absence of high-frequency tasks, and the third execution strategy of the heterogeneous drone in the sub-area is simultaneously recorded as directly executing the second execution strategy of the heterogeneous drone in the sub-area.

[0096] It should be noted that if all tasks in a sub-region are low-frequency tasks, it means that the overall importance of tasks in the sub-region is low and they do not occur frequently. Therefore, the existing second execution strategy can be directly used without further analysis, saving computing and execution resources.

[0097] Traverse each sub-area, and determine the third execution strategy of the heterogeneous UAVs in each sub-area based on the existence information of high-frequency tasks in each sub-area.

[0098] When the third execution strategy of the sub-region heterogeneous UAV is task reanalysis, the ranking penalty factor of each task of the heterogeneous UAV is analyzed, and the fourth execution strategy of the sub-region heterogeneous UAV is determined in combination with the corresponding second execution strategy of the heterogeneous UAV.

[0099] In this embodiment, the ranking penalty factors of each task of heterogeneous drones are analyzed. The specific analysis process is as follows:

[0100] During the preset monitoring period, the status information of each task is collected, including the path overlap rate, resource cache hit rate and urgency of each task.

[0101] It should be noted that the path overlap rate of a task refers to the degree of overlap between the current task path and the paths of other executed tasks in the sub-area within the preset monitoring period. The resource cache hit rate refers to whether the resources required during task execution (such as map fragments, image data, navigation information, etc.) already exist in the local cache. If a resource request for a task can be directly hit from the cache without reloading from the remote end, it is called a "hit". Within the preset monitoring period, the ratio of the number of task hits to the total number of requests is used as the numerical result of the resource cache hit rate. The remaining execution time of the task at the end of the monitoring period is used as the numerical result of the urgency.

[0102] It should be added that the status information of each task can be collected from the program log.

[0103] The reference path overlap rate, reference resource cache hit rate and reference urgency stored in the database are extracted.

[0104] Extract the path overlap rate allocation factor, resource cache hit rate allocation factor and urgency allocation factor preset in the database.

[0105] It should be noted that the path overlap rate allocation factor, resource cache hit rate allocation factor and urgency allocation factor all have value ranges between 0 and 1, and the sum of the path overlap rate allocation factor, resource cache hit rate allocation factor and urgency allocation factor is 1. When used, the pre-set values ​​can be directly extracted from the database. The specific extraction method is, for example, to construct a one-to-one mapping set with the path overlap rate, resource cache hit rate and urgency respectively and the corresponding path overlap rate allocation factor, resource cache hit rate allocation factor and urgency allocation factor. When used, the obtained path overlap rate, resource cache hit rate and urgency are respectively input into the corresponding mapping set, thereby extracting the path overlap rate allocation factor, resource cache hit rate allocation factor and urgency allocation factor.

[0106] Analyze the sorting penalty factor of each task based on the status information of each task.

[0107] The ranking penalty factor of each task is the quantitative data of the degree of influence of the path overlap rate, resource cache hit rate and urgency of each task on the execution constraints and interference of the task. The specific analysis process is as follows: the reference values ​​of the resource cache hit rate and urgency are compared with the resource cache hit rate and urgency of the corresponding tasks, the reference value of the path overlap rate is compared with the path overlap rate of each task, and the comparison results are coupled with the corresponding allocation factors to obtain the ranking penalty factor of each task.

[0108] In a specific embodiment, the ranking penalty factor of each task is specifically expressed as follows:

[0109]

[0110] Among them, A i is the ranking penalty factor of the i-th task, a i is the path overlap rate of the i-th task, b i is the resource cache hit rate of the i-th task, c i is the urgency of the i-th task, a vef is the reference path overlap ratio, b vef is the reference resource cache hit rate, c vef is the reference urgency, α1 is the path overlap rate allocation factor, α2 is the resource cache hit rate allocation factor, α3 is the urgency allocation factor, i is the task number, i = 1, 2, ..., m, m is the number of tasks.

[0111] In this embodiment, the fourth execution strategy for heterogeneous UAVs in the sub-region is determined. The specific analysis process is as follows:

[0112] A dynamic task scheduling optimization network based on deep reinforcement learning is constructed, which takes the second execution strategy of heterogeneous UAVs in the corresponding sub-region and the ranking penalty factor of each task in the corresponding sub-region as joint input, and outputs the task priority adjustment result of the corresponding sub-region.

[0113] The result of adjusting the task priority of the sub-region is recorded as the fourth execution strategy of the heterogeneous UAVs in the sub-region.

[0114] The sorting penalty factor is used to quantify the potential interference risk during the task execution order adjustment process.

[0115] Based on the fourth execution strategy of sub-region heterogeneous UAVs, sub-region task execution conflicts are determined, and the first execution strategy of task allocation is determined. When the first execution strategy of task allocation is to analyze and adapt UAVs, each hotspot task is extracted, and the heterogeneous UAV adaptation index of each hotspot task is analyzed to determine the fifth execution strategy of sub-region heterogeneous UAVs.

[0116] In this embodiment, the first execution strategy of task allocation is determined, and the specific analysis process is as follows:

[0117] According to the fourth execution strategy of the heterogeneous drones in the sub-region, the heat count of each task is obtained.

[0118] It should be explained that the heat count refers to the number of times the task ranks in the top k in the priority list of all drones, where k is a preset value in the database.

[0119] Extract the preset heat count threshold in the database.

[0120] Tasks whose heat count is less than the heat count threshold are recorded as low-heat tasks.

[0121] If the heat count is less than the heat count threshold, it means that the competition degree of this type of task is small in the current scheduling cycle and the probability of task conflict is low.

[0122] Tasks whose heat count is greater than or equal to the heat count threshold are recorded as hot tasks.

[0123] If the heat count is greater than or equal to the heat count threshold, it means that it has high importance or prevalence in the current mission environment, and multiple drones may attempt to perform the mission at the same time, thus creating the risk of resource preemption and execution conflict.

[0124] Traverse each task in the sub-area. If there is a hot task, the first execution strategy of the task assignment is recorded as the analysis and adaptation drone.

[0125] If there are hotspot tasks, this indicates that the current task distribution has high-attention and high-overlap scheduling hotspots, which may trigger competition for the same task among multiple heterogeneous drones. In this case, the first execution strategy for task allocation is recorded as analyzing and adapting drones. This requires further intelligent adaptation based on information such as the resource capabilities, task conflict probability, and penalty factors of each drone to develop a conflict avoidance optimization strategy, achieving dynamic adjustment and conflict suppression for collaborative scheduling.

[0126] If all tasks are low-heat tasks, the first execution strategy of task assignment is recorded as the fourth execution strategy of directly executing the heterogeneous drones in the sub-region.

[0127] If all tasks are low-heat tasks, this indicates low overlap between tasks under the current scheduling state, decentralized task allocation, low resource competition, and negligible conflict risk. In this case, the previously generated first execution strategy for task allocation can be directly adopted as the fourth execution strategy for the heterogeneous sub-region UAVs. This allows for stable and efficient task execution without further analysis or adaptation, improving scheduling response efficiency and system throughput.

[0128] In this embodiment, the heterogeneous UAV adaptation index of each hotspot task is analyzed. The specific analysis steps are as follows:

[0129] Obtain heterogeneous drone adaptation data for hotspot tasks, including each drone's distance, remaining resources, and historical success rate.

[0130] It should be noted that the distance between each drone's current location and the starting point of the corresponding hotspot mission is the distance. The remaining battery power of each drone is used as the numerical value of the remaining resources. The historical success rate refers to the ratio of the number of successful executions of each drone for the corresponding hotspot mission to the total number of successful executions.

[0131] It should also be noted that heterogeneous drone adaptation data can be obtained from the program log.

[0132] Extract the reference distance, reference remaining resource and reference historical success rate stored in the database.

[0133] Extract the preset distance allocation coefficient, remaining resource allocation coefficient and historical success rate allocation coefficient in the database.

[0134] It should be noted that the distance allocation coefficient, the remaining resource allocation coefficient and the historical success rate allocation coefficient all have value ranges between 0 and 1, and the sum of the distance allocation coefficient, the remaining resource allocation coefficient and the historical success rate allocation coefficient is 1. When used, the pre-set value can be directly extracted from the database. The specific extraction method is, for example, to construct a one-to-one mapping set of distance, remaining resource and historical success rate with the corresponding distance allocation coefficient, remaining resource allocation coefficient and historical success rate allocation coefficient respectively. When used, the obtained distance, remaining resource and historical success rate are respectively input into the corresponding mapping set to extract the distance allocation coefficient, remaining resource allocation coefficient and historical success rate allocation coefficient.

[0135] The adaptation index of heterogeneous UAVs for hot tasks is analyzed based on the adaptation data of heterogeneous UAVs for hot tasks.

[0136] The heterogeneous UAV adaptation index for hotspot missions is a quantitative indicator of the impact of the distance, remaining resources, and historical success rate of each UAV on the degree of adaptation between the hotspot mission and each UAV. The specific analysis process is as follows: the reference value of the distance is compared with the distance of each UAV, the remaining resources and historical success rate of each UAV are compared with the corresponding reference values, and the comparison results are coupled with the corresponding allocation coefficient to obtain the heterogeneous UAV adaptation index for hotspot missions.

[0137] In a specific embodiment, the heterogeneous UAV adaptation index of a hotspot task is specifically expressed as follows:

[0138]

[0139] Among them, B is the heterogeneous UAV adaptation index of hotspot tasks, d is the UAV distance, f is the remaining UAV resources, g is the UAV historical success rate, d vef is the reference distance, f vef is the reference remaining resource, g vef is the reference historical success rate, β1 is the distance allocation coefficient, β2 is the remaining resource allocation coefficient, and β3 is the historical success rate allocation coefficient.

[0140] Traverse each hotspot task to obtain the heterogeneous drone adaptation index of each hotspot task.

[0141] In this embodiment, the fifth execution strategy for heterogeneous UAVs in the sub-region is determined. The specific analysis steps are as follows:

[0142] A1, the drone corresponding to the maximum value of the heterogeneous drone adaptation index of the hotspot task is used as the execution drone of the corresponding hotspot task.

[0143] A2, traverses each hotspot task and obtains the executing drone of each hotspot task.

[0144] If the executing drones of each hotspot task do not overlap, the drone execution confirmation information of each hotspot task is generated.

[0145] If the drones executing each hotspot task do not overlap, meaning each hotspot task is executed by a different drone, this indicates that the system has achieved a preliminary match between task decentralization and resource independence at the hotspot task scheduling level, and there is no direct competition for execution resources between drones. At this point, drone execution confirmation information can be directly generated for each hotspot task to clarify the execution entity of each task within the current scheduling cycle, ensuring the stability and consistency of the task execution plan.

[0146] If there are overlapping execution drones, the hotspot tasks corresponding to the overlapping execution drones are obtained and recorded as the hotspot tasks to be confirmed.

[0147] If there are overlapping drones, meaning the same drone is assigned to two or more hotspot tasks, this indicates a task aggregation conflict or resource allocation competition in the system scheduling, putting the drone at risk of task execution order conflicts or resource shortages. In this case, all hotspot tasks associated with the overlapping drone should be extracted, marked as pending hotspot tasks, and then undergo a priority readjustment or resource reallocation analysis process to alleviate scheduling pressure and improve overall task completion rates.

[0148] Obtain the second execution strategy for overlapping drones, and use the hotspot task with the highest priority among all the hotspot tasks to be confirmed as the execution task of the overlapping drone, generate drone execution confirmation information for the hotspot task, and record the remaining tasks in all the hotspot tasks to be confirmed as the second-confirmed hotspot tasks.

[0149] A3, for each secondary confirmation hotspot task, re-obtain the drone corresponding to the maximum value of the heterogeneous drone adaptation index and execute A2 until the corresponding execution drone is determined, and generate the drone execution confirmation information of the corresponding hotspot task.

[0150] A4 receives the drone execution confirmation information of each hotspot task and removes the corresponding hotspot task from the remaining drone task priority list, thereby obtaining the updated task priority list of heterogeneous drones and recording it as the fifth execution strategy of the sub-region heterogeneous drones.

[0151] See Figure 2 As shown, the second aspect of the present invention provides a heterogeneous UAV collaborative task allocation system based on a genetic algorithm, comprising:

[0152] The first execution strategy generation module is used to record the heterogeneous UAV task execution area as the target area, divide the target area into sub-areas through k-means clustering, and generate the first execution strategy of the heterogeneous UAV in each sub-area through an improved genetic algorithm.

[0153] The third execution strategy determination module is used to execute tasks based on the first task execution strategy of the heterogeneous UAVs in each sub-area, share task information in real time, generate the second execution strategy of the heterogeneous UAVs in each sub-area through the deep reinforcement learning network, and simultaneously analyze the existence information of high-frequency tasks in each sub-area. Based on the existence information of high-frequency tasks in each sub-area, the third execution strategy of the heterogeneous UAVs in each sub-area is determined.

[0154] The fourth execution strategy determination module is used to analyze the sorting penalty factors of each task of the heterogeneous drone in the sub-region when the third execution strategy of the heterogeneous drone in the sub-region is task reanalysis, and determine the fourth execution strategy of the heterogeneous drone in the sub-region in combination with the second execution strategy of the corresponding heterogeneous drone.

[0155] The fifth execution strategy determination module is used to determine the sub-area task execution conflict based on the fourth execution strategy of the sub-area heterogeneous UAV, determine the first execution strategy for task assignment, and when the first execution strategy for task assignment is to analyze and adapt the UAV, extract each hotspot task, analyze the heterogeneous UAV adaptation index of each hotspot task, and determine the fifth execution strategy for the sub-area heterogeneous UAV.

[0156] It should be added that the heterogeneous drones of the present invention include but are not limited to charging drones and heterogeneous drones.

[0157] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0158] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0159] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0160] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0161] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0162] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A collaborative task allocation method for heterogeneous UAVs based on genetic algorithm, characterized by: The following steps are involved: The heterogeneous UAV mission execution area is recorded as the target area, and the target area is divided into sub-areas through k-means clustering. The first execution strategy of the heterogeneous UAVs in each sub-area is generated by the improved genetic algorithm. Based on the first mission execution strategy of the heterogeneous UAVs in each sub-region, the mission information is shared in real time. The second execution strategy of the heterogeneous UAVs in each sub-region is generated through the deep reinforcement learning network. The existence information of high-frequency tasks in each sub-region is simultaneously analyzed. Based on the existence information of high-frequency tasks in each sub-region, the third execution strategy of the heterogeneous UAVs in each sub-region is determined; When the third execution strategy of the heterogeneous UAVs in the sub-region is task reanalysis, the ranking penalty factor of each task of the heterogeneous UAVs is analyzed, and the fourth execution strategy of the heterogeneous UAVs in the sub-region is determined in combination with the second execution strategy of the corresponding heterogeneous UAVs; Based on the fourth execution strategy of sub-region heterogeneous UAVs, sub-region task execution conflicts are determined, and the first execution strategy of task allocation is determined. When the first execution strategy of task allocation is to analyze and adapt UAVs, each hotspot task is extracted, and the heterogeneous UAV adaptation index of each hotspot task is analyzed to determine the fifth execution strategy of sub-region heterogeneous UAVs.

2. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 1, characterized in that: The specific analysis method for generating the first execution strategy of heterogeneous drones in each sub-region is as follows: Obtain a set of tasks to be assigned, extract spatial features from the set of tasks to be assigned, and divide the tasks into several spatial task sub-regions based on the geographical location information of the tasks using the k-means clustering algorithm, which are recorded as sub-regions; Treat each sub-region as an independent optimization unit, construct a sub-task description vector including task number, task location coordinates, and resource requirement information, and statistically obtain the sub-task description vector of each sub-region; For each sub-region, a dual-objective fitness function is constructed with the goals of minimizing the total flight range of UAVs and minimizing the standard deviation of task resource load. The location coordinates, resource requirements and time constraints contained in the description vector of each task are used as input features for the individual construction and fitness calculation of the genetic algorithm. By combining the improved genetic algorithm with elite retention, roulette wheel selection and dynamic crossover mutation, the optimal initial task allocation and path execution plan for the heterogeneous UAVs in each sub-region is generated, and it is recorded as the first execution strategy for the heterogeneous UAVs in each sub-region.

3. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 1, characterized in that: The second execution strategy of heterogeneous drones in each sub-region is generated by the deep reinforcement learning network. The specific analysis process is as follows: The task information includes task priority, task time window, task location information and task completion status; The task information is input into the pre-trained deep reinforcement learning network to generate task execution order adjustment decisions for heterogeneous drones in each sub-area; According to the task execution order adjustment decision of the heterogeneous UAVs in each sub-area, the priorities of the currently assigned tasks of the heterogeneous UAVs in each sub-area are dynamically adjusted to form an updated task priority list of the heterogeneous UAVs in each sub-area, which is recorded as the second execution strategy of the heterogeneous UAVs in each sub-area.

4. The method for collaborative task allocation among heterogeneous UAVs based on genetic algorithm according to claim 1, characterized in that: The third execution strategy of the heterogeneous UAVs in each sub-area is determined based on the high-frequency task existence information of each sub-area. The specific process is as follows: Extract the task execution count threshold preset in the database; Obtain the number of times each task is executed using a preset monitoring cycle; The tasks whose execution times are less than the task execution times threshold are recorded as low-frequency tasks; The tasks whose execution times are greater than or equal to the task execution times threshold are recorded as high-frequency tasks; Traverse each task in the sub-region. If there is a high-frequency task, record the existence of the high-frequency task in the sub-region as the existence of a high-frequency task. Simultaneously, record the third execution strategy of the heterogeneous UAV in the sub-region as a task for re-analysis. If all tasks in the sub-region are low-frequency tasks, the existence of high-frequency tasks in the sub-region is recorded as no high-frequency tasks, and the third execution strategy of the heterogeneous drones in the sub-region is recorded as directly executing the second execution strategy of the heterogeneous drones in the sub-region; Traverse each sub-area, and determine the third execution strategy of the heterogeneous UAVs in each sub-area based on the existence information of high-frequency tasks in each sub-area.

5. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 1, characterized in that: The specific analysis process of analyzing the ranking penalty factors of each task of heterogeneous drones is as follows: During the preset monitoring period, the status information of each task is collected, including the path overlap rate, resource cache hit rate and urgency of each task; Analyze the sorting penalty factor of each task based on the status information of each task; The sorting penalty factor of each task is the quantitative data of the degree of influence of the path overlap rate, resource cache hit rate and urgency of each task on the execution constraints and interference of the task. The specific analysis process is: comparing the reference values ​​of the resource cache hit rate and urgency with the resource cache hit rate and urgency of each corresponding task, comparing the reference value of the path overlap rate with the path overlap rate of each task, and coupling the comparison results with the corresponding allocation factor to obtain the sorting penalty factor of each task.

6. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 5, characterized in that: The specific analysis process of determining the fourth execution strategy of heterogeneous UAVs in the sub-region is as follows: A dynamic task scheduling optimization network based on deep reinforcement learning is constructed. The second execution strategy of heterogeneous drones in the corresponding sub-region and the ranking penalty factor of each task in the corresponding sub-region are used as joint inputs, and the task priority adjustment result of the corresponding sub-region is output. The result of adjusting the task priority of the sub-region is recorded as the fourth execution strategy of the heterogeneous UAVs in the sub-region; The sorting penalty factor is used to quantify the potential interference risk during the task execution order adjustment process.

7. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 1, characterized in that: The specific analysis process of determining the first execution strategy of task allocation is as follows: According to the fourth execution strategy of the heterogeneous drones in the sub-region, the heat count of each task is obtained; Extract the preset heat count threshold in the database; The tasks whose heat count is less than the heat count threshold are recorded as low heat tasks; Tasks whose heat count is greater than or equal to the heat count threshold are recorded as hot tasks; Traverse each task in the sub-area. If there is a hot task, the first execution strategy of the task assignment is recorded as the analysis and adaptation drone; If all tasks are low-heat tasks, the first execution strategy of task assignment is recorded as the fourth execution strategy of directly executing the heterogeneous drones in the sub-region.

8. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 1, characterized in that: The specific analysis steps for analyzing the heterogeneous drone adaptation index of each hotspot task are as follows: Obtain heterogeneous drone adaptation data for hotspot tasks, including each drone's distance, remaining resources, and historical success rate; Analyze the adaptation index of heterogeneous drones for hot tasks based on the adaptation data of heterogeneous drones for hot tasks; The heterogeneous UAV adaptation index of the hotspot mission is a quantitative indicator of the impact of the distance, remaining resources and historical success rate of each UAV on the degree of adaptation between the hotspot mission and each UAV. The specific analysis process is as follows: the reference value of the distance is compared with the distance of each UAV, the remaining resources and historical success rate of each UAV are compared with the corresponding reference values, and the results of each comparison are coupled with the corresponding allocation coefficient to obtain the heterogeneous UAV adaptation index of the hotspot mission. Traverse each hotspot task to obtain the heterogeneous drone adaptation index of each hotspot task.

9. The method for heterogeneous UAV collaborative task allocation based on genetic algorithm according to claim 8, characterized in that: The specific analysis steps for determining the fifth execution strategy of heterogeneous UAVs in the sub-region are as follows: A1, the UAV with the maximum adaptation index of the heterogeneous UAVs of the hotspot task is selected as the executing UAV of the corresponding hotspot task; A2, traverses each hotspot task and obtains the executing drone of each hotspot task; If the executing drones of each hotspot task do not overlap, then the drone execution confirmation information of each hotspot task is generated; If there are overlapping execution drones, obtain the hotspot tasks corresponding to the overlapping execution drones and record them as the hotspot tasks to be confirmed; Obtain the second execution strategy for the overlapping UAVs, thereby taking the hotspot task with the highest priority among all the pending hotspot tasks as the execution task for the overlapping UAVs, generating UAV execution confirmation information for the hotspot task, and recording the remaining tasks among all the pending hotspot tasks as the second-confirmed hotspot tasks; A3: For each secondary confirmation hotspot task, re-obtain the drone corresponding to the maximum heterogeneous drone adaptation index and execute A2 until the corresponding execution drone is determined, generating the drone execution confirmation information for the corresponding hotspot task; A4 receives the drone execution confirmation information of each hotspot task and removes the corresponding hotspot task from the remaining drone task priority list, thereby obtaining the updated task priority list of heterogeneous drones and recording it as the fifth execution strategy of the sub-region heterogeneous drones.

10. A system using the genetic algorithm-based heterogeneous UAV collaborative task allocation method according to any one of claims 1 to 9, characterized in that: include: The first execution strategy generation module is used to record the heterogeneous UAV mission execution area as the target area, divide the target area into sub-areas through k-means clustering, and generate the first execution strategy of the heterogeneous UAVs in each sub-area through an improved genetic algorithm; The third execution strategy determination module is used to execute tasks based on the first task execution strategy of the heterogeneous UAVs in each sub-region, share task information in real time, generate the second execution strategy of the heterogeneous UAVs in each sub-region through a deep reinforcement learning network, and simultaneously analyze the existence information of high-frequency tasks in each sub-region. Based on the existence information of high-frequency tasks in each sub-region, the module determines the third execution strategy of the heterogeneous UAVs in each sub-region; a fourth execution strategy determination module, configured to analyze the ranking penalty factors of the tasks of the heterogeneous UAVs in the sub-region when the third execution strategy of the heterogeneous UAVs in the sub-region is task reanalysis, and determine the fourth execution strategy of the heterogeneous UAVs in the sub-region in combination with the second execution strategy of the corresponding heterogeneous UAVs; The fifth execution strategy determination module is used to determine the sub-area task execution conflict based on the fourth execution strategy of the sub-area heterogeneous UAV, determine the first execution strategy for task assignment, and when the first execution strategy for task assignment is to analyze and adapt the UAV, extract each hotspot task, analyze the heterogeneous UAV adaptation index of each hotspot task, and determine the fifth execution strategy for the sub-area heterogeneous UAV.

Images