Shared unmanned aerial vehicle task allocation method, device and system and unmanned aerial vehicle

Abstract

The invention discloses a shared unmanned aerial vehicle task allocation method, device and system and an unmanned aerial vehicle, and belongs to the field of task allocation, and the method comprises the steps: obtaining a to-be-allocated task, and calculating the similarity between the to-be-allocated task and an allocated task; when any similarity is not greater than a preset similarity, selecting N allocated tasks with the highest similarity as target allocated tasks; obtaining candidate unmanned aerial vehicles corresponding to the target allocated tasks, and obtaining a target executed task set of each candidate unmanned aerial vehicle; calculating the similarity between the to-be-allocated task and each executed task in all the target executed task sets, determining the executed task with the maximum similarity as a target task, and determining the candidate unmanned aerial vehicle executing the target task as a target unmanned aerial vehicle; and distributing the to-be-distributed task to the target unmanned aerial vehicle. According to the scheme, there is no need to construct a label for the unmanned aerial vehicle, matching is directly performed according to the task executed by the unmanned aerial vehicle, the task can be accurately allocated even if the task is complex, and the task execution efficiency is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of task allocation, in particular, to a shared unmanned aerial vehicle task allocation method, device, system and unmanned aerial vehicle. BACKGROUND

[0002] With the development of unmanned aerial vehicle technology, existing unmanned aerial vehicles do not need a pilot, and truly achieve the effect of unmanned operation. After sharing bicycles and sharing cars, shared unmanned aerial vehicles are also becoming more and more popular. Unlike shared bicycles and shared cars, shared unmanned aerial vehicles can be applied in multiple fields, such as tourism and consumption, the agricultural field, urban management and emergency response. Since the application in different fields may have great differences in the capabilities or equipment possessed by the unmanned aerial vehicle, for example, detection tasks require the unmanned aerial vehicle to have target description capabilities (such as having visual cameras, multispectral, lidar, sensors, etc.), while delivery tasks require the unmanned aerial vehicle to have transportation capabilities (such as having grippers, slings, cargo holds, etc.). Due to cost problems, it is impossible for an unmanned aerial vehicle to have multiple capabilities at the same time, and in practice, an unmanned aerial vehicle generally only has one or two capabilities. Although the unmanned aerial vehicle can be replaced with a pod to change the carried equipment and have different capabilities in other application scenarios, in the field of shared unmanned aerial vehicles, the carried equipment of the unmanned aerial vehicle is generally not replaced or said pod is fixed. Because replacing the carried equipment means that manual replacement costs need to be increased before the automated equipment replacement technology is put into use, and some equipment needs to be tested after replacement, in addition, the unmanned aerial vehicle cannot perform tasks during the replacement of the pod, which increases the non-working time of the unmanned aerial vehicle and affects the task execution efficiency.

[0003] Therefore, for user tasks, it is necessary to accurately allocate the tasks to unmanned aerial vehicles that have the ability to complete the tasks. The existing allocation method is to build labels for unmanned aerial vehicles, and then allocate the tasks to unmanned aerial vehicles with corresponding labels after determining the type of the tasks. However, this method can achieve accurate allocation when allocating simple types of tasks, but it may cause misallocation problems when involving complex tasks, which affects the task execution efficiency. SUMMARY

[0004] In order to overcome the deficiencies of the prior art, the present application provides a shared unmanned aerial vehicle task allocation method, device, system and unmanned aerial vehicle to solve the existing shared unmanned aerial vehicle task allocation, which uses label allocation, but the label has the problem of being prone to errors.

[0005] The technical solution adopted by the present application to solve its technical problems is: In a first aspect, a shared unmanned aerial vehicle task allocation method is provided, comprising: obtaining a task to be allocated, and calculating the similarity between the task to be allocated and an allocated task; When similarity of any one is not greater than the preset similarity, select N target allocated tasks with the highest similarity as target allocated tasks, N being a positive integer; Obtain candidate unmanned aerial vehicles for executing the target allocated tasks, and obtain a target executed task set of each candidate unmanned aerial vehicle; Calculate similarity between the task to be allocated and each executed task in all target executed task sets, determine an executed task with the greatest similarity as a target task, and determine a candidate unmanned aerial vehicle for executing the target task as a target unmanned aerial vehicle; Allocate the task to be allocated to the target unmanned aerial vehicle.

[0006] As an optional implementation manner of the application, the method further includes: Carry out word segmentation on the task to obtain an initial word vector of each word, the task being any one of a task to be allocated, an allocated task and an executed task; For each target word, calculate a cosine similarity between an initial word vector of the target word and an initial word vector of each other word, the target word being any one word after word segmentation of the task, and the other word being a word other than the target word after word segmentation of the task; Determine a first weight of the target word and the other word based on the cosine similarity, wherein the greater the cosine similarity, the greater the first weight, and take the other word with the cosine similarity greater than a preset cosine similarity as a related word; Take a product of the first weight and an initial word vector of the related word as a vector increment of the target word; Obtain a final word vector of the target word based on the initial word vector of the target word and the vector increment of the target word; Obtain a text vector of the task based on final word vectors of all target words, the text vector being used to calculate similarity with other tasks.

[0007] As an optional implementation manner of the application, the obtaining of the final word vector of the target word based on the initial word vector of the target word and the vector increment of the target word includes: Take a sum of the initial word vector of the target word and vector increments of all target words as an intermediate word vector of the target word; Sort words after word segmentation of the task, and obtain a sorting sequence number of the target word, and determine a second weight based on the sorting sequence number, wherein the sorting sequence number is different, and the second weight is different; Take a product of the second weight and the intermediate word vector as the final word vector of the target word.

[0008] As an optional implementation of the present application, the final word vector of the target word is obtained based on the initial word vector of the target word and the vector increment of the target word, comprising: The sum of the initial word vector of the target word and the vector increment of all target words is taken as the final word vector of the target word.

[0009] As an optional implementation of the present application, the method further comprises: The executed task set of each unmanned aerial vehicle is filtered to obtain a preset number of representative tasks; The representative task set of all unmanned aerial vehicles is taken as the assigned task.

[0010] As an optional implementation of the present application, the filtering of the executed task set of each unmanned aerial vehicle to obtain a preset number of representative tasks comprises: For each executed task set, a n*1 vector matrix is constructed based on the text vector of each task in the executed task set, where n is the number of tasks in the executed task set, and each element in the vector matrix corresponds to the text vector of a task; The sub-task similarity of each task in the vector matrix with any other task is calculated, and the overall task similarity of the task is obtained by adding the sub-task similarity of each task with all other tasks; The task with the highest overall task similarity is taken as a representative task; After obtaining at least one representative task, the representative similarity of each remaining task is calculated, the remaining task with the smallest representative similarity is taken as a representative task, and the step is repeated until a preset number of representative tasks are obtained; Wherein, the representative similarity is the sum of the sub-representative similarity of each remaining task, the sub-representative similarity of each remaining task is the similarity between the text vector of the remaining task and a representative task, and the remaining task is a task in the executed task set except the representative task.

[0011] As an optional implementation of the present application, the method further comprises: The size of the preset number is determined based on the number of tasks in the executed task set, and the more the number of tasks in the executed task set, the greater the value of the preset number.

[0012] In a second aspect, a shared unmanned aerial vehicle task allocation device is provided, comprising: A task similarity calculation module is configured to obtain a to-be-allocated task and calculate the similarity between the to-be-allocated task and the assigned task; A target task determination module is configured to select the N assigned tasks with the highest similarity as target assigned tasks when any similarity is not greater than a preset similarity, and N is a positive integer. a candidate UAV determination module configured to obtain candidate UAVs that execute target assigned tasks, and obtain a target executed task set of each of the candidate UAVs; a target UAV determination module configured to calculate a similarity between the task to be assigned and each executed task in all target executed task sets, determine an executed task with the largest similarity as a target task, and determine a candidate UAV that executes the target task as a target UAV; a task to be assigned assignment module configured to assign the task to be assigned to the target UAV.

[0013] In a third aspect, a shared UAV task assignment system is provided, comprising: at least one processor and at least one memory; the memory stores executable instructions of the processor; the processor is configured to execute the shared UAV task assignment method described in any of the above aspects.

[0014] In a fourth aspect, a shared UAV is provided, which applies the shared UAV task assignment method described in any of the above aspects.

[0015] Advantages: The technical solution of the present application provides a shared UAV task assignment method, device, system and UAV. The shared UAV task assignment method comprises obtaining a task to be assigned, and calculating a similarity between the task to be assigned and an assigned task. When any similarity is not greater than a preset similarity, N assigned tasks with the highest similarity are selected as target assigned tasks. Candidate UAVs that execute the target assigned tasks are obtained, and a target executed task set of each of the candidate UAVs is obtained. A similarity between the task to be assigned and each executed task in all target executed task sets is calculated, an executed task with the largest similarity is determined as a target task, and a candidate UAV that executes the target task is determined as a target UAV. The task to be assigned is assigned to the target UAV. The present application does not need to build a label for a UAV, but directly matches the tasks that have been executed by the UAV, so that the tasks can be accurately assigned even if the tasks are complex, and the task execution efficiency is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0017] Figure 1 is a shared unmanned aerial vehicle task allocation method flowchart provided by an embodiment of the present application; Figure 2 is a shared unmanned aerial vehicle task allocation device structure schematic diagram provided by an embodiment of the present application; Figure 3 is a shared unmanned aerial vehicle task allocation system structure schematic diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0018] To make the purposes, technical solutions and advantages of the present application clearer, the technical solutions of the present application are described in detail below in combination with the drawings and embodiments. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0019] To solve this problem, with reference to Figure 1 An embodiment of the present application provides a shared unmanned aerial vehicle task allocation method, which comprises the following steps: S11: obtaining a to-be-allocated task, and calculating the similarity between the to-be-allocated task and an allocated task; It should be noted that the present application scheme is applied to a server in the following system, and the application of the system is that a user sends a task to the server through a client (including but not limited to an APP and a mini program), and the server allocates the task to an unmanned aerial vehicle for execution.

[0020] When the server receives the to-be-allocated task, the text vector of the to-be-allocated task is obtained first, and then the cosine similarity between the text vector of the to-be-allocated task and the text vector of the allocated task is calculated, as the similarity between the to-be-allocated task and the allocated task. The main advantage of using the cosine similarity is that it can effectively measure the consistency of the direction of the text vector, ignore the absolute length difference, and be suitable for high-dimensional sparse data. In the high-dimensional space of the text vector (such as TF-IDF), the cosine similarity can effectively capture the semantic association, and is especially suitable for sparse data. In particular, the format or word habit input by the user is different, and the same task may use different words, so the cosine similarity can effectively ignore the length and word problems.

[0021] Conventional methods for obtaining text vectors include word2Vec and BERT models. However, in practical applications, while word2Vec is simple, it cannot consider text context and word order, making it prone to errors. For example, word2Vec might produce the same text vector for "A to B" and "B to A," but these actually belong to different tasks (because various factors influence flight path planning, causing the routes from A to B and B to A to differ, thus placing different demands on the drone). While BERT models can consider context and word order, they require training, resulting in higher costs.

[0022] Based on this, this application proposes a new method for obtaining text vectors, specifically including: The task is segmented to obtain the initial word vector for each word. The task can be any one of the following: a task to be assigned, a task already assigned, or a task already executed. The initial word vector can be obtained using methods such as word2Vec.

[0023] For each target word, the cosine similarity between the initial word vector of the target word and the initial word vector of each other word is calculated. The target word is any word after the task segmentation, and the other words are words after the task segmentation excluding the target word; that is, the relationship between each word in the text and its context is considered.

[0024] Other words with a cosine similarity greater than a preset cosine similarity are used as associated words. The first weight between the target word and the associated words is determined based on the cosine similarity. The larger the cosine similarity, the larger the first weight. The purpose of setting associated words is that in actual operation, words with a small cosine similarity have little semantic impact on the target word. Therefore, in order to reduce the amount of subsequent calculations, this application only takes associated words in the context for subsequent processing.

[0025] The product of the first weight and the initial word vector of the associated word is used as the vector increment of the target word; the final word vector of the target word is obtained based on the initial word vector of the target word and the vector increment of the target word; that is, the initial word vector of the target word is adjusted based on the vector of the associated word in the context, so that the adjusted target word vector can take into account the connection between the context, and the final text vector is more accurate.

[0026] The text vector for the task is obtained based on the final word vectors of all target words, and the text vector is used to calculate the similarity with other tasks.

[0027] In one embodiment, the sum of the initial word vector of the target word and the vector increments of all target words is used as the final word vector of the target word. However, the final word vector obtained in this way does not take into account the influence of word order and may be inaccurate.

[0028] Based on this, in another embodiment of this application: the sum of the initial word vector of the target word and the vector increments of all target words is used as the intermediate word vector of the target word; The words segmented by the task are sorted, and the sorting index of the target word is obtained. A second weight is determined based on the sorting index, wherein different sorting indices result in different second weights; the earlier the sorting index, the larger or smaller the second weight can be, and only one method needs to be used in practice. It should be noted that the second weight is set to a small value to avoid having a significant impact on the overall semantics of the target word. However, since each target word in the text has been adjusted, the final text vector will differ significantly when there are identical characters but different word orders (when there are many words in the text).

[0029] The product of the second weight and the intermediate word vector is taken as the final word vector of the target word.

[0030] Furthermore, in actual use, when no tasks have been executed, assigned tasks are configured manually (i.e., a task template is set for each type of task that each drone can perform, and this task template is used as an assigned task), or tasks are assigned manually. After a period of use, when the number of executed tasks is greater than a first preset value but less than a second preset value, all executed tasks are used as assigned tasks. When the number of executed tasks is greater than or equal to the second preset value, assigned tasks are determined based on the following method: The number of representative tasks is obtained by filtering the executed task set of each UAV; the representative task sets of all UAVs are then used as assigned tasks.

[0031] While manual screening can be used, its efficiency is low when the number of tasks already executed is large. Therefore, as a preferred implementation method in this application: For each set of executed tasks, construct an n*1 vector matrix A based on the text vector of each task in the set of executed tasks, where n is the number of tasks in the set of executed tasks, and each element in the vector matrix corresponds to the text vector of a task. Calculate the subtask similarity of each task in the vector matrix with any other task, and sum the subtask similarities of each task with all other tasks to obtain the overall task similarity. In practice, multiply matrix A by its transpose to obtain an n*n similarity matrix B (with all diagonals being 1). Then, sum the elements of each row of similarity matrix B to obtain an n*1 overall task similarity matrix C, where each element represents the overall task similarity of each task plus 1 (its own similarity). Since 1 is added to all elements, it does not affect their relative similarities. The representative task can then be obtained based on the overall task similarity matrix C.

[0032] The task with the highest similarity to the overall task is taken as a representative task; the resulting representative task is the one that best represents the set of executed tasks.

[0033] After obtaining at least one representative task, calculate the representative similarity of each remaining task, and take the remaining task with the smallest representative similarity as the representative task. Repeat this step until a preset number of representative tasks are obtained. The representative similarity is the sum of the sub-representative similarities of each remaining task. The sub-representative similarity of each remaining task is the similarity between the text vector of the remaining task and a representative task. The remaining tasks are those other than the representative task in the executed task set. It is understood that after obtaining a representative task, other representative tasks need to be obtained from the remaining executed task set. If the overall similarity is calculated at this point, all representative tasks would belong to the same category, ignoring the possibility that the drone can perform other tasks, resulting in a waste of drone resources. Therefore, this application introduces representative similarity. By calculating the representative similarity between the remaining task and all representative tasks, the smaller the representative similarity, the greater the probability that the remaining task and the representative task are of different types. Therefore, the remaining task with the smallest representative similarity is selected as the new representative task.

[0034] As an optional embodiment of this application, after obtaining at least one representative task, the representative similarity of each remaining task is calculated, and the comprehensive similarity of each remaining task is obtained based on the overall task similarity and the representative similarity. The remaining task with the highest comprehensive similarity is selected as the representative task, and this step is repeated until a preset number of representative tasks are obtained. Wherein, comprehensive similarity = first preset weight * overall similarity - second preset weight * representative similarity. This embodiment considers overall similarity in addition to representative similarity, aiming to select tasks among the remaining tasks that are of a different type from the representative task and can represent the other remaining tasks.

[0035] In one embodiment, the preset quantity is a fixed value.

[0036] In another embodiment, the preset number is determined based on the number of tasks in the executed task set, and the larger the number of tasks in the executed task set, the larger the value of the preset number.

[0037] S12: When any similarity is not greater than the preset similarity, select the N assigned tasks with the highest similarity as the target assigned task, where N is a positive integer; Understandably, if any similarity score is no greater than a preset similarity score, it indicates that there is no already assigned task that is completely identical to the task to be assigned. Therefore, further confirmation is needed.

[0038] If any similarity score is greater than a preset similarity score, the drone corresponding to the assigned task with a similarity score greater than the preset similarity score is identified as the target drone, and the task to be assigned is directly assigned to the target drone. If any similarity score is greater than the preset similarity score, it indicates that there is an already assigned task that is completely identical to the task to be assigned. In this case, no further confirmation is needed, and the task can be assigned directly.

[0039] S13: Obtain candidate drones corresponding to the assigned tasks of the execution target, and obtain the target executed task set for each candidate drone; that is, further confirm that the calculation is performed on the target executed task set of candidate drones with high similarity. When there are many drones and each drone has a large number of executed tasks, calculating only the target executed task set of candidate drones is less computationally intensive, faster in task allocation, and more efficient than calculating the executed task set of all drones.

[0040] S14: Calculate the similarity between the task to be assigned and each executed task in the set of all target executed tasks, determine the executed task with the highest similarity as the target task, and determine the candidate drone to execute the target task as the target drone; in actual calculation, there may be multiple executed tasks with the highest similarity, in which case any one can be selected, or a relatively idle drone can be arranged based on the current number of tasks of the drone.

[0041] S15: Assign the task to be assigned to the target drone.

[0042] It should be noted that, in practical use, the accuracy of task allocation in this application depends on the accuracy of previously assigned or executed tasks. Therefore, this application aims to ensure the accuracy of previously assigned tasks as much as possible. For example, when no tasks have been assigned, each UAV is configured with executable tasks, and each type of executable task is considered an assigned task. When the number of executed tasks is large (greater than a preset value), the assigned task is determined based on the executed tasks (see the above embodiment for details). Furthermore, to further ensure the accuracy of task allocation, task execution feedback is added (e.g., user rating or evaluation after task completion). A matching degree is obtained based on the task execution feedback (higher scores indicate higher matching degrees). After calculating the similarity between the task to be assigned and the executed tasks, the executed task with the largest product of similarity and matching degree is selected as the target task.

[0043] It should be noted that any process or method description in the flowchart or otherwise described herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order according to the functions involved, as should be understood by those skilled in the art to which the embodiments of this application pertain.

[0044] Furthermore, in the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0045] Based on the same inventive concept, such as Figure 2 As shown, this application embodiment provides a shared drone task allocation device 20, including: The task similarity calculation module 21 is used to obtain the task to be assigned and calculate the similarity between the task to be assigned and the assigned task. Specifically, the task is segmented into words to obtain the initial word vector for each word, wherein the task is any one of the following: a task to be assigned, a task already assigned, or a task already executed. For each target word, the cosine similarity between the initial word vector of the target word and the initial word vectors of each other word is calculated. The target word is any word after word segmentation in the task, and the other words are words other than the target word after word segmentation in the task. Other words with a cosine similarity greater than a preset cosine similarity are designated as associated words. A first weight between the target word and the associated words is determined based on the cosine similarity, wherein the larger the cosine similarity, the larger the first weight. The product of the first weight and the initial word vector of the associated word is used as the vector increment of the target word. The final word vector of the target word is obtained based on the initial word vector of the target word and the vector increment of the target word. The text vector for the task is obtained based on the final word vectors of all target words, and the text vector is used to calculate the similarity with other tasks.

[0046] In one embodiment, the sum of the initial word vector of the target word and the vector increments of all target words is used as the final word vector of the target word. However, the final word vector obtained in this way does not take into account the influence of word order and may be inaccurate.

[0047] Based on this, in another embodiment of this application: the sum of the initial word vector of the target word and the vector increments of all target words is used as the intermediate word vector of the target word; The words after word segmentation are sorted, and the sorting number of the target word is obtained. A second weight is determined based on the sorting number, wherein the second weight is different for different sorting numbers. The product of the second weight and the intermediate word vector is taken as the final word vector of the target word.

[0048] In addition, it also includes: filtering the executed task set of each UAV to obtain a preset number of representative tasks; and treating the representative task set of all UAVs as assigned tasks.

[0049] The step of filtering the executed task set of each UAV to obtain a preset number of representative tasks includes: For each executed task set, an n*1 vector matrix is ​​constructed based on the text vector of each task in the executed task set, where n is the number of tasks in the executed task set, and each element in the vector matrix corresponds to the text vector of a task; the subtask similarity of each task in the vector matrix with any other task is calculated, and the subtask similarities of each task with all other tasks are added together to obtain the overall task similarity of the task; the task with the highest overall task similarity is taken as a representative task; after obtaining at least one representative task, the representative similarity of each remaining task is calculated, and the remaining task with the smallest representative similarity is taken as a representative task. This step is repeated until a preset number of representative tasks are obtained; wherein, the representative similarity is the sum of the sub-representative similarities of each remaining task, and the sub-representative similarity of each remaining task is the similarity between the remaining task and the text vector of a representative task, and the remaining tasks are the tasks in the executed task set excluding the representative tasks.

[0050] Furthermore, the preset quantity is determined based on the number of tasks in the executed task set; the more tasks in the executed task set, the larger the value of the preset quantity.

[0051] The target task determination module 22 is used to select the N assigned tasks with the highest similarity as the target assigned tasks when any similarity is not greater than a preset similarity, where N is a positive integer.

[0052] The candidate drone determination module 23 is used to obtain the candidate drones corresponding to the assigned tasks of the execution target, and to obtain the set of executed tasks of each candidate drone.

[0053] The target drone determination module 24 is used to calculate the similarity between the task to be assigned and each executed task in the set of all target executed tasks, determine the executed task with the highest similarity as the target task, and determine the candidate drone to execute the target task as the target drone.

[0054] The task allocation module 25 is used to allocate the task to be assigned to the target UAV.

[0055] Based on the same inventive concept, such as Figure 3 As shown, this application also provides a shared drone task allocation system 30, comprising: At least one processor 31 and at least one memory 32; The memory stores the executable instructions of the processor; The processor is configured to execute the shared drone task allocation method provided in the above embodiments.

[0056] Based on the same inventive concept, this application also provides a shared drone, applying the shared drone task allocation method described in any of the above claims.

[0057] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0058] It should be noted that in the description of this application, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this application, unless otherwise stated, "a plurality of" means at least two.

[0059] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A method for allocating shared drone tasks, characterized in that, include: Obtain the tasks to be assigned and calculate the similarity between the tasks to be assigned and the tasks already assigned; When any similarity is not greater than a preset similarity, select the N assigned tasks with the highest similarity as the target assigned task, where N is a positive integer; Obtain candidate drones corresponding to the assigned tasks of the execution target, and obtain the set of executed tasks for each candidate drone; Calculate the similarity between the task to be assigned and each executed task in the set of all target executed tasks, determine the executed task with the highest similarity as the target task, and determine the candidate UAVs to execute the target task as the target UAVs; The task to be assigned is assigned to the target drone.

2. The method according to claim 1, characterized in that, Also includes: The task is segmented into words to obtain the initial word vector for each word, where the task can be any one of the following: a task to be assigned, a task already assigned, or a task already executed. For each target word, calculate the cosine similarity between the initial word vector of the target word and the initial word vector of each other word, where the target word is any word after word segmentation by the task, and the other words are words after word segmentation by the task other than the target word; Other words with a cosine similarity greater than a preset cosine similarity are used as associated words. The first weight between the target word and the associated words is determined based on the cosine similarity, wherein the greater the cosine similarity, the greater the first weight. The product of the first weight and the initial word vector of the associated word is used as the vector increment of the target word; The final word vector of the target word is obtained based on the initial word vector of the target word and the vector increment of the target word; The text vector for the task is obtained based on the final word vectors of all target words, and the text vector is used to calculate the similarity with other tasks.

3. The method according to claim 2, characterized in that: The process of obtaining the final word vector of the target word based on the initial word vector of the target word and the vector increment of the target word includes: The sum of the initial word vector of the target word and the vector increments of all target words is used as the intermediate word vector of the target word; The words after word segmentation are sorted and the sorting number of the target word is obtained. A second weight is determined based on the sorting number, wherein the second weight is different if the sorting number is different. The product of the second weight and the intermediate word vector is taken as the final word vector of the target word.

4. The method according to claim 2, characterized in that: The process of obtaining the final word vector of the target word based on the initial word vector of the target word and the vector increment of the target word includes: The sum of the initial word vector of the target word and the vector increments of all target words is taken as the final word vector of the target word.

5. The method according to claim 1, characterized in that, Also includes: The preset number of representative tasks are obtained by filtering the set of tasks executed by each drone. Treat all representative mission sets of all drones as assigned missions.

6. The method according to claim 5, characterized in that: The step of filtering the executed task set of each UAV to obtain a preset number of representative tasks includes: For each set of executed tasks, an n*1 vector matrix is ​​constructed based on the text vector of each task in the set of executed tasks, where n is the number of tasks in the set of executed tasks, and each element in the vector matrix corresponds to the text vector of a task. Calculate the subtask similarity between each task in the vector matrix and any other task, and sum the subtask similarities of each task with all other tasks to obtain the overall task similarity of the task; The task with the highest overall task similarity is taken as a representative task; After obtaining at least one representative task, calculate the representative similarity of each remaining task, and take the remaining task with the smallest representative similarity as the representative task. Repeat this step until a preset number of representative tasks are obtained. Wherein, the representative similarity is the sum of the sub-representative similarities of each remaining task, and the sub-representative similarity of each remaining task is the similarity between the text vector of the remaining task and a representative task, and the remaining tasks are the tasks in the executed task set other than the representative tasks.

7. The method according to claim 5, characterized in that, Also includes: The preset number is determined based on the number of tasks in the executed task set. The more tasks in the executed task set, the larger the value of the preset number.

8. A shared drone task allocation device, characterized in that, include: The task similarity calculation module is used to obtain tasks to be assigned and calculate the similarity between the tasks to be assigned and the assigned tasks. The target task determination module is used to select the N most similar assigned tasks as target assigned tasks when any similarity is not greater than a preset similarity, where N is a positive integer; The candidate drone determination module is used to obtain candidate drones corresponding to the assigned tasks of the execution target, and to obtain the set of executed tasks of each candidate drone. The target drone determination module is used to calculate the similarity between the task to be assigned and each executed task in the set of all target executed tasks, determine the executed task with the highest similarity as the target task, and determine the candidate drones that execute the target task as the target drones; The task allocation module is used to assign the task to be assigned to the target UAV.

9. A shared unmanned aerial vehicle (UAV) task allocation system, characterized in that, include: At least one processor and at least one memory; The memory stores the executable instructions of the processor; The processor is configured to perform the method according to any one of claims 1-7.

10. A shared drone, characterized in that, The method described in any one of claims 1-7.

Images