Structured instruction semantic interaction method oriented to unmanned system cluster task allocation

Abstract

The invention discloses an unmanned system cluster task allocation-oriented structured instruction semantic interaction method. The method comprises the following steps of S1, cluster task voice instruction recognition; converting cluster task voice instruction data issued by an operator into a cluster task text instruction through a voice recognition technology; s2, performing semantic analysis on the cluster tasks; keywords in the cluster task text instruction are extracted through natural language processing, and then a corresponding cluster task structured instruction is generated through slot filling; s3, distributing and solving cluster tasks; and a corresponding cluster task allocation result is obtained by matching a corresponding cluster task allocation model and algorithm solution. According to the method, the unmanned cluster command and control quintuple structured instruction slot normal form with universality is constructed, and an adaptive task allocation scheme matching mechanism of a task-level instruction is designed, so that unmanned system cluster command and control are more natural, simpler and more efficient.

Description

Technical Field

[0001] The present invention relates to the technical field of human - computer interaction, in particular to a task - level structured instruction semantic interaction method for unmanned system cluster task allocation. Background Art

[0002] With the rapid development of artificial intelligence technology, intelligent unmanned systems are being more and more widely used in various fields. The successful application of unmanned systems is inseparable from the remote task planning and interactive control of personnel. Currently, operating an unmanned system often requires multiple ground operators, resulting in a huge workload and cognitive load for personnel. Compared with the traditional method of operating unmanned systems using a ground station and a joystick for remote control, the unmanned system control technology based on structured instruction semantic interaction has the advantages of natural interaction mode, high interaction efficiency, convenient control method, and low cognitive load of operators. However, the existing structured instruction semantic interaction technology can only perform action - level control on a single unmanned system. In the future, unmanned system clusters are likely to be applied in various industries. To reduce the control load of unmanned systems and improve the control efficiency of unmanned systems, there is an urgent need for a human - computer interaction and unmanned system interaction control technology that enables an operator to efficiently and agilely command multiple unmanned systems.

[0003] Research institutions at home and abroad have conducted research on action - level structured instructions, but task - level structured instructions are still blank. In the research on action - level structured instructions for single - unmanned - system control, the research team of Fu Kai from the Xi'an Flight Automatic Control Research Institute of Aviation Industry proposed a method for identifying drone instruction intentions based on deep learning, which can accurately identify instruction intentions, and by extracting key instruction information, obtain structured instructions that drones can directly execute, realizing direct interaction between air traffic controllers and drones.

[0004] The existing action - level structured instruction semantic interaction methods for single - unmanned - system control can only control a single unmanned system at the action level. Although they have the advantage of a more natural way of controlling a single unmanned system, they cannot significantly reduce the cognitive load of unmanned - system operators. Moreover, the existing action - level structured instruction semantic interaction methods cannot control unmanned system clusters. Summary of the Invention

[0005] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide a structured instruction semantic interaction method for unmanned system cluster task allocation, construct a general unmanned cluster command - and - control five - tuple structured instruction slot paradigm, and design an adaptive task - allocation scheme matching mechanism for task - level instructions, so as to make the unmanned system cluster command - and - control more natural, simpler, and more efficient for natural semantic interaction in unmanned system cluster task allocation.

[0006] To achieve the above object, the present invention provides a structured instruction semantic interaction method for unmanned system cluster task allocation, including the following steps:

[0007] S1. Cluster task voice instruction recognition; converting the cluster task voice instruction data issued by the operator into cluster task text instructions through voice recognition technology;

[0008] S2. Cluster task semantic parsing; extracting keywords in the cluster task text instructions through natural language processing, and then generating corresponding cluster task structured instructions through slot filling;

[0009] S3. Cluster task allocation solution; solving through matching corresponding cluster task allocation models and algorithms to obtain corresponding cluster task allocation results.

[0010] Further, the unmanned system cluster is a drone cluster.

[0011] Further, in step S2, the unmanned system cluster task allocation structured instruction slots include two-level nested slots, namely sibling slots and dependency slots.

[0012] Further, taking the unmanned system cluster task allocation five-tuple as the slot paradigm for task-level structured instruction semantic interaction, the unmanned system cluster task allocation five-tuple includes battlefield environment, unmanned system cluster set, target object set, task type set, and constraint conditions.

[0013] Further, the number of dependency slots of each sibling slot can be dynamically increased or decreased according to the actual situation, and the value of each dependency slot is assigned or defaulted according to the issued voice instruction.

[0014] Further, in step S3, the general task allocation scheme for unmanned system task-level instructions based on the auction algorithm is generated based on the auction algorithm.

[0015] Further, the auction algorithm is specifically set as follows in the model for unmanned system cluster task allocation: N unmanned systems execute tasks for M targets. One unmanned system can only receive one task, and one task can only be assigned to one unmanned system; each unmanned system has a corresponding benefit for executing the corresponding task, and each task has its own cost. The net benefit of the drone for executing the corresponding task is the benefit minus the cost. By optimizing the task allocation, the overall benefit of the unmanned system cluster for executing tasks is maximized.

[0016] Further, in the model for unmanned system cluster task allocation, the corresponding mathematical symbol definitions are as follows:

[0017] p ij: The cost when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j-th (j = 0, 1, 2..., M - 1) target; the cost includes energy consumption and / or ammunition consumption;

[0018] a ij : The benefit when the i-th unmanned system executes a task on the j-th target; the benefit is related to the value of the target;

[0019] V ij : The net benefit when the i-th unmanned system executes a task on the j-th target, V ij =a ij -p ij ;

[0020] Each task can be assigned to at most one unmanned system;

[0021] Each unmanned system can receive at most one task;

[0022] x ij ∈{0, 1} represents the decision variable of whether the i-th unmanned system executes a task on the j-th target. If the i-th unmanned system executes a task on the j-th target, then x ij =1. If the i-th unmanned system does not execute a task on the j-th target, then x ij =0;

[0023]

[0024] Furthermore, in the model for unmanned system cluster task allocation, the optimization objectives include minimizing losses, maximizing benefits, and maximizing net benefits.

[0025] Furthermore, from the perspective of the comparison between the number of unmanned systems and the number of target objects, it is divided into three cases: N > M, N = M, and N < M. The comparison situation between the number of unmanned systems and the number of target objects and the three optimization objectives are cross-combined to form a total of 9 types of cluster task allocation models, realizing the matching of unmanned system cluster task-level structured instructions and cluster task allocation schemes.

[0026] The beneficial effects of the present invention are as follows:

[0027] The present invention comprehensively applies speech recognition, semantic analysis, and unmanned system task allocation technologies, and innovatively proposes a task-level structured instruction interaction method for unmanned system cluster task allocation, mainly including the design of a five-tuple structured instruction slot paradigm that is common and standard for unmanned cluster control, an adaptive general task allocation scheme generation mechanism for task-level instructions, etc. Compared with the previous cluster control methods where multiple operators use remote controls to operate unmanned system clusters, or load task algorithms for each unmanned system in advance to control unmanned system clusters, the results of the present invention are expected to enable an unmanned system operator to control the entire unmanned system cluster through voice / semantic instructions by applying the task-level structured instruction interaction method, thereby making the command and control of unmanned system clusters simpler and more efficient, and significantly reducing the workload of unmanned system operators. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 Schematic diagrams of action-level structured instructions and task-level structured instructions;

[0029] Figure 2 Overall flowchart of task-level structured instruction semantic interaction for unmanned system cluster task allocation;

[0030] Figure 3 Schematic diagram of the five-tuple structured instruction slot paradigm for unmanned system cluster task allocation. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0032] In the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be construed as indicating or implying relative importance.

[0033] In the description of the present invention, it should be noted that, unless otherwise clearly specified and defined, the terms "installation", "connection", and "coupling" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0034] The following Figures 1 - 3 will describe in detail the specific embodiments of the present invention. It should be understood that the specific embodiments described herein are only for the purpose of illustration and explanation of the present invention, and are not intended to limit the present invention.

[0035] Structured instructions are in the process of human-machine interaction. Through interaction algorithms such as natural language processing, the natural language interaction instructions issued by humans are processed and transformed into structured instructions that can be executed by unmanned systems. Such instructions usually have specific formats and grammars to ensure their accuracy and executability. According to the number of controlled targets and instruction characteristics, structured instructions can be divided into two levels: action level and task level, as Figure 1 shown. The action level mainly targets individual unmanned systems, and the instruction content mainly includes specific action steps such as "ascend, turn right, hover". The task level mainly targets unmanned system clusters, and the instruction content mainly includes target tasks such as "search, action".

[0036] As Figure 2 shown, the structured instruction semantic interaction method for unmanned system cluster task allocation is mainly generated according to three major steps: cluster task voice instruction recognition, cluster task semantic parsing, and cluster task allocation solution. Specifically, according to a structured instruction semantic interaction method for unmanned system cluster task allocation of the present invention, it includes the following steps:

[0037] S1. Cluster task voice instruction recognition; establish a corpus containing information related to the five-tuple of unmanned system cluster task allocation, and convert the cluster task voice instruction data issued by the operator into cluster task text instructions through online / offline voice recognition technology;

[0038] S2. Cluster task semantic parsing; extract keywords in the cluster task text instructions through natural language processing, and then generate corresponding cluster task structured instructions through slot filling according to the slot paradigm of the five-tuple of unmanned system cluster task allocation;

[0039] S3. Cluster task allocation solution; solve through matching the corresponding cluster task allocation model and algorithm to obtain the corresponding cluster task allocation result.

[0040] Specifically, in step S2, the slot paradigm of the structured instruction for the unmanned system cluster task allocation is set as follows:

[0041] The unmanned system cluster task allocation generally needs to consider the battlefield environment, the unmanned system cluster set, the target object set, the task type set, and the constraint conditions, that is, the five-tuple of the unmanned system cluster task allocation. Since the task-level structured instruction semantic interaction method of the present invention is oriented to the unmanned system cluster task allocation, taking the five-tuple of the unmanned system cluster task allocation as the slot paradigm of the task-level structured instruction semantic interaction can make the structured instruction for the unmanned system cluster task allocation proposed by the present invention have broader generality. The designed slot paradigm of the five-tuple structured instruction for the unmanned system cluster task allocation of the present invention is as Figure 3 shown.

[0042] The slot of the five-tuple structured instruction for the unmanned system cluster task allocation mainly includes two-level nested slot frameworks of parallel slots (slots) and dependent slots (sub-slots). Note that Figure 3 the number of dependent slots of each parallel slot can be dynamically increased or decreased according to the actual situation, and the value of each dependent slot is assigned or defaulted according to the issued voice instruction.

[0043] In a specific embodiment, the settings are as follows: The parallel slots include slot 1: battlefield environment, slot 2: unmanned system cluster set, slot 3: target object set; slot 4: task type of the target object, slot 5: constraint conditions. Slot 1 includes sub-slots 1-1: weather, sub-slots 1-2: electromagnetism, sub-slots 1-3: geography..., slot 2 includes sub-slots 2-1: unmanned system 1, sub-slots 2-2: unmanned system 2..., slot 3 includes sub-slots 3-1: target object 1, sub-slots 3-2: target object 2..., slot 4 includes sub-slots 4-1: reconnaissance, sub-slots 4-2: tracking, sub-slots 4-3: strike..., slot 5 includes sub-slots 5-1: minimum loss, sub-slots 5-2: maximum benefit, sub-slots 5-3: maximum net benefit. Thus, the number of unmanned systems can be obtained according to the number of sub-slots of slot 2, the number of target objects can be obtained according to the number of sub-slots of slot 3, the task type can be obtained according to the filled content of slot 4, and the constraint conditions can be obtained according to the filled content of slot 5. The constraint conditions are minimum loss, maximum benefit, or maximum net benefit.

[0044] Among them, the content of slot 1 is the environmental background of the entire task allocation, which affects the entire task allocation process according to the actual situation. For example, it will affect the content of the voice instruction issued by the operator, thereby determining the specific content filled in the slot. For example, when the environmental conditions are harsh, it can be set that the unmanned system refuses to execute the task, etc. How the content of slot 1 specifically affects the result of the task allocation needs to be further set according to the actual situation.

[0045] The content of slot 4 is the task type of the target object, which exerts an influence on the entire task allocation process according to the actual situation. For example, when performing a reconnaissance task, the task allocation should be carried out according to the optimization goal of minimizing losses, and when performing a strike task, the task allocation should be carried out according to the optimization goal of maximizing benefits, and so on.

[0046] In step S3, the generation mechanism of the general task allocation scheme for the task-level instructions of the unmanned system based on the auction algorithm is as follows:

[0047] After the generation of the structured instructions for the unmanned system cluster task allocation, it is necessary to calculate and solve them to give a specific task allocation scheme. In order to make the task allocation scheme generated based on the task-level structured instructions be able to adapt to various task allocation scenarios more generally, it is urgent to design a general task allocation scheme generation mechanism for task-level instructions. In the aspect of unmanned system cluster task allocation, although there have been a large number of research works related to task allocation algorithms at home and abroad, each task allocation algorithm has its own scope of application, and there are few task allocation algorithms that can be applied to multiple task allocation scenarios.

[0048] In the task allocation algorithm, the auction algorithm is a computational mechanism that realizes resource matching through the way of buyer bidding under a series of clear rules. The basic idea of the auction algorithm is that the auction item corresponds to the task, and the task allocator and the receiver respectively auction and bid for the task according to their own profit functions and bidding strategies. The mathematical derivation of the auction algorithm is rigorous, the computational complexity can be proved, the theoretical explanation is clear, the restrictions on the network scale and topology are small, and in practical applications, the cluster individuals only need to calculate the bidding or tendering information, with small time complexity and computational volume, and high operating efficiency, which is very suitable for distributed task allocation problems.

[0049] In the conventional task allocation model based on the auction algorithm, the number of task allocation subjects is often equal to the number of target objects to be allocated. When the number of task allocation subjects is greater than or equal to the number of target objects to be allocated, the solution ideas and steps of the model will change. From the perspective of sets, assuming that in the task allocation model, the number of task allocation subjects is N and the number of target objects to be allocated is M, from the perspective of the comparison between the number of task allocation subjects N and the number of target objects M to be allocated, all task allocation models based on the auction algorithm are nothing more than three situations: N > M, N = M, and N < M. If the auction algorithm task allocation models in these three scenarios are solved simultaneously, various task allocation models based on the auction algorithm can be matched, making the task allocation models based on the auction algorithm have a certain degree of generality.

[0050] In the model for task allocation of unmanned system clusters, N unmanned systems execute tasks on M targets. One unmanned system can only receive one task, and one task can only be assigned to one unmanned system. Each unmanned system has a corresponding benefit when executing the corresponding task, and each task has its own cost. The net benefit of the unmanned aerial vehicle executing the corresponding task is the benefit minus the cost. By optimizing the task allocation, the overall benefit of the unmanned system cluster in executing tasks is maximized.

[0051] The corresponding mathematical symbol definitions are as follows:

[0052] p ij : The cost (such as energy consumption, ammunition consumption, etc. of the unmanned system) when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j-th (j = 0, 1, 2..., M - 1) target;

[0053] a ij : The benefit when the i-th unmanned system executes a task on the j-th target. a ij is mainly related to the production cost c of the j-th target j and the system importance degree β. a ij = βc j (0 ≤ β ≤ 1);

[0054] V ij : The net benefit when the i-th unmanned system executes a task on the j-th target. V ij = a ij - p ij ;

[0055] Each task can be assigned to at most one unmanned system;

[0056] Each unmanned system can receive at most one task;

[0057] x ij ∈ {0, 1} represents the decision variable of whether the i-th unmanned system executes a task on the j-th target. If the i-th unmanned system executes a task on the j-th target, then x ij = 1. If the i-th unmanned system does not execute a task on the j-th target, then x ij = 0.

[0058]

[0059] For the optimization objectives, the present invention altogether considers three optimization objectives: minimum loss, maximum benefit, and maximum net benefit.

[0060] The optimization objective expression for minimum loss is as follows.

[0061]

[0062] The optimization objective expression with the maximum benefit is as follows.

[0063]

[0064] The optimization objective expression with the maximum net benefit is as follows.

[0065]

[0066] In summary, the optimization model group for task allocation of the unmanned system cluster of the present invention is shown in the following table. From the perspective of the comparison between the number of unmanned systems and the number of target objects, it can be divided into three cases: N > M, N = M, and N < M. From the perspective of the optimization objective, it can be divided into three cases: minimum loss, maximum benefit, and maximum net benefit. By cross-combining the comparison of the number of unmanned systems and the number of target objects with the three optimization objectives, a model group containing 9 types of cluster task allocation models can be formed in total (Table 1), realizing the matching of the task-level structured instructions of the unmanned system cluster and the cluster task allocation scheme, and also making the model group for task allocation of the unmanned system cluster provided by the present invention more general and applicable.

[0067] Table 1 Optimization model group for task allocation of the unmanned system cluster based on the auction algorithm

[0068]

[0069] The conventional solution of the auction algorithm mainly targets the case where the number of allocation subjects is equal to the number of allocation objects. The present invention expands and improves the solution steps of the auction algorithm, enabling the auction algorithm to solve various cases where the number of unmanned systems is greater than, equal to, or less than the number of target objects. By cross-combining the comparison of the number of unmanned systems and the number of target objects with the three optimization objectives, a model group containing 9 types of cluster task allocation models is designed in total, which can handle various cases of unmanned system cluster task allocation. All parameters or formulas in Table 1 are unique to this project.

[0070] Specifically, the specific implementation method of step S3 is as follows:

[0071] S3.1 In the task allocation model, obtain the number N of task allocation subjects and the number M of target objects to be allocated. Among them, the number N of task allocation subjects is the number of unmanned systems in slot 2, and the number M of target objects to be allocated is the number of available target objects in slot 3; determine the relationship between the number of allocation subjects and the number of target objects to be allocated as N > M, N = M, or N < M;

[0072] S3.2 Determine the optimization objective, where the optimization objective is the constraint condition in slot 5; the constraint conditions include the least loss, the highest benefit, and the highest net benefit;

[0073] S3.3 Cross - combines the comparison of the number of unmanned systems and the number of target objects with three optimization objectives, and designs a model group containing 9 types of cluster task allocation models, specifically including:

[0074] ① When it is judged that the relationship between the number of allocation entities and the number of target objects to be allocated is N > M, and the optimization objective is to minimize the loss, calculate the loss cost when the i - th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j - th (j = 0, 1, 2..., M - 1) target. The cost includes the energy consumption and ammunition consumption of the unmanned system; at this time, the optimization objective is The constraint conditions are The solution steps of the task allocation model are adjusted accordingly according to the optimization objective and the constraint conditions.

[0075] ② When it is judged that the relationship between the number of allocation entities and the number of target objects to be allocated is N > M, and the optimization objective is to maximize the benefit, calculate the benefit when the i - th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j - th (j = 0, 1, 2..., M - 1) target. The benefit is related to the value of the target; at this time, the optimization objective is The constraint conditions are The solution steps of the task allocation model are adjusted accordingly according to the optimization objective and the constraint conditions.

[0076] ③ When it is judged that the relationship between the number of allocation entities and the number of target objects to be allocated is N > M, and the optimization objective is to maximize the net benefit, calculate the net benefit when the i - th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j - th (j = 0, 1, 2..., M - 1) target. The net benefit is equal to the benefit minus the loss; at this time, the optimization objective is The constraint conditions are The solution steps of the task allocation model are adjusted accordingly according to the optimization objective and the constraint conditions.

[0077] ④ When it is judged that the relationship between the number of allocation entities and the number of target objects to be allocated is N = M, and the optimization objective is to minimize the loss, calculate the loss cost when the i - th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j - th (j = 0, 1, 2..., M - 1) target. The cost includes the energy consumption and ammunition consumption of the unmanned system; at this time, the optimization objective is The constraint conditions are The solution steps of the task allocation model are adjusted accordingly according to the optimization objective and the constraint conditions.

[0078] ⑤ When it is determined that the relationship between the number of allocation entities and the number of target objects to be allocated is N = M, and the optimization goal is to maximize the benefit, calculate the benefit when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task for the j-th (j = 0, 1, 2..., M - 1) target. The benefit is related to the value of the target. At this time, the optimization goal is The constraint condition is The solution steps of the task allocation model are adjusted accordingly according to the optimization goal and the constraint condition.

[0079] ⑥ When it is determined that the relationship between the number of allocation entities and the number of target objects to be allocated is N = M, and the optimization goal is to maximize the net benefit, calculate the net benefit when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task for the j-th (j = 0, 1, 2..., M - 1) target. The net benefit is equal to the benefit minus the loss. At this time, the optimization goal is The constraint condition is The solution steps of the task allocation model are adjusted accordingly according to the optimization goal and the constraint condition.

[0080] ⑦ When it is determined that the relationship between the number of allocation entities and the number of target objects to be allocated is N < M, and the optimization goal is to minimize the loss, calculate the loss cost when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task for the j-th (j = 0, 1, 2..., M - 1) target. The cost includes the energy consumption and ammunition consumption of the unmanned system. At this time, the optimization goal is The constraint condition is The solution steps of the task allocation model are adjusted accordingly according to the optimization goal and the constraint condition.

[0081] ⑧ When it is determined that the relationship between the number of allocation entities and the number of target objects to be allocated is N < M, and the optimization goal is to maximize the benefit, calculate the benefit when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task for the j-th (j = 0, 1, 2..., M - 1) target. The benefit is related to the value of the target. At this time, the optimization goal is The constraint condition is The solution steps of the task allocation model are adjusted accordingly according to the optimization goal and the constraint condition.

[0082] ⑨ When it is determined that the relationship between the number of allocation entities and the number of target objects to be allocated is N < M, and the optimization goal is to maximize the net benefit, calculate the net benefit when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task for the j-th (j = 0, 1, 2..., M - 1) target. The net benefit is equal to the benefit minus the loss. At this time, the optimization goal is The constraint condition is The solution steps of the task allocation model are adjusted accordingly according to the optimization objectives and constraint conditions.

[0083] The content of Slot 1 and Slot 4 will affect the optimization objectives and constraint conditions of task allocation. Generally speaking, the optimization objective of the reconnaissance task is to select the maximum benefit, the optimization objective of the tracking task is to select the maximum loss, and the optimization objective of the strike task is to select the maximum net benefit. By comprehensively applying speech recognition, semantic parsing, and unmanned system task allocation technologies, the present invention innovatively proposes a task-level structured instruction interaction method for unmanned system cluster task allocation, which is expected to make the command and control of unmanned system clusters simpler and more efficient, and greatly reduce the workload of unmanned system operators.

[0084] Any process or method description in the flowchart of the present invention or described in other ways herein can be understood as representing a module, segment, or part of code including one or more executable instructions for implementing a specific logical function or process, which can be implemented on any computer-readable medium for an instruction execution system, apparatus, or device. The computer-readable medium can be any medium including storage, communication, propagation, or transmission programs for use by an execution system, apparatus, or device, including read-only memory, magnetic disks, or optical discs, etc.

[0085] In the description of this specification, the description with reference to terms such as "embodiment", "example", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. In addition, those skilled in the art can combine or combine different embodiments or examples described in this specification and the features therein without conflict.

[0086] Although the above content has shown and described the embodiments of the present invention, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those of ordinary skill in the art can perform update operations such as changes, modifications, substitutions, and variations on the above embodiments within the scope of the present invention.

Claims

1. A structured instruction semantic interaction method for unmanned system cluster task allocation, characterized in that It includes the following steps: S1. Cluster task voice command recognition: Convert the cluster task voice command data issued by the operator into cluster task text commands through voice recognition technology; S2. Cluster task semantic parsing: Extract keywords from the cluster task text commands through natural language processing, and then generate corresponding cluster task structured commands through slot filling; S3. Cluster task assignment solution: Solve through matching the corresponding cluster task assignment model and algorithm to obtain the corresponding cluster task assignment result.

2. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 1, characterized in that The unmanned system cluster is a drone cluster.

3. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 1, characterized in that, In step S2, the slot of the unmanned system cluster task assignment structured command includes two-level nested slots, namely the peer slot and the dependency slot.

4. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 3, characterized in that, Taking the unmanned system cluster task assignment five-tuple as the slot paradigm for semantic interaction of task-level structured commands, the unmanned system cluster task assignment five-tuple includes the battlefield environment, the unmanned system cluster set, the target object set, the task type set, and the constraint conditions.

5. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 3, characterized in that The number of dependency slots of each peer slot can be dynamically increased or decreased according to the actual situation, and the value of each dependency slot is assigned or defaulted according to the issued voice command.

6. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 3, characterized in that In step S3, the general task assignment scheme for the unmanned system task-level command based on the auction algorithm is generated based on the auction algorithm.

7. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 6, characterized in that, The specific settings of the auction algorithm in the model for unmanned system cluster task assignment are as follows: N unmanned systems execute tasks on M targets. One unmanned system can only receive one task, and one task can only be assigned to one unmanned system; Each unmanned system has a corresponding benefit for executing the corresponding task, and each task has its own cost. The net benefit of the drone for executing the corresponding task is the benefit minus the cost. By optimizing the task assignment, the overall benefit of the unmanned system cluster for executing tasks is maximized.

8. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 7, characterized in that, In the model for unmanned system cluster task assignment, the definitions of the corresponding mathematical symbols are as follows: p ij : The cost when the i-th (i = 0, 1, 2..., N - 1) unmanned system executes a task on the j-th (j = 0, 1, 2..., M - 1) target; the cost includes energy consumption and / or ammunition consumption; a ij : The benefit when the i-th unmanned system executes a task on the j-th target; the benefit is related to the value of the target; V ij : Net benefit when the i-th unmanned system executes a mission on the j-th target, V ij = a ij - p ij ; Each task can be assigned to at most one unmanned system; Each unmanned system can only receive one task at most; x ij ∈ {0, 1} is a decision variable indicating whether the i-th unmanned system executes a task on the j-th target. If the i-th unmanned system executes a task on the j-th target, then x ij = 1. If the i-th unmanned system does not execute a task on the j-th target, then x ij = 0; 9. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 8, characterized in that In the model for unmanned system cluster task assignment, the optimization objectives include three types: minimum loss, maximum benefit, and maximum net benefit.

10. The structured instruction semantic interaction method for unmanned system cluster task allocation according to claim 9, characterized in that From the perspective of the comparison between the number of unmanned systems and the number of target objects, it is divided into three situations: N > M, N = M, and N < M. The comparison situation between the number of unmanned systems and the number of target objects and the three optimization objectives are cross-combined to achieve the matching of the unmanned system cluster task-level structured commands and the cluster task assignment scheme.

Images