Prediction method and device of multi-agent action system cooperative completion target, and medium
By performing binary modeling and implicit relationship modeling on the multi-agent action system, we predict the accessibility of the multi-agent system's cooperative completion goals, solving the problem that traditional methods are difficult to characterize the multi-agent action dependency, and achieving efficient and accurate accessibility analysis.
Patent Information
- Application Number
- CN202510594183.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-09
- Publication Date
- 2025-06-10
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
In multi-agent systems, it is a complex issue to determine whether the system can reach the target state from the initial state. Traditional methods are difficult to effectively characterize the complex dependencies between the actions of multi-agents, resulting in low accessibility analysis and insufficient accuracy.
By binary modeling of the multi-agent action system, the binary model is obtained, and the implicit relationship between the multi-agent action system is modeled according to the model to obtain the correlation relationship model. Then, based on the association relationship model and the goals to be completed by the goals to be completed by multiple agents, the accessibility of the goal of collaborative completion of the multi-agent action system is predicted, and the corresponding attributed reasoning process and strategies are obtained.
This method can clearly and accurately characterize the complex dependence relationship between multi-agent movements, improve the efficiency of multi-agent systems in the planning and decision-making process, and ensure the accuracy and efficiency of accessibility analysis.
Smart Images
Figure CN120124752A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a prediction method, device and medium for a multi-agent action system to collaboratively achieve a goal. Background Art
[0002] In a multi-agent system, multiple agents need to collaborate with each other to complete a specific task. However, determining whether the system can reach the target state from the initial state is a complex problem. Traditional methods often only stay at the representation of a single production rule when dealing with the dependencies between multi-agent actions, and cannot effectively characterize the complex dependencies between multi-agent actions, resulting in low efficiency and insufficient accuracy in solving the reachability problem.
[0003] With the widespread application of multi-agent systems in government services, industrial control, intelligent transportation, distributed robots and other fields, how to efficiently and accurately judge whether the multi-agent action system can collaborate to achieve the goal is an urgent problem to be solved. Summary of the invention
[0004] The technical task of the present invention is to provide a prediction method, device and medium for a multi-agent action system to collaboratively achieve a goal, so as to solve the problem of how to efficiently and accurately judge whether a multi-agent action system can collaboratively achieve a goal.
[0005] The technical task of the present invention is achieved in the following way: a prediction method for a multi-agent action system to collaboratively complete a goal, the method is as follows: Perform binary modeling on the multi-agent action system to obtain a binary model; According to the binary model, the implicit relationship between the multi-agent action systems is modeled to obtain the association relationship model; According to the association relationship model and the target to be completed by collaboration, predict the reachability of the multi-agent action system to collaboratively complete the target to be completed, and obtain the corresponding inference process; Develop a reductive reasoning strategy based on the reductive reasoning process.
[0006] As a preferred method, the binary modeling of the multi-agent action system is as follows: The prerequisite for the agent's action T is P, and the postcondition is Q, then T=(P,Q); Each precondition P corresponds to a production rule, where P={p 1 ,p 2 ,⋯,p n}; Q = {q 1 ,q 2 ,⋯,q m}, n and m are positive integers; The production rule structure corresponding to a single action is: if each condition in the premise condition P is met, the corresponding postcondition is generated, that is, each condition c∈P or Q corresponds to an independent production rule to form an atomic rule unit; for example, if p1˄p2˄⋯˄pn holds, then q1˄q2˄⋯˄qm holds.
[0007] As a preferred method, the implicit relationship between the multi-agent action system is modeled as follows: For two different agent action systems T 1 and T 2 , when the production expression T of a single action has been obtained 1 =(P 1 ,Q 1 ), T 2 =(P 2 ,Q 2 ); if q 1i =p 2j , then T 1 A postcondition of T 2 The premise is that there is an implication relation R(T1→T2) between the production rules of the actions; where P 1 ={p 11 ,p 12 ,⋯,p 1n}, Q 1 ={q 11 ,q 12 ,⋯,q 1m}, P 2 ={p 21 ,p 22 ,⋯,p 2k}, Q 2 ={q 21 ,q 22 ,⋯,q 2l}; i∈[1,m], j∈[1,k]; k and l are both positive integers; Describe the dependencies between multi-agent action systems one by one, and then obtain the production rule system of multi-agent action systems; Through the logical associations between rules (such as transitive closure), the global action dependency graph is dynamically generated.
[0008] As a preferred method, according to the association relationship model and the target to be completed by collaboration, the reachability of the multi-agent action system to collaboratively complete the target to be completed is predicted, and the corresponding conclusion reasoning process is obtained as follows: State encoding: If the action can occur in the initial state, then the production rule corresponding to the action in the initial state belongs to the initial state set I. If the conditions involved in the target state to be achieved by the multi-agent action system are G={g1 , g 2 , ⋯, g s}, then each production rule corresponding to a condition belongs to the target state set N. Incorporate the production rules corresponding to the conditions involved in the target state into the target state set N, and map the initial state set I and the target state N to the production rule set; Incorporate the negation ¬N of the state corresponding to the target state set N into the initial state set I to obtain the production rule set S = I ∪ ¬N corresponding to the current state set of the system; where ¬N is a symbol representing the negation of the state corresponding to the target state set N; Based on the production rule system of the multi-agent action system, select the production rules executable in the current state and update the production rule set corresponding to the current system state; when selecting rules, determine the triggerable production rules based on the preconditions already satisfied in the current system state, apply the triggerable production rules to the current state set, and update the conditions in the state set; Execute the resolution inference strategy on the rules in the production rule set S.
[0009] More preferably, the execution of the resolution inference strategy on the rules in the production rule set S is as follows: ① Suppose the production rule corresponding to an action is P → Q, and P holds in the current state, then Q is deduced; ② Suppose the production rule corresponding to an action is P 1 ∧P 2 →Q, and P 1 and P 2 hold in the current state, then Q is deduced; ③ Suppose the production rule corresponding to an action is P → Q, P holds in the current state, and Q is included in the target state N, then directly resolve Q in N. If all items have been completely resolved by resolution, the target state is reachable; otherwise, continue to execute step ②; ④ If none of the production rules of the multi-agent action system can be made to occur in the production rule set S corresponding to the current system state, and there are still the negations of the production rules in ¬N, it means that the target state is unreachable.
[0010] As a preference, the resolution inference strategy is formulated according to the resolution inference process as follows: Execute resolution inference according to the conditions of the current state and the target state, and gradually resolve the conditions; If the conditions of the target state are completely satisfied, it is determined that the target state is reachable; otherwise, it is determined to be unreachable.
[0011] An electronic device includes: a memory and at least one processor; Wherein, a computer program is stored on the memory; The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the prediction method for the multi-agent action system to cooperate to complete the target as described above.
[0012] A computer-readable storage medium stores a computer program, and the computer program can be executed by a processor to implement the prediction method for the multi-agent action system to cooperate to complete the target as described above.
[0013] The prediction method, device and medium for the multi-agent action system to cooperate to complete the target of the present invention have the following advantages: (1) Through the innovative action representation and the construction method of the production rule system, the present invention can clearly and accurately describe the complex dependence relationship between multi-agent actions. Compared with the traditional method, it is more comprehensive and in-depth in representing the multi-agent cooperation logic, providing a solid foundation for subsequent reachability analysis; (2) The present invention can efficiently determine the reachability of the multi-agent action system. By incorporating the negation of the target state into the initial state set and performing step-by-step reasoning and resolution based on the production rule system, it can quickly determine whether it is reachable from the initial state to the target state, greatly improving the efficiency of the multi-agent system in the planning and decision-making process; (3) The present invention has good versatility and scalability, and can be applied to various multi-agent system scenarios, such as distributed robot cooperation, intelligent transportation scheduling, industrial automation control and other fields, and can effectively improve the performance and reliability of the multi-agent system in practical applications; (4) By constructing a production rule system and the corresponding resolution reasoning strategy, the present invention clearly describes the dependence relationship between multi-agent actions, efficiently and accurately determines the reachability of the multi-agent action system from the initial state to the target state, thereby improving the cooperation efficiency and reliability of the multi-agent system in practical applications; (5) The present invention is used to solve the problem of determining the reachability from the initial state to the target state in the process of multiple agents cooperating to complete the target, effectively processes the complex relationship between multi-agent actions, and improves the cooperation efficiency and reliability of the multi-agent system in practical applications. Brief Description of the Drawings
[0014] The present invention will be further described below with reference to the accompanying drawings.
[0015] Attached Figure 1 is a flow block diagram of the prediction method for the multi-agent action system to cooperate to complete the target. Detailed Embodiments
[0016] The following provides a detailed description of the method, device, and medium for predicting the collaboration of a multi-agent action system to complete a target with reference to the accompanying drawings of the specification and specific embodiments.
[0017] Example 1: As shown in the appendix Figure 1 This example provides a method for predicting the collaboration of a multi-agent action system to complete a target, which is as follows: S1. Perform binary component modeling on the multi-agent action system to obtain a binary tuple model; S2. Based on the binary tuple model, model the implicative relationship between multi-agent action systems to obtain an association relationship model; S3. According to the association relationship model and the target to be collaboratively completed, predict the reachability of the multi-agent action system to collaboratively complete the target to be collaboratively completed, and obtain the corresponding resolution reasoning process; S4. Develop a resolution reasoning strategy based on the resolution reasoning process.
[0018] The binary component modeling of the multi-agent action system in step S1 of this example is specifically as follows: S101. If the precondition for the agent action T to occur is P and the postcondition is Q, then T = (P, Q); S102. Each precondition P corresponds to a production rule, where P = {p 1 , p 2 , ⋯, p n}; Q = {q 1 , q 2 , ⋯, q m}, and n and m are positive integers; S103. The production rule structure corresponding to a single action is: if each condition in the precondition P is satisfied, then the corresponding postcondition is generated, that is, each condition c ∈ P or Q corresponds to an independent production rule, forming an atomic rule unit; for example, if p1 ˄ p2 ˄ ⋯ ˄ pn holds, then q1 ˄ q2 ˄ ⋯ ˄ qm holds.
[0019] The modeling of the implicative relationship between multi-agent action systems in step S2 of this example is specifically as follows: S201. For two different agent action systems T 1 and T 2 , when the production representation of a single action has been obtained as T 1 = (P 1 , Q 1 ), T 2 = (P 2 , Q 2 ); if q 1i = p 2j , then it means that T 1A postcondition of is T 2 The precondition, at this time, there is an implication relationship R(T1→T2) between the production rules of actions; among them, P 1 ={p 11 , p 12 , ⋯, p 1n}}, Q 1 ={q 11 , q 12 , ⋯, q 1m}}, P 2 ={p 21 , p 22 , ⋯, p 2k}}, Q 2 ={q 21 , q 22 , ⋯, q 2l}}; i ∈ [1, m], j ∈ [1, k]; both k and l are positive integers; S202. Characterize the existing dependency relationships between multi-agent action systems one by one, and then obtain the production rule system of the multi-agent action system; S203. Dynamically generate a global action dependency graph through the logical association between rules (such as transitive closure).
[0020] In step S3 of this embodiment, according to the association relationship model and the target to be collaboratively completed, predict the reachability of the multi-agent action system to collaboratively complete the target to be collaboratively completed, and obtain the corresponding resolution reasoning process as follows: S301. State encoding: If the action can occur in the initial state, the production rule corresponding to the action in the initial state belongs to the initial state set I. If the conditions involved in the target state to be reached by the multi-agent action system are G = {g 1 , g 2 , ⋯, g s}}, then the production rule corresponding to each condition belongs to the target state set N. Incorporate the production rules corresponding to the conditions involved in the target state into the target state set N, and map the initial state set I and the target state N to the production rule set; S302. Incorporate the negation ¬N of the state corresponding to the target state set N into the initial state set I, and obtain the production rule set S = I ∪ ¬N corresponding to the current state set of the system; where ¬N is a symbol representing the negation of the state corresponding to the target state set N; S303. Select the production rules that can be executed in the current state by the production rule system based on the multi-agent action system, and update the set of production rules corresponding to the current system state; when selecting rules, determine the triggerable production rules according to the preconditions that have been satisfied in the current system state, apply the triggerable production rules to the current state set, and update the conditions in the state set. S304. Execute the resolution inference strategy on the rules in the production rule set S.
[0021] The specific method of executing the resolution inference strategy on the rules in the production rule set S in step S304 of this embodiment is as follows: ① Suppose the production rule corresponding to the action is P → Q, and P holds in the current state, then Q is obtained by resolution; for example, if there is a production rule "If agent A is at position X and has tool Y, then agent A can complete task Z", and in the current state, agent A is at position X and has tool Y, then it can be concluded by resolution that agent A can complete task Z. ② Suppose the production rule corresponding to the action is P 1 ∧P 2 →Q, and P 1 and P 2 hold in the current state, then Q is obtained by resolution. ③ Suppose the production rule corresponding to the action is P → Q, P holds in the current state, and the target state N contains Q, then directly resolve Q in N. If all items have been resolved and eliminated, the target state is reachable; otherwise, continue to execute step ②. ④ If none of the production rules of the multi-agent action system can be triggered in the production rule set S corresponding to the current system state, and there are still negations of the production rules in ¬N, it means that the target state is unreachable.
[0022] The specific method of formulating the resolution inference strategy according to the resolution inference process in step S4 of this embodiment is as follows: S401. According to the conditions of the current state and the target state, execute resolution inference and gradually eliminate the conditions. S402. If the conditions of the target state are fully satisfied, it is determined that the target state is reachable; otherwise, it is determined that it is unreachable.
[0023] Embodiment 2: This embodiment also provides an electronic device, including: a memory and at least one processor; Wherein, the memory stores computer execution instructions; The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the prediction method for multi-agent action systems to cooperate to complete the target in any embodiment of the present invention.
[0024] The processor can be a central processing unit (CPU), or can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), off-the-shelf programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor can be a microprocessor or the processor can also be any conventional processor, etc.
[0025] The memory can be used to store computer programs and / or modules. By running or executing the computer programs and / or modules stored in the memory, and calling the data stored in the memory, the processor realizes various functions of the electronic device. The memory mainly includes a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function, etc.; the data storage area can store data created according to the use of the terminal, etc. In addition, the memory can also include high-speed random access memory, and can also include non-volatile memory, such as a hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash memory card, at least one magnetic disk storage period, flash memory device, or other volatile solid-state storage devices.
[0026] Embodiment 3: This embodiment also provides a computer-readable storage medium, in which multiple instructions are stored. The instructions are loaded by the processor to enable the processor to execute the prediction method for the multi-agent action system to cooperate to complete the target in any embodiment of the present invention. Specifically, a system or device equipped with a storage medium can be provided. On this storage medium, software program codes for implementing the functions in any one of the above embodiments are stored, and the computer (or CPU or MPU) of the system or device reads and executes the program codes stored in the storage medium.
[0027] In this case, the program code read from the storage medium itself can implement the functions in any one of the above embodiments. Therefore, the program code and the storage medium storing the program code constitute a part of the present invention.
[0028] Embodiments of the storage medium for providing program codes include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RYM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Optionally, the program code can be downloaded from a server computer through a communication network.
[0029] In addition, it should be clear that not only can the actual operations be completed in part or in whole by executing the program codes read by the computer, but also by means of instructions based on the program codes, the operating system operating on the computer, etc., so as to realize the functions in any one of the above embodiments.
[0030] In addition, it can be understood that the program code read out from the storage medium is written into the memory provided in the expansion board inserted into the computer or into the memory provided in the expansion unit connected to the computer, and then based on the instructions of the program code, the CPU or the like installed on the expansion board or the expansion unit is made to execute part or all of the actual operations, thereby implementing the functions of any one of the above embodiments.
[0031] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A prediction method for a multi-agent action system to achieve a goal through collaboration, characterized in that: The method is as follows: Perform binary modeling on the multi-agent action system to obtain a binary model; According to the binary model, the implicit relationship between the multi-agent action systems is modeled to obtain the association relationship model; According to the association relationship model and the target to be completed by collaboration, predict the reachability of the multi-agent action system to collaboratively complete the target to be completed, and obtain the corresponding inference process; Develop a reductive reasoning strategy based on the reductive reasoning process.
2. The method for predicting the goal of collaborative completion of a multi-agent action system according to claim 1, characterized in that: The binary modeling of the multi-agent action system is as follows: The prerequisite for the agent's action T is P, and the postcondition is Q, then T=(P,Q); Each precondition P corresponds to a production rule, where P={p1,p2,⋯,p n }; Q = {q1,q2,⋯,q m }, n and m are positive integers; The production rule structure corresponding to a single action is: if each condition in the precondition P is met, the corresponding postcondition is generated, that is, each condition c∈P or Q corresponds to an independent production rule, forming an atomic rule unit.
3. The method for predicting the goal of collaborative completion of a multi-agent action system according to claim 1, characterized in that: The modeling of the implicit relationship between multi-agent action systems is as follows: For two different agent action systems T1 and T2, when the production expression of a single action is obtained, T1=(P1,Q1), T2=(P2,Q2); if q 1i =p 2j , it means that a postcondition of T1 is a precondition of T2. At this time, there is an implication relation R(T1→T2) between the production rules of actions. Among them, P1={p 11 ,p 12 ,⋯,p 1n }, Q1={q 11 ,q 12 ,⋯,q 1m }, P2={p 21 ,p 22 ,⋯,p 2k }, Q2={q 21 ,q 22 ,⋯,q 2l }; i∈[1,m], j∈[1,k]; k and l are both positive integers; Describe the dependencies between multi-agent action systems one by one, and then obtain the production rule system of multi-agent action systems; Through the logical association between rules, the global action dependency graph is dynamically generated.
4. The method for predicting the goal of collaborative completion of a multi-agent action system according to claim 1, characterized in that: According to the association relationship model and the target to be completed by collaboration, the reachability of the multi-agent action system to collaboratively complete the target to be completed is predicted, and the corresponding conclusion reasoning process is obtained as follows: State encoding: If the action can occur in the initial state, then the production rule corresponding to the action in the initial state belongs to the initial state set I. If the conditions involved in the target state to be achieved by the multi-agent action system are G={g1,g2,⋯,g s }, then the production rules corresponding to each condition belong to the target state set N, the production rules corresponding to the conditions involved in the target state are included in the target state set N, and the initial state set I and the target state N are mapped to the production rule set; The negation ¬N of the state corresponding to the target state set N is merged into the initial state set I, and the production rule set S=I∪¬N corresponding to the current state set of the system is obtained; where ¬N is a symbol, which represents the negation of the state corresponding to the target state set N; The production rule system based on the multi-agent action system selects the production rules that can be executed in the current state and updates the production rule set corresponding to the current system state; when selecting rules, it determines the triggerable production rules based on the prerequisites that have been met in the current system state, applies the triggerable production rules to the current state set, and updates the conditions in the state set; Execute the resolution reasoning strategy on the rules in the production rule set S.
5. The method for predicting the goal of collaborative completion of a multi-agent action system according to claim 4, characterized in that: The specific strategy for executing the reductive reasoning on the rules in the production rule set S is as follows: ① Suppose the production rule corresponding to the action is P→Q, and if P is true in the current state, then Q is obtained; ② Assume that the production rule corresponding to the action is P1∧P2→Q, and the current state has P1 and P2, then the result is Q; ③ Assume that the production rule corresponding to the action is P→Q, P is established in the current state, and the target state N contains Q, then directly resolve Q in N. If all items have been resolved, the target state is reachable, otherwise continue to execute step ②; ④ If the production rule set S corresponding to the current system state fails to make any production rule of the multi-agent action system occur, and there are still negations of the production rules in ¬N, it means that the target state is unreachable.
6. The method for predicting the goal of collaborative completion of a multi-agent action system according to claim 1, characterized in that: According to the inductive reasoning process, the inductive reasoning strategy is formulated as follows: According to the conditions of the current state and the target state, perform reductive reasoning and gradually resolve the conditions; If the conditions of the target state are fully satisfied, the target state is determined to be reachable, otherwise it is determined to be unreachable.
7. An electronic device, characterized in that: include: memory and at least one processor; Wherein, the memory stores a computer program; The at least one processor executes the computer program stored in the memory, so that the at least one processor executes the prediction method for the multi-agent action system to collaboratively achieve a goal as described in any one of claims 1 to 6.
8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which can be executed by a processor to implement a method for predicting a goal achieved by a multi-agent action system in collaboration as described in any one of claims 1 to 6.