A Trustworthy and Controllable Interaction Method and Device for Swarm Intelligence Agents

Abstract

This invention relates to the field of artificial intelligence technology. It provides a method and apparatus for trusted and controllable interaction among swarm intelligence agents. The method includes: sending task information about an agent to be evaluated performing the target task to a first target intelligence agent, based on a target task, so that the first target intelligence agent performs a trust assessment of the agent to be evaluated based on the task information and obtains an assessment result; the first target intelligence agent and the agent to be evaluated are agents under the target task; receiving the assessment result, determining the trust value of the agent to be evaluated at the next moment based on the assessment result and the trust value of the agent to be evaluated at the current moment; simultaneously determining the controllable value of the agent to be evaluated at the next moment based on the temporary controllable values ​​of the agent to be evaluated and the agents preceding it on the task execution path at the current moment, thereby ensuring the overall trustworthiness and controllability of the intelligence agent system that needs to complete the task at the next moment and improving the security of the intelligence agent system's execution behavior.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a reliable and controllable interaction method and apparatus for swarm intelligence agents. Background Technology

[0002] In the field of artificial intelligence, an agent is any computing system or entity that can perceive its environment through sensors (such as cameras, microphones, thermometers) or data interfaces (such as APIs, databases) and autonomously take actions through actuators based on the perceived information to achieve a preset goal.

[0003] The core of an intelligent agent lies in the cycle of "perception-thinking-action": it continuously acquires information from the environment, makes decisions (such as reasoning, planning or learning) using internal algorithms or models, and then outputs actions to change the environment, thereby gradually approaching and ultimately achieving its design goals.

[0004] Intelligent agents can be pure software programs or robots embedded in the physical world. Their level of "intelligence" depends on their decision-making ability, ranging from simple conditioned reflexes to complex learning and adaptation.

[0005] For example, self-driving cars use cameras and LiDAR to perceive roads, pedestrians, and vehicles, and then plan their routes to achieve intelligent navigation. Similarly, chatbots receive text input from users, perform resource searches, and provide intelligent responses.

[0006] However, such fully autonomous intelligent operation could lead to uncontrollable collective behavior of the intelligent agents, potentially causing harm and serious losses to humans or the task at hand. Therefore, trust between intelligent agents and control over each agent are particularly important and necessary in the face of behaviors with significant impact or harm. Summary of the Invention

[0007] To address the problems existing in the prior art, the present invention provides a reliable and controllable interaction method and apparatus for swarm intelligence agents.

[0008] This invention provides a trusted and controllable interaction method for swarm intelligence agents, comprising:

[0009] Based on the target task, a first target agent is identified, and task information for the agent to be evaluated to perform the target task is sent to the first target agent, so that the first target agent can perform a trust evaluation of the agent to be evaluated based on the task information and obtain an evaluation result; the first target agent and the agent to be evaluated are agents under the target task.

[0010] Receive the evaluation result, and determine the trust value of the agent to be evaluated at the next moment based on the evaluation result and the trust value of the agent to be evaluated at the current moment;

[0011] Determine the execution path of the target task, wherein the execution path is a path constructed by multiple agents performing task actions in sequence;

[0012] Based on the execution path, a second target agent is determined that precedes the agent to be evaluated;

[0013] Obtain the temporary controllable value of the second target agent at the current moment;

[0014] Based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time, the controllable value of the agent to be evaluated at the next time is determined.

[0015] According to the swarm intelligence agent trusted and controllable interaction method provided by the present invention, before the agents perform tasks, the method further includes:

[0016] Configure the initial controllable values ​​corresponding to the agent to be evaluated;

[0017] Based on the initial controllable value and the trust value of the agent to be evaluated at the current moment, the temporary controllable value of the agent to be evaluated at the current moment is obtained.

[0018] According to the present invention, a swarm intelligence agent trusted and controllable interaction method is provided, the method further includes:

[0019] When the trustworthiness value of the agent to be evaluated is less than the first threshold at the current moment, the input controllable value is used as the initial controllable value of the agent to be evaluated.

[0020] According to the swarm intelligence agent trusted and controllable interaction method provided by the present invention, before the agents perform tasks, the method further includes:

[0021] When the controllable value of the agent to be evaluated is 0 at the next moment, a task pause message and an authorization reminder message are issued.

[0022] According to the present invention, a swarm intelligence agent trusted and controllable interaction method is provided, the method further includes:

[0023] When the trust value decay time is triggered, the trust value of the agent to be evaluated at the current time is decayed.

[0024] According to the present invention, a swarm intelligence agent trusted and controllable interaction method is provided, the method further includes:

[0025] Send a query request for a trust value to a third target intelligent agent, so that the third target intelligent agent can provide its trust value at the current moment;

[0026] Based on the trustworthiness value of the third target agent at the current moment, determine the trust result for the third target agent.

[0027] The present invention also provides a trusted and controllable interaction device for swarm intelligence agents, comprising:

[0028] The sending module is configured to determine a first target intelligent agent based on the target task, and send task information of the intelligent agent to be evaluated performing the target task to the first target intelligent agent, so that the first target intelligent agent can perform a trust evaluation of the intelligent agent to be evaluated based on the task information and obtain an evaluation result; the first target intelligent agent and the intelligent agent to be evaluated are intelligent agents under the target task;

[0029] A trust determination module is used to receive the evaluation result and, based on the evaluation result and the trust value of the agent to be evaluated at the current moment, determine the trust value of the agent to be evaluated at the next moment.

[0030] A generation module is used to determine the execution path of the target task, wherein the execution path is a path constructed by multiple agents performing task actions in sequence;

[0031] The selection module is used to determine a second target agent located before the agent to be evaluated, based on the execution path.

[0032] The acquisition module is used to acquire the temporary controllable value of the second target agent at the current moment;

[0033] The controllable determination module is used to determine the controllable value of the agent to be evaluated at the next time step based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time step.

[0034] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement any of the above-described swarm intelligence agent trusted and controllable interaction methods.

[0035] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements any of the above-described swarm intelligence agent trusted and controllable interaction methods.

[0036] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements any of the above-described swarm intelligence agent trusted and controllable interaction methods.

[0037] This invention provides a trusted and controllable interaction method and apparatus for swarm intelligence agents. The method involves an agent to be evaluated sending task information to other agents working together to complete the task. Each agent then performs a trust assessment based on the task information. The agent to be evaluated determines its trust value for the next moment based on the assessment results and its current trust value. Simultaneously, based on the temporary controllable values ​​of the agents along the task execution path and those preceding it at the current moment, the controllable value for the agent to be evaluated for the next moment is determined. This ensures the overall trustworthiness and controllability of the agent system that needs to complete the task in the next moment, thereby improving the security of the agent system's execution behavior. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0039] Figure 1 This is a flowchart illustrating the trusted and controllable interaction method for swarm intelligence agents provided by the present invention.

[0040] Figure 2 This is a schematic diagram of the structure of the swarm intelligence agent trusted and controllable interaction device provided by the present invention.

[0041] Figure 3 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0043] Figure 1 This diagram illustrates a flowchart of a trusted and controllable interaction method for swarm intelligence agents provided by the present invention. (See attached diagram.) Figure 1 The method includes the following steps:

[0044] Step 11: Based on the target task, determine the first target agent and send the task information of the agent to be evaluated to perform the target task to the first target agent, so that the first target agent can perform a trust evaluation of the agent to be evaluated based on the task information and obtain the evaluation result; the first target agent and the agent to be evaluated are agents under the target task.

[0045] Step 12: Receive the evaluation results, and determine the trust value of the agent to be evaluated at the next moment based on the evaluation results and the trust value of the agent to be evaluated at the current moment.

[0046] Step 13: Determine the execution path of the target task. The execution path is a path constructed by multiple agents executing task actions in sequence.

[0047] Step 14: Based on the execution path, determine the second target agent located before the agent to be evaluated.

[0048] Step 15: Obtain the temporary controllable value of the second target agent at the current moment;

[0049] Step 16: Based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time, determine the controllable value of the agent to be evaluated at the next time.

[0050] Regarding steps 11 to 16, it should be noted that in the field of artificial intelligence, an intelligent agent refers to any computing system or entity that can perceive its environment through sensors (such as cameras, microphones, thermometers) or data interfaces (such as APIs, databases) and autonomously take actions through actuators based on the perceived information to achieve a preset goal.

[0051] The core of an intelligent agent lies in the cycle of "perception-thinking-action": it continuously acquires information from the environment, makes decisions (such as reasoning, planning or learning) using internal algorithms or models, and then outputs actions to change the environment, thereby gradually approaching and ultimately achieving its design goals.

[0052] Intelligent agents can be pure software programs or robots embedded in the physical world. Their level of "intelligence" depends on their decision-making ability, ranging from simple conditioned reflexes to complex learning and adaptation.

[0053] For example, self-driving cars use cameras and LiDAR to perceive roads, pedestrians, and vehicles, and then plan their routes to achieve intelligent navigation. Similarly, chatbots receive text input from users, perform resource searches, and provide intelligent responses.

[0054] With the development of AI technologies, exemplified by large-scale models, model-driven agents are now capable of completing complex tasks and collaborating across multiple agents. Consequently, communication protocols between agents have emerged, namely the A2A protocol. A2A stands for Agent-to-Agent, and it refers to a set of rules, standards, and conventions governing communication and interaction between agents in a multi-agent system (MAS) for collaboration, negotiation, competition, or coordination. In the absence of centralized control, the A2A protocol is the cornerstone for the orderly operation of agents in a multi-agent system. It enables previously independent agents to form an efficient and organic whole, and is one of the key technologies enabling distributed artificial intelligence.

[0055] The A2A protocol defines a comprehensive set of security mechanisms, employing TLS end-to-end encryption, DID, and JWS digital signatures to achieve two-way authentication and message integrity verification. This, combined with short-lived capability tokens and a least privilege policy, prevents unauthorized calls. Agent cards are enforced with versioning and signing, and CI / CD and GitOps ensure updates are tamper-proof. Upon termination, tokens are automatically revoked, connections are closed, and centralized auditing is implemented to prevent credential remnants. While these security mechanisms guarantee the correctness, security, and integrity of communication between agents...

[0056] However, such fully autonomous intelligent operation could lead to uncontrollable collective behavior of the intelligent agents, potentially causing harm and serious losses to humans or the task at hand. Therefore, trust between intelligent agents and control over each agent are particularly important and necessary in the face of behaviors with significant impact or harm.

[0057] Therefore, it is necessary to achieve more precise control over the individual agents in a multi-agent system when they execute collective behaviors. This invention incorporates trusted and controllable technology into the A2A protocol, enabling the agent swarm to quantitatively acquire mutual trust when completing a task, thereby ensuring the overall security of agent behavior control.

[0058] In this invention, under the constraints of the A2A protocol, agents in a multi-agent system performing a common task communicate and interact. During this communication, the trustworthiness and even controllability of the agents need to be considered. Regarding trustworthiness, this invention adds a trust value to the A2A protocol. The trust value reflects the degree of trustworthiness of the agent's actions. Regarding controllability, this invention adds a controllability value to the A2A protocol. The controllability value reflects the degree to which human intervention is required by the agent.

[0059] To address this, the multi-agent system of this invention, in order to accomplish a collective action target, can identify the individual agents required to complete that target task. That is, each agent can know which agents will cooperate with it to complete the target task.

[0060] In this invention, the trust value is a systematic and dynamically evolving value evaluation system. Its core objective is to accurately quantify and continuously update the trustworthiness of each agent. To this end, each agent in a multi-agent system that jointly completes a target task needs to update its own trust value after each task execution. The trust value of an agent is a decimal between [0, 1], with higher values ​​indicating greater trustworthiness. 0 represents completely untrustworthy, and 1 represents completely trustworthy. The trust value can be stored in the agent's Card.

[0061] In this invention, each agent in a multi-agent system that works together to complete a target task can be referred to as an agent to be evaluated during the process of updating its trust value.

[0062] The agent to be evaluated can identify the first target agent based on the target task. Both the first target agent and the agent to be evaluated are agents belonging to the target task. That is, the agent to be evaluated needs to identify other agents that will jointly complete the target task.

[0063] An agent's trustworthiness value is dynamically adjusted based on its behavioral history, reflecting its reliability. The higher an agent's trustworthiness value, the more easily it is trusted by other agents.

[0064] At this point, the agent to be evaluated sends task information about its performance on the target task to the first target agent. This task information represents the performance status of the agent to be evaluated during the execution of the target task. This performance status can be analyzed by the first target agent to determine the level of trust in the agent to be evaluated, i.e., to assign a trust score to the agent to be evaluated.

[0065] In this invention, the first target agent can perform a trust assessment on the agent to be evaluated based on task information and obtain an assessment result. Then, the first target agent feeds back the assessment result to the agent to be evaluated.

[0066] In this invention, the agent to be evaluated needs to update its trust value after completing a task (i.e., executing the task at the current moment). At this time, the agent to be evaluated receives the evaluation result and determines its trust value at the next moment based on the evaluation result and the agent's trust value at the current moment.

[0067] In this invention, the intelligent agent updates its own trust value in real time. When the task is executed in the next moment, each intelligent agent can view the trust values ​​of other intelligent agents by sharing the trust value. Other intelligent agents can know what kind of quantitative trust value the intelligent agent they are using has, and select the intelligent agent based on the trust value, thereby ensuring the overall trust level of the intelligent agent system that needs to complete the task in the next moment.

[0068] The application of controllable values ​​is mainly reflected in the control process of the execution path of a multi-agent system executing a task. An agent's controllable value is a decimal between [0, 1], with higher values ​​indicating greater controllability. 0 represents completely uncontrollable, and 1 represents completely controllable. Controllable values ​​can be stored in the agent's Card. Each agent has two controllable values: an initial controllable value and a temporary controllable value. When an agent begins executing a task action on the execution path, the initial controllable value is incorporated into the calculation of the agent's temporary controllable value, which becomes the controllable value after the task action is executed. If an agent needs to execute multiple task actions on the execution path, the temporary controllable value at each execution of a task action is incorporated into the calculation of the next temporary controllable value. In other words, as task actions are executed repeatedly, the agent's temporary controllable value can be modified multiple times, but the initial controllable value remains unchanged. The initial controllable value can be incorporated into the calculation of the temporary controllable value again when the agent executes a task action on the next task's execution path. When an agent's temporary controllable value falls below a certain threshold, the agent must report to the user and obtain permission to execute. Once the user's instruction is received, the agent becomes fully controllable again.

[0069] Therefore, the agent to be evaluated can determine the execution path of the target task based on the target task. This execution path is constructed by multiple agents working together to complete the target task, executing task actions sequentially. The agent to be evaluated is also part of this execution path.

[0070] Based on the execution path, a second target agent is identified that precedes the agent to be evaluated. From the perspective of sequential execution, all agents that precede the agent to be evaluated on the execution path are represented as the second target agent.

[0071] Agents that complete their actions along the execution path will receive a temporary controllable value, meaning that the agent to be evaluated can obtain the temporary controllable value of the second target agent at the current moment.

[0072] Based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time, the controllable value of the agent to be evaluated at the next time step is determined.

[0073] The swarm intelligence agent trusted and controllable interaction method provided by this invention involves the agent to be evaluated sending task information to the agents jointly completing the task. The agents then perform a trust assessment of the agent to be evaluated based on the task information. The agent to be evaluated then determines its trust value for the next moment based on the assessment result and its current trust value. Simultaneously, based on the temporary controllable values ​​of the agents to be evaluated along the task execution path and the agents preceding the agent to be evaluated at the current moment, the controllable value of the agent to be evaluated for the next moment is determined. This method ensures the overall trustworthiness and controllability of the agent system that needs to complete the task in the next moment, thereby improving the security of the agent system's execution behavior.

[0074] In a further step of the above method, the trust value setting of the agent includes an initial trust value setting and a trust value update rule.

[0075] An initial trust value is assigned to the agent. The trust value is a decimal between [0, 1], so the initial trust value is neutral, such as 0.5. The initial trust value can also be adjusted based on the agent's metadata (such as creator and type).

[0076] In this invention, the agent to be evaluated determines its trustworthiness at the next moment by using a trustworthiness update rule based on the evaluation result and the trustworthiness value of the agent to be evaluated at the current moment.

[0077] Trustworthy value rules include weighted average update method and incremental update method.

[0078] Weighted average update method:

[0079] Let Agent B's current trust value be... After interacting with Agent A, Agent A provides a score s ( Then the update formula is:

[0080]

[0081] in, Forgetting factor ( This involves controlling the weight of historical data to ensure a smooth change in trustworthiness values.

[0082] Incremental update method:

[0083] In this invention, based on the task information, it can be determined whether the agent to be evaluated and the first target agent have successfully cooperated in completing the target task. If the cooperation is successful, it indicates that the task is successful, and the evaluation result fed back by the first target agent is that the task is successful.

[0084] If the task is successful, the trustworthiness of the agent being evaluated increases. If the task fails, the trust value of the agent to be evaluated decreases. The updated formula is:

[0085]

[0086] When the task succeeds... When the task fails, δ is a preset fixed value, but it can be adjusted according to the importance of the task.

[0087] In a further step of the above method, the agent to be evaluated collects controllable values ​​from multiple agents and uses the minimum value to represent all controllable values.

[0088] .

[0089] in, These are the controllable values ​​input to the agent being evaluated. Let be the controllable value of the i-th agent.

[0090] The controllable values ​​of the agent to be evaluated at the next time step will be updated as follows:

[0091] .

[0092] in, This represents the controllable value at the next moment. It is the controllable value accumulation factor, which reflects the impact of the controllable values ​​of other agents on the agent being evaluated.

[0093] Furthermore, the initial controllable value corresponding to the agent to be evaluated is configured, and the temporary controllable value of the agent to be evaluated at the current moment is obtained based on the initial controllable value and the trust value of the agent to be evaluated at the current moment.

[0094] Furthermore, the controllability value of the agent is set to 0 or 1, representing completely uncontrollable or completely controllable, respectively. For an agent with a controllability value of 0, interaction with the user is mandatory to obtain the right to execute the task. When the controllability value of the agent to be evaluated equals 0 at the next moment, a task pause message and authorization reminder message are issued to ensure the safety of the agent's execution behavior.

[0095] When the trust value of the agent to be evaluated is less than the first threshold at the current moment, the input controllable value is used as the initial controllable value of the agent to be evaluated. In other words, for some agents with low trust values, the user can set a lower controllable value.

[0096] In this invention, the trustworthiness value and controllability value of an intelligent agent are correlated. For an untrustworthy intelligent agent, users often prefer to focus on its performance. Therefore, it is set that the controllability value of an untrustworthy intelligent agent should also be reduced.

[0097] .

[0098] in, It is a controllable value. It is the influence factor of the trust value on the controllable value, where C is the trust value.

[0099] In a further step of the above method, when the trust value decays, the trust value of the agent to be evaluated at the current moment is decayed. That is, to ensure that the trust value reflects the agent's recent actions, a time decay mechanism for the trust value is established:

[0100] Every time period T (e.g., daily), the trustworthiness of all agents decays:

[0101]

[0102] in, Let be the trustworthiness value of the i-th agent at time t+1. It is a decay factor that causes the trustworthiness value to decrease slowly, maintaining good behavior.

[0103] In a further step of the above method, when an agent (such as the agent to be evaluated) needs to match other agents to complete a target task, the agent sends a query request for a trustworthy value to a third target agent, so that the third target agent can provide its trustworthy value at the current moment. In this invention, the third target agent is any agent whose trustworthy value has been requested by the agent.

[0104] The agent to be evaluated determines its trust in the third target agent based on the trust value of the third target agent at the current moment.

[0105] For example, when agent A needs to assess whether it trusts agent B, it can query agent B's trustworthiness value. .if If the trust threshold T is met, then agent A trusts agent B; otherwise, it does not trust it. Therefore, the trust result is either trust or distrust.

[0106] In addition, when multiple agents form a group, agent A collects the trust values ​​of multiple agents, calculates a weighted average, obtains a comprehensive trust value, and makes an overall judgment on the trust results of multiple agents.

[0107] The following describes the swarm intelligence agent trusted and controllable interaction device provided by the present invention. The swarm intelligence agent trusted and controllable interaction device described below and the swarm intelligence agent trusted and controllable interaction method described above can be referred to in correspondence.

[0108] Figure 2 This diagram illustrates the structure of a trusted and controllable interaction device for swarm intelligence agents provided by the present invention. (See attached diagram.) Figure 2 The device includes a sending module 21, a reliable determination module 22, a generation module 23, a selection module 24, an acquisition module 25, and a controllable determination module 26, wherein:

[0109] The sending module is used to determine a first target intelligent agent based on the target task, and send task information of the intelligent agent to be evaluated to perform the target task to the first target intelligent agent, so that the first target intelligent agent can perform a trust evaluation of the intelligent agent to be evaluated based on the task information and obtain the evaluation result; the first target intelligent agent and the intelligent agent to be evaluated are intelligent agents under the target task;

[0110] The determination module is used to receive the evaluation results and, based on the evaluation results and the trustworthiness value of the agent to be evaluated at the current moment, determine the trustworthiness value of the agent to be evaluated at the next moment.

[0111] The generation module is used to determine the execution path of the target task. The execution path is a path constructed by multiple agents performing task actions in sequence.

[0112] The selection module is used to determine the second target agent that precedes the agent to be evaluated, based on the execution path.

[0113] The acquisition module is used to acquire the temporary controllable values ​​of the second target agent at the current moment;

[0114] The controllable determination module is used to determine the controllable value of the agent to be evaluated at the next time step based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time step.

[0115] Since the principle of the swarm intelligence agent trusted and controllable interaction device in this embodiment is the same as that of the swarm intelligence agent trusted and controllable interaction method in the above embodiment, more detailed explanations will not be repeated here.

[0116] It should be noted that, in the embodiments of the present invention, the relevant functional modules can be implemented by a hardware processor.

[0117] The swarm intelligence agent trusted and controllable interaction device provided by this invention sends task information of the agent to be evaluated to other agents working together to complete the task. The agents then perform a trust assessment of the agent to be evaluated based on the task information. The agent to be evaluated then determines its trust value for the next moment based on the assessment result and its current trust value. Simultaneously, based on the temporary controllable values ​​of the agents to be evaluated along the task execution path and the agents preceding it at the current moment, the controllable value of the agent to be evaluated for the next moment is determined. This ensures the overall trustworthiness and controllability of the agent system that needs to complete the task in the next moment, thereby improving the security of the agent system's execution behavior.

[0118] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include: a processor 31, a communication interface 32, a memory 33, and a communication bus 34, wherein the processor 31, the communication interface 32, and the memory 33 communicate with each other through the communication bus 34. The processor 31 can call logical instructions in the memory 33 to execute a swarm intelligence agent trusted and controllable interaction method. This method includes: determining a first target intelligence agent based on a target task; sending task information about the intelligence agent to be evaluated performing the target task to the first target intelligence agent, so that the first target intelligence agent can perform a trust evaluation of the intelligence agent to be evaluated based on the task information and obtain an evaluation result; the first target intelligence agent and the intelligence agent to be evaluated are intelligence agents under the target task; receiving the evaluation result; determining the trust value of the intelligence agent to be evaluated at the next time step based on the evaluation result and the trust value of the intelligence agent to be evaluated at the current time step; determining the execution path of the target task, where the execution path is a path constructed by multiple intelligence agents executing task actions sequentially; determining a second target intelligence agent located before the intelligence agent to be evaluated based on the execution path; obtaining the temporary controllable value of the second target intelligence agent at the current time step; and determining the controllable value of the intelligence agent to be evaluated at the next time step based on the temporary controllable values ​​of the intelligence agent to be evaluated and the second target intelligence agent at the current time step, respectively.

[0119] Furthermore, the logical instructions in the aforementioned memory 33 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0120] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the swarm intelligence agent trusted and controllable interaction method provided by the above methods. The method includes: determining a first target intelligence agent according to a target task; sending task information of the intelligence agent to be evaluated to perform the target task to the first target intelligence agent, so that the first target intelligence agent can perform a trust evaluation of the intelligence agent to be evaluated according to the task information and obtain an evaluation result; the first target intelligence agent and the intelligence agent to be evaluated are intelligence agents under the target task; receiving the evaluation result; determining the trust value of the intelligence agent to be evaluated at the next time step according to the evaluation result and the trust value of the intelligence agent to be evaluated at the current time step; determining the execution path of the target task, the execution path being a path constructed by multiple intelligence agents performing task actions in sequence; determining a second target intelligence agent located before the intelligence agent to be evaluated according to the execution path; obtaining the temporary controllable value of the second target intelligence agent at the current time step; and determining the controllable value of the intelligence agent to be evaluated at the next time step according to the temporary controllable values ​​of the intelligence agent to be evaluated and the second target intelligence agent at the current time step, respectively.

[0121] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon. When executed by a processor, the computer program implements a swarm intelligence agent trusted and controllable interaction method provided by the above methods. The method includes: determining a first target intelligence agent according to a target task; sending task information of an intelligence agent to be evaluated performing the target task to the first target intelligence agent, so that the first target intelligence agent performs a trust evaluation of the intelligence agent to be evaluated based on the task information and obtains an evaluation result; the first target intelligence agent and the intelligence agent to be evaluated are intelligence agents under the target task; receiving the evaluation result; determining the trust value of the intelligence agent to be evaluated at the next time step based on the evaluation result and the trust value of the intelligence agent to be evaluated at the current time step; determining the execution path of the target task, wherein the execution path is a path constructed by multiple intelligence agents performing task actions in sequence; determining a second target intelligence agent located before the intelligence agent to be evaluated based on the execution path; obtaining the temporary controllable value of the second target intelligence agent at the current time step; and determining the controllable value of the intelligence agent to be evaluated at the next time step based on the temporary controllable values ​​of the intelligence agent to be evaluated and the second target intelligence agent at the current time step, respectively.

[0122] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0123] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A trusted and controllable interaction method for swarm intelligence agents, characterized in that, include: Based on the target task, a first target agent is identified, and task information for the agent to be evaluated to perform the target task is sent to the first target agent, so that the first target agent can perform a trust evaluation of the agent to be evaluated based on the task information and obtain an evaluation result; the first target agent and the agent to be evaluated are agents under the target task. Receive the evaluation result, and determine the trust value of the agent to be evaluated at the next moment based on the evaluation result and the trust value of the agent to be evaluated at the current moment; Determine the execution path of the target task, wherein the execution path is a path constructed by multiple agents performing task actions in sequence; Based on the execution path, a second target agent is determined that precedes the agent to be evaluated; Obtain the temporary controllable value of the second target agent at the current moment; Based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time, the controllable value of the agent to be evaluated at the next time is determined. Before the agent performs the task, the method further includes: Configure the initial controllable values ​​corresponding to the agent to be evaluated; Based on the initial controllable value and the trust value of the agent to be evaluated at the current moment, the temporary controllable value of the agent to be evaluated at the current moment is obtained.

2. The swarm intelligence agent trusted and controllable interaction method according to claim 1, characterized in that, The method further includes: When the trustworthiness value of the agent to be evaluated is less than the first threshold at the current moment, the input controllable value is used as the initial controllable value of the agent to be evaluated.

3. The swarm intelligence agent trusted and controllable interaction method according to claim 1, characterized in that, Before the agent performs the task, the method further includes: When the controllable value of the agent to be evaluated is 0 at the next moment, a task pause message and an authorization reminder message are issued.

4. The swarm intelligence agent trusted and controllable interaction method according to claim 1, characterized in that, The method further includes: When the trust value decay time is triggered, the trust value of the agent to be evaluated at the current time is decayed.

5. The swarm intelligence agent trusted and controllable interaction method according to claim 1, characterized in that, The method further includes: Send a query request for a trust value to a third target intelligent agent, so that the third target intelligent agent can provide its trust value at the current moment; Based on the trustworthiness value of the third target agent at the current moment, determine the trust result for the third target agent.

6. A swarm intelligence agent trusted and controllable interaction device based on any one of the swarm intelligence agent trusted and controllable interaction methods of claims 1-5, characterized in that, include: The sending module is configured to determine a first target intelligent agent based on the target task, and send task information of the intelligent agent to be evaluated performing the target task to the first target intelligent agent, so that the first target intelligent agent can perform a trust evaluation of the intelligent agent to be evaluated based on the task information and obtain an evaluation result; the first target intelligent agent and the intelligent agent to be evaluated are intelligent agents under the target task; A trust determination module is used to receive the evaluation result and, based on the evaluation result and the trust value of the agent to be evaluated at the current moment, determine the trust value of the agent to be evaluated at the next moment. A generation module is used to determine the execution path of the target task, wherein the execution path is a path constructed by multiple agents performing task actions in sequence; The selection module is used to determine a second target agent located before the agent to be evaluated, based on the execution path. The acquisition module is used to acquire the temporary controllable value of the second target agent at the current moment; The controllable determination module is used to determine the controllable value of the agent to be evaluated at the next time step based on the temporary controllable values ​​of the agent to be evaluated and the second target agent at the current time step.

7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the swarm intelligence agent trusted and controllable interaction method as described in any one of claims 1-5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the swarm intelligence agent trusted and controllable interaction method as described in any one of claims 1-5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the swarm intelligence agent trusted and controllable interaction method as described in any one of claims 1 to 5.

Images