A communication method and system based on multi-agent semantic collaborative evolution

Abstract

The embodiment of the application relates to the technical field of multi-agent communication, and specifically discloses a communication method and system based on multi-agent semantic cooperative evolution. The embodiment of the application initializes semantic states for multiple agents; locally updates the semantic states; when a communication triggering condition is met, corresponding agents generate communication messages according to semantic change amounts; cooperatively updates the semantic states; and dynamically adjusts the communication triggering condition, communication frequency and communication content selection strategy of subsequent time steps. Through semantic state modeling and a cooperative evolution mechanism, the embodiment of the application can depict semantic interaction relationships between agents, enable semantic information to continuously evolve at a group level along with a cooperation process, combine communication strategies with semantic cooperative evolution states, realize adaptive adjustment of communication parameters, thereby reducing communication redundancy, improving semantic consistency, and enhancing the cooperation efficiency and operation stability of multi-agents in a complex dynamic environment.

Description

Technical Field

[0001] This invention belongs to the field of multi-agent communication technology, and in particular relates to a communication method and system based on multi-agent semantic cooperative evolution. Background Technology

[0002] In existing technologies, multiple agents typically achieve information sharing and collaborative control through communication mechanisms, primarily based on traditional data communication models or their improvements. These solutions usually encode the perception, state, or decision information generated by the agents and send it to other agents or a central node via a pre-defined communication link to support collaborative task execution among multiple agents. In this process, communication design focuses on channel modeling, bandwidth allocation, scheduling strategies, and ensuring the reliability of information transmission. With the increase in the number of agents and the complexity of tasks, some existing technologies have introduced learning-based methods to optimize the communication process. For example, reinforcement learning or game theory strategies can be used to adjust communication frequency, transmission timing, or information selection to reduce communication load and improve overall system performance. These solutions typically use communication parameters as optimization targets, training to obtain communication strategies adapted to specific environments or tasks. On the other hand, semantic communication technologies proposed in recent years attempt to break through the traditional bit-level communication paradigm by extracting semantic features of information at the sending end and performing semantic reconstruction at the receiving end, thereby reducing the amount of data transmitted while ensuring task effectiveness. Existing semantic communication schemes mostly employ predefined or trained semantic representation models to map raw information into a semantic space before transmission. The receiving end then recovers the required semantic content based on the corresponding model to serve applications such as object detection, state estimation, or task decision-making. In multi-agent scenarios, some existing technologies combine semantic communication with multi-agent systems, enabling agents to interact based on semantic-level information. However, these schemes typically treat the semantic processing of each agent as a relatively independent unit. The updating and use of semantic representations during communication mainly revolve around individual agents, lacking a unified modeling of semantic interaction relationships at the group level. Furthermore, the updating methods of semantic representations or semantic models in existing technologies often rely on local data or fixed strategies, making it difficult to reflect the overall evolutionary characteristics of semantics over time and interaction relationships during multi-agent collaboration.

[0003] Although existing multi-agent communication and semantic communication technologies have improved the information exchange efficiency of distributed systems to some extent, they still reveal several shortcomings in complex dynamic environments and long-term collaborative tasks, limiting their application effectiveness and scalability in multi-agent systems. These shortcomings are mainly reflected in the following aspects: (1) Existing multi-agent communication technologies generally treat communication objects as independent and discrete information units. The communication process mainly revolves around the integrity and efficiency of a single information transmission, lacking systematic modeling of long-term semantic relationships between agents. In the process of multi-agent collaboration, the information content that each agent is concerned with is often highly related and changes continuously with the progress of the task. However, existing technologies are unable to characterize such semantic relationships across agents and time scales, resulting in a large amount of repeated transmission and semantic redundancy in the communication process, which reduces the overall communication efficiency. (2) Existing semantic communication methods mostly focus on semantic extraction and reconstruction under a single communication subject or fixed communication relationship, and usually assume that the semantic representation model remains relatively stable during the communication process. In multi-agent systems, different agents have different environmental states, perception capabilities, and task roles, and semantic understanding has obvious individual differences. Existing technologies lack effective mechanisms to dynamically coordinate the semantic differences between different agents, which can easily lead to inconsistent semantic understanding, thereby affecting the accuracy of collaborative decision-making and system stability; (3) Although some existing technologies have introduced learning or adaptive mechanisms to optimize communication parameters, their optimization targets are mostly focused on low-level indicators such as communication frequency, bandwidth usage, or transmission delay, and they fail to deeply couple the communication strategy with the semantic state changes between multiple agents. Such solutions usually separate the adjustment of the communication strategy from the evolution of semantic content, making it difficult for the communication process to be adjusted in conjunction with the cooperation relationship of multiple agents and the requirements of semantic consistency, thus limiting the adaptability of the communication system to complex collaborative tasks; (4) In multi-agent collaborative tasks, the interaction relationships and collaboration patterns between agents are usually dynamic and uncertain, and semantic information at the group level exhibits the characteristic of gradual evolution over time. However, existing technologies mostly rely on local observations or static rules for semantic updates, lacking a unified description and control mechanism for the collaborative evolution of semantics in a multi-agent group. This makes it difficult to guarantee the consistency and stability of the semantic evolution direction, which may lead to a decrease in collaboration efficiency or fluctuations in system performance. Summary of the Invention

[0004] The purpose of this invention is to provide a communication method and system based on multi-agent semantic cooperative evolution, aiming to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the embodiments of the present invention provide the following technical solutions: A communication method based on multi-agent semantic cooperative evolution, the method specifically includes the following steps: During system startup or task initiation, semantic states are initialized for multiple agents; At each time step, the multiple agents update their semantic state locally based on their real-time perception results, internal state changes, and task progress, thus obtaining the updated semantic state. Each of the aforementioned agents performs a phase comparison and a neighbor comparison of the semantic state, calculates the amount of semantic change and the degree of semantic difference with neighboring agents, and determines whether the communication triggering condition is met. When the communication triggering condition is met, the corresponding agent generates a communication message based on the semantic change and sends the communication message to one or more neighboring agents; After receiving a communication message, multiple intelligent agents combine the communication message with their own semantic state and update the semantic state in a collaborative manner according to a preset semantic collaborative evolution rule. Based on the updated semantic state, the semantic consistency among the corresponding multiple agents is evaluated, and the communication triggering conditions, communication frequency, and communication content selection strategy for subsequent time steps are dynamically adjusted.

[0006] As a further limitation of the technical solution of the embodiment of the present invention, in the process of initializing semantic states for multiple agents during the system startup or task start phase, the multiple agents construct corresponding semantic states based on corresponding initial perception information, task objectives or prior knowledge.

[0007] As a further limitation of the technical solution of this embodiment of the invention, the semantic state is represented as being located in the semantic space. The vector in the vector is expressed as: ; in, Semantic representation dimension Indicates by A vector space consisting of n real numbers Represents a time step. For time step Time The semantic state corresponding to each agent; Based on the semantic states corresponding to multiple agents, a group semantic state set is introduced to support semantic interaction among multiple agents, expressed as: ; in, Indicates shared ownership An intelligent agent.

[0008] As a further limitation of the technical solution of this embodiment of the invention, each of the plurality of intelligent agents performs stage comparison and neighbor comparison on semantic state, calculates the amount of semantic change and the degree of semantic difference with neighboring intelligent agents, and determines whether the communication triggering condition is met, specifically including the following steps: The multiple agents compare the current semantic state with the semantic state of the previous time step and calculate the amount of semantic change. Multiple agents compare their current semantic state with their neighbors and calculate the semantic difference between themselves and their neighboring agents. Compare the semantic change or semantic difference with the corresponding preset threshold; When the amount of semantic change or the degree of semantic difference exceeds the corresponding preset threshold, it is determined that the communication triggering condition is met. When both the semantic change and semantic difference do not exceed the corresponding preset thresholds, it is determined that the communication triggering conditions are not met.

[0009] As a further limitation of the technical solution of this embodiment of the invention, the formula for calculating the semantic change is: ; in, Represents time step Time The semantic changes corresponding to each agent; The formula for calculating the semantic difference is: ; in, It can be a distance function, a difference function, or a combination function. Representative and time step Time Each agent corresponds to a neighboring agent. Represents time step Time The semantic difference between an agent and its corresponding neighboring agents; The expression for the communication triggering condition is: or ; in, and These are preset thresholds corresponding to semantic change and semantic difference.

[0010] As a further limitation of the technical solution of this embodiment of the invention, the communication message focuses on describing the key changes in the semantic state, rather than the complete semantic state, in order to reduce communication redundancy.

[0011] As a further limitation of the technical solution of this embodiment of the invention, the expression of the communication message is: ; in, For time step Time Communication messages corresponding to each intelligent agent This represents a communication content generation map used to extract information that contributes to co-evolution from semantic state changes.

[0012] As a further limitation of the technical solution of this embodiment of the invention, the expression for the collaborative evolution update of semantic state is: ; in, The semantic state is updated after co-evolution. This represents the amount of local semantic updates caused by changes in the agent's own perception or the task itself. This represents the amount of collaborative evolutionary updates generated by the interaction of multiple intelligent agents. , These are time-related weighting parameters; The collaborative evolution update amount is constructed based on the semantic state of the neighbors, and its expression is: ; in, Represents intelligent agents With intelligent agents The collaborative weights between them It represents the set of communicating neighbors.

[0013] As a further limitation of the technical solution of this embodiment of the invention, the formula for calculating the degree of semantic consistency is as follows: ; in, Represents time step Time-based intelligent agents The degree of overall semantic consistency with neighboring groups. This represents the semantic similarity function.

[0014] A communication system based on multi-agent semantic cooperative evolution for executing any of the communication methods based on multi-agent semantic cooperative evolution described above, the system comprising a semantic state construction module, a semantic change detection module, a semantic cooperative evolution module, a communication policy generation module, a communication execution module, and a feedback and update module, wherein: The semantic state construction module is used to build and maintain the semantic state of the agent; The semantic change detection module is used to calculate the amount of semantic change and the semantic difference between the agent and its neighboring agents, and to determine whether the communication triggering conditions are met. The semantic co-evolution module is used to perform co-evolution updates of semantic states according to preset semantic co-evolution rules. The communication strategy generation module is used to dynamically generate communication strategies. The communication execution module is used to generate communication messages according to the communication strategy and send them to one or more intelligent agents, while receiving communication messages from other intelligent agents. The feedback and update module is used to obtain the communication results and feed them back to the semantic state construction module and the communication strategy generation module.

[0015] Compared with the prior art, the beneficial effects of the present invention are: (1) This invention introduces semantic state modeling and co-evolution mechanism into a multi-agent system to characterize the semantic interaction relationship between agents, so that semantic information can continuously evolve at the group level with the cooperation process; at the same time, by combining communication strategy with semantic co-evolution state, adaptive adjustment of communication parameters is achieved, thereby reducing communication redundancy, improving semantic consistency, and enhancing the cooperation efficiency and operational stability of the multi-agent system in complex dynamic environments. (2) The semantic states of multiple agents in this invention are no longer isolated and static expressions, but form a dynamic structure that is continuously adjusted and mutually influenced during the interaction process, so that the semantic understanding at the group level can be gradually aligned with the collaboration process, providing a stable basis for the subsequent generation of communication strategies based on semantic states. (3) The multiple intelligent agents of the present invention can realize the continuous collaborative evolution of semantic state in dynamic environment and collaborative task, so that the communication behavior can be adaptively matched with the semantic consistency requirements, and the communication overhead can be effectively reduced while ensuring the collaborative effect. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention.

[0017] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0018] Figure 2 An application architecture diagram of the system provided in an embodiment of the present invention is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] It is understandable that existing technologies for communication between multiple agents have several shortcomings, which limit their application effect and scalability in multi-agent systems. These shortcomings are mainly reflected in the following aspects: (1) Existing multi-agent communication technologies generally regard communication objects as independent and discrete information units. The communication process mainly revolves around the integrity and efficiency of a single information transmission, and lacks systematic modeling of long-term semantic relationships between agents. In the process of multi-agent collaboration, the information content that each agent is concerned with is often highly related and changes continuously with the progress of the task. However, existing technologies are unable to characterize such semantic relationships across agents and time scales, resulting in a large amount of repeated transmission and semantic redundancy in the communication process, which reduces the overall communication efficiency; (2) Existing semantic communication methods mostly focus on semantic extraction and reconstruction under a single communication subject or fixed communication relationship, and usually assume that the semantic representation model remains relatively stable in the communication process. In multi-agent systems, the environmental state, perception ability and task role of different agents are different, and semantic understanding has obvious individual differences. Existing technologies lack effective mechanisms to dynamically coordinate semantic differences between different agents, which can easily lead to inconsistent semantic understanding, thereby affecting the accuracy of collaborative decision-making and system stability; (3) Although some existing technologies have introduced learning or adaptive mechanisms to optimize communication parameters, their optimization targets are mostly focused on low-level indicators such as communication frequency, bandwidth usage, or transmission delay, and they fail to deeply couple communication strategies with semantic state changes between multiple agents. Such schemes usually separate the adjustment of communication strategies from the evolution of semantic content, making it difficult for the communication process to be adjusted in conjunction with the collaborative relationship and semantic consistency requirements of multiple agents, thus limiting the adaptability of the communication system to complex collaborative tasks; (4) In multi-agent collaborative tasks, the interaction relationship and collaboration mode between agents are usually dynamic and uncertain, and semantic information at the group level shows the characteristics of gradual evolution over time. However, the semantic update methods in existing technologies mostly rely on local observation or static rules, lacking a unified description and control mechanism for the collaborative evolution process of semantics in a multi-agent group, making it difficult to guarantee the consistency and stability of semantic evolution direction, which may lead to a decrease in collaborative efficiency or fluctuations in system performance.

[0021] To address the aforementioned issues, this invention initializes semantic states for multiple agents during system startup or task initiation. At each time step, each agent locally updates its semantic state based on its real-time perception results, internal state changes, and task progress, resulting in an updated semantic state. Each agent performs stage-wise and neighbor-wise comparisons of its semantic state, calculates the amount of semantic change and the semantic difference with neighboring agents, and determines whether communication triggering conditions are met. When communication triggering conditions are met, the corresponding agent generates a communication message based on the amount of semantic change and sends the message to one or more neighboring agents. Upon receiving the communication message, each agent combines it with its own semantic state and updates its semantic state according to preset semantic co-evolution rules. Based on the updated semantic state, the semantic consistency among the agents is evaluated, and the communication triggering conditions, communication frequency, and communication content selection strategy for subsequent time steps are dynamically adjusted. It can characterize the semantic interaction relationship between agents through semantic state modeling and co-evolution mechanism, so that semantic information can continuously evolve at the group level with the collaboration process. By combining communication strategy with semantic co-evolution state, it can achieve adaptive adjustment of communication parameters, thereby reducing communication redundancy, improving semantic consistency, and enhancing the collaboration efficiency and operational stability of multiple agents in complex dynamic environments.

[0022] Figure 1 A flowchart of the method provided by an embodiment of the present invention is shown.

[0023] Specifically, a communication method based on multi-agent semantic cooperative evolution includes the following steps: Step S101: During system startup or task start phase, initialize semantic states for multiple agents.

[0024] In this embodiment of the invention, during the system startup or task start phase, when initializing semantic states for multiple agents, the multiple agents construct corresponding semantic states based on corresponding initial perceptual information, task objectives, or prior knowledge. These semantic states are represented as being located in the semantic space. The vector in the vector is expressed as: ; in, Semantic representation dimension Indicates by A vector space consisting of n real numbers Represents a time step. For time step Time The semantic state corresponding to each agent; Furthermore, based on the semantic states corresponding to multiple agents, a group semantic state set is introduced to support semantic interaction among multiple agents, expressed as: ; in, Indicates shared ownership An intelligent agent.

[0025] Understandably, to support the collaborative evolution and adaptive communication of semantic information among multiple agents, a unified model is used to model the semantic information of interactions between multiple agents. The concept of semantic state is introduced to characterize the semantic understanding of task-related information by an agent at a specific moment. In this embodiment of the invention, it consists of N agents, denoted as... At time step t, each agent Based on its perceived information, internal state, and historical interaction results, a corresponding semantic state is constructed. Semantic state is not raw data or bit-level information, but a high-level abstract representation of task-related information, used to reflect the agent's current semantic cognition.

[0026] It is understandable that semantic state can comprehensively reflect information such as environmental state, task progress, collaborative relationships, and uncertainties.

[0027] In step S102, at each time step, the multiple agents update their semantic state locally based on their real-time perception results, internal state changes, and task progress, thus obtaining the updated semantic state.

[0028] In this embodiment of the invention, at each time step, each agent updates its current semantic state locally based on its real-time perception results, internal state changes, and task progress, and obtains the updated semantic state, which reflects the agent's latest semantic understanding of the external environment and task state.

[0029] In step S103, all the agents perform phase comparison and neighbor comparison on the semantic state, calculate the semantic change amount and the semantic difference degree between the agents and the neighbor agents, and determine whether the communication triggering condition is met.

[0030] In this embodiment of the invention, multiple agents compare their current semantic state with the semantic state of the previous time step, calculate the semantic change and the semantic difference with neighboring agents, and compare the semantic change or semantic difference with a corresponding preset threshold to determine whether the communication triggering condition is met. Specifically, when the semantic change or semantic difference exceeds the corresponding preset threshold, the communication triggering condition is determined to be met; when neither the semantic change nor the semantic difference exceeds the corresponding preset threshold, the communication triggering condition is determined not to be met. The formula for calculating the semantic change is as follows: ; in, Represents time step Time The semantic change quantity corresponding to each intelligent agent is used to characterize the evolution trend of the agent's semantic cognition over time, and can serve as an important part of the communication content, thereby reducing redundant information transmission. The formula for calculating semantic dissimilarity is: ; in, It can be a distance function, a difference function, or a combination function. Representative and time step Time Each agent corresponds to a neighboring agent. Represents time step Time The semantic difference between an agent and its corresponding neighboring agents; The expression for the communication trigger condition is: or ; in, and These are preset thresholds corresponding to semantic change and semantic difference.

[0031] Understandably, by judging the communication triggering conditions, communication is only triggered when there is a significant semantic disagreement or a significant change in the semantic state, thereby avoiding unnecessary frequent communication.

[0032] Understandably, by constructing a unified semantic state representation framework, semantic information can be modeled in a stateful and structured form, laying the foundation for subsequent semantic co-evolution mechanisms and the design of semantic state-based communication strategies.

[0033] In step S104, when the communication triggering condition is met, the corresponding agent generates a communication message based on the semantic change and sends the communication message to one or more neighboring agents.

[0034] In this embodiment of the invention, when the communication triggering condition is met, the agent generates a communication message based on the semantic change and sends the communication message to one or more neighboring agents. The communication message focuses on describing the key changes in the semantic state, rather than the complete semantic state, to reduce communication redundancy. The expression of the communication message is: ; in, For time step Time Communication messages corresponding to each intelligent agent This represents a communication content generation map used to extract information that contributes to co-evolution from semantic state changes.

[0035] Understandably, communication messages can contain only semantic dimensions that change significantly or semantic components that are of high importance to the collaborative task, thereby further compressing the communication load and making the communication content more focused on "key changes in semantic evolution" rather than the repeated transmission of redundant information.

[0036] In step S105, after receiving the communication message, the multiple agents combine the communication message with their own semantic state and perform a collaborative evolution update of the semantic state according to the preset semantic collaborative evolution rules.

[0037] In this embodiment of the invention, after receiving a communication message from a neighboring intelligent agent, each agent combines the communication message with its own semantic state and updates the current semantic state collaboratively according to a preset semantic collaborative evolution rule. This allows the semantic state to evolve towards group consistency while maintaining individual characteristics. Specifically, the expression for the collaborative evolution update of the semantic state is as follows: ; in, The semantic state is updated after co-evolution. This represents the amount of local semantic updates caused by changes in the agent's own perception or the task itself. This represents the amount of collaborative evolutionary updates generated by the interaction of multiple intelligent agents. , These are time-related weighting parameters; The co-evolution update quantity is constructed based on the semantic state of neighbors, and the expression is: ; in, Represents intelligent agents With intelligent agents The collaborative weights between them It represents the set of communication neighbors, which is determined by communication topology, task relationships, or distance constraints.

[0038] It is understandable that the collaborative weight can be dynamically adjusted based on factors such as semantic similarity, historical collaboration effect, or communication reliability. To avoid oscillations or divergence in the semantic state during the evolution process, evolutionary consistency constraints can be further introduced to limit the semantic update amplitude. When the semantic difference between agents is lower than a preset threshold, the collaborative evolution weight is reduced; when the semantic difference exceeds the threshold, the collaborative constraint strength is increased, thereby achieving smooth evolution of the semantic state.

[0039] Understandably, allowing semantic co-evolution mechanisms to execute via event triggering or periodic triggering, triggering one or more co-evolution updates when changes in task stage, significant environmental changes, or decreased semantic consistency are detected, can enhance the adaptability of multi-agents in dynamic environments. Furthermore, based on the construction of multi-agent semantic states and the semantic co-evolution mechanism, a communication strategy generation method based on semantic co-evolution states is proposed. This method dynamically matches communication behavior with the degree of semantic consistency and evolution states among multi-agents, thereby reducing communication overhead while ensuring collaborative effectiveness. Among multiple agents, not all agents need to exchange an equal amount of information at all times. When semantic understanding among agents is highly consistent, frequent communication introduces redundancy; conversely, when semantic differences are significant or collaborative states change, insufficient communication may lead to collaboration failure. Therefore, by directly linking communication decisions to semantic co-evolution states, communication becomes an adaptive process driven by semantic needs.

[0040] Understandably, based on the construction of multi-agent semantic state representation, in order to solve the problem of the accumulation of semantic understanding differences among multi-agents over time and the difficulty in maintaining collaborative consistency, a multi-agent semantic co-evolution mechanism is used to guide the orderly and stable dynamic evolution of semantic states at the group level. Due to differences in perception perspective, environmental state, and task role, the semantic states of each agent often differ. If there is a lack of an effective coordination mechanism, these differences may continue to amplify during the interaction process, thereby affecting the collaborative effect. Therefore, through co-evolution constraints, the semantic states of each agent can gradually evolve towards group consistency while maintaining individual characteristics.

[0041] Step S106: Based on the updated semantic state, evaluate the degree of semantic consistency among the corresponding multiple agents, and dynamically adjust the communication triggering conditions, communication frequency, and communication content selection strategy for subsequent time steps.

[0042] In this embodiment of the invention, the updated semantic state is used to evaluate the degree of semantic consistency among the current multiple agents, and accordingly dynamically adjusts the communication triggering conditions, communication frequency, and communication content selection strategy for subsequent time steps, thereby forming a closed-loop process of "semantic evolution - communication - feedback". This enables multiple agents to achieve continuous collaborative evolution of semantic states in dynamic environments and collaborative tasks, allowing communication behavior to adaptively match semantic consistency requirements, effectively reducing communication overhead while ensuring collaborative effectiveness. Specifically, the formula for calculating the degree of semantic consistency is as follows: ; in, Represents time step Time-based intelligent agents The degree of overall semantic consistency with neighboring groups. This represents the semantic similarity function.

[0043] Furthermore, Figure 2 An application architecture diagram of the system provided in an embodiment of the present invention is shown.

[0044] In another preferred embodiment of the present invention, a communication system based on multi-agent semantic cooperative evolution for executing any of the above-described communication methods based on multi-agent semantic cooperative evolution includes: Semantic state construction module 101 is used to construct and maintain the semantic state of the agent.

[0045] In this embodiment of the invention, the semantic state construction module 101 constructs and maintains the semantic state of the agent based on the agent's perception information, internal state and task-related information. By mapping the original data into a structured semantic state representation, it provides basic input for subsequent semantic co-evolution and communication decision-making.

[0046] The semantic change detection module 102 is used to calculate the amount of semantic change and the degree of semantic difference with neighboring intelligent agents, and to determine whether the communication triggering conditions are met.

[0047] In this embodiment of the invention, the semantic change detection module 102 compares the current semantic state with the historical semantic state, calculates the amount of semantic change, and determines whether the semantic change meets the communication triggering condition, so as to avoid redundant communication when the semantic change is small or the semantic consistency is high.

[0048] The semantic co-evolution module 103 is used to perform co-evolution updates of semantic states according to preset semantic co-evolution rules.

[0049] In this embodiment of the invention, after receiving semantic change information from other agents, the semantic co-evolution module 103 updates the local semantic state according to the preset co-evolution rules, so that the semantic state gradually aligns in the multi-agent group and maintains the stability and consistency of the evolution process.

[0050] The communication strategy generation module 104 is used to dynamically generate communication strategies.

[0051] In this embodiment of the invention, the communication strategy generation module 104 dynamically generates a communication strategy based on the current semantic state and semantic co-evolution state. The communication strategy includes, but is not limited to, parameters such as communication triggering conditions, communication frequency, communication object selection, and communication content scale, so that the communication behavior matches the semantic consistency requirements.

[0052] The communication execution module 105 is used to generate communication messages according to the communication strategy and send them to one or more intelligent agents, while receiving communication messages from other intelligent agents.

[0053] In this embodiment of the invention, the communication execution module 105 generates a communication message according to the communication strategy output by the communication strategy generation module and sends it to one or more target intelligent agents through the communication interface. At the same time, it receives communication messages from other intelligent agents. The communication messages preferentially contain semantic state change related information to reduce the communication load.

[0054] The feedback and update module 106 is used to obtain the communication results and feed them back to the semantic state construction module and the communication strategy generation module.

[0055] In this embodiment of the invention, the feedback and update module 106 feeds back the communication results to the semantic state construction module and the communication strategy generation module, which can continuously adjust subsequent communication behaviors based on the latest semantic evolution state, thereby forming a closed-loop linkage between semantic evolution and communication strategy.

[0056] It is understandable that through the collaborative work of multiple modules, the semantic state can be continuously evolved and aligned in a dynamic environment, enabling communication behavior to be adaptively adjusted according to semantic consistency and collaboration requirements, thereby effectively reducing communication resource consumption while ensuring the collaborative effect of multi-agents.

[0057] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0058] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0059] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0060] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0061] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A communication method based on multi-agent semantic cooperative evolution, characterized in that, The method specifically includes the following steps: During system startup or task initiation, semantic states are initialized for multiple agents; At each time step, the multiple agents update their semantic state locally based on their real-time perception results, internal state changes, and task progress, thus obtaining the updated semantic state. Each of the aforementioned agents performs a phase comparison and a neighbor comparison of the semantic state, calculates the amount of semantic change and the degree of semantic difference with neighboring agents, and determines whether the communication triggering condition is met. When the communication triggering condition is met, the corresponding agent generates a communication message based on the semantic change and sends the communication message to one or more neighboring agents; After receiving a communication message, multiple intelligent agents combine the communication message with their own semantic state and update the semantic state in a collaborative manner according to a preset semantic collaborative evolution rule. Based on the updated semantic state, the semantic consistency among the corresponding multiple agents is evaluated, and the communication triggering conditions, communication frequency, and communication content selection strategy for subsequent time steps are dynamically adjusted.

2. The communication method based on multi-agent semantic cooperative evolution according to claim 1, characterized in that, During the system startup or task start phase, when initializing semantic states for multiple agents, the multiple agents construct corresponding semantic states based on their respective initial perception information, task objectives, or prior knowledge.

3. The communication method based on multi-agent semantic cooperative evolution according to claim 1, characterized in that, Semantic states are represented as being located in the semantic space The vector in the vector is expressed as: ； in, Semantic representation dimension Indicates by A vector space consisting of n real numbers Represents a time step. For time step Time The semantic state corresponding to each agent; Based on the semantic states corresponding to multiple agents, a group semantic state set is introduced to support semantic interaction among multiple agents, expressed as: ； in, Indicates shared ownership An intelligent agent.

4. The communication method based on multi-agent semantic cooperative evolution according to claim 1, characterized in that, Each of the multiple agents performs phase comparison and neighbor comparison on semantic state, calculates semantic change and semantic difference with neighboring agents, and determines whether the communication triggering condition is met. Specifically, this includes the following steps: The multiple agents compare the current semantic state with the semantic state of the previous time step and calculate the amount of semantic change. Multiple agents compare their current semantic state with their neighbors and calculate the semantic difference between themselves and their neighboring agents. Compare the semantic change or semantic difference with the corresponding preset threshold; When the amount of semantic change or the degree of semantic difference exceeds the corresponding preset threshold, it is determined that the communication triggering condition is met. When both the semantic change and semantic difference do not exceed the corresponding preset thresholds, it is determined that the communication triggering conditions are not met.

5. The communication method based on multi-agent semantic cooperative evolution according to claim 3, characterized in that, The formula for calculating the semantic change is: ； in, Represents time step Time The semantic changes corresponding to each agent; The formula for calculating the semantic difference is: ； in, It can be a distance function, a difference function, or a combination function. Representative and time step Time Each agent corresponds to a neighboring agent. Represents time step Time The semantic difference between an agent and its corresponding neighboring agents; The expression for the communication triggering condition is: or ; in, and These are preset thresholds corresponding to semantic change and semantic difference.

6. The communication method based on multi-agent semantic cooperative evolution according to claim 1, characterized in that, The communication messages focus on describing key changes in the semantic state, rather than the complete semantic state, in order to reduce communication redundancy.

7. The communication method based on multi-agent semantic cooperative evolution according to claim 5, characterized in that, The expression for the communication message is: ； in, For time step Time Communication messages corresponding to each intelligent agent This represents a communication content generation map used to extract information that contributes to co-evolution from semantic state changes.

8. The communication method based on multi-agent semantic cooperative evolution according to claim 5, characterized in that, The expression for the collaborative evolution update of semantic state is: ； in, The semantic state is updated after co-evolution. This represents the amount of local semantic updates caused by changes in the agent's own perception or the task itself. This represents the amount of collaborative evolutionary updates generated by the interaction of multiple intelligent agents. , These are time-related weighting parameters; The collaborative evolution update amount is constructed based on the semantic state of the neighbors, and its expression is: ； in, Represents intelligent agents With intelligent agents The collaborative weights between them It represents the set of communicating neighbors.

9. The communication method based on multi-agent semantic cooperative evolution according to claim 8, characterized in that, The formula for calculating the degree of semantic consistency is as follows: ； in, Represents time step Time-based intelligent agents The degree of overall semantic consistency with neighboring groups. This represents the semantic similarity function.

10. A communication system based on multi-agent semantic cooperative evolution for executing the communication method based on multi-agent semantic cooperative evolution as described in any one of claims 1-9, characterized in that, The system includes a semantic state construction module, a semantic change detection module, a semantic co-evolution module, a communication strategy generation module, a communication execution module, and a feedback and update module, wherein: The semantic state construction module is used to build and maintain the semantic state of the agent; The semantic change detection module is used to calculate the amount of semantic change and the semantic difference between the agent and its neighboring agents, and to determine whether the communication triggering conditions are met. The semantic co-evolution module is used to perform co-evolution updates of semantic states according to preset semantic co-evolution rules. The communication strategy generation module is used to dynamically generate communication strategies. The communication execution module is used to generate communication messages according to the communication strategy and send them to one or more intelligent agents, while receiving communication messages from other intelligent agents. The feedback and update module is used to obtain the communication results and feed them back to the semantic state construction module and the communication strategy generation module.

Images