A multi-agent environment operation system based on adaptive social structure

Abstract

The present invention provides a multi-agent environment operation system based on an adaptive social structure, including: respectively obtaining the intelligent attributes corresponding to each operating agent to construct the interaction methods corresponding to different operating agents, building a social simulation environment model according to the social structure, respectively analyzing the interaction resources corresponding to each interaction method in the social simulation environment model, analyzing the operation value characteristics of each interaction resource to different operating agents in the social simulation environment model, synthesizing the operation value characteristics in the social simulation environment model according to the interaction method, obtaining the operation agent interaction results and displaying them, and promoting continuous compound innovation of artificial intelligence systems by simulating multi-level interactions in human society and culture.

Description

Technical Field

[0001] The present invention relates to the field of virtual environment operation technology, and in particular to a multi-agent environment operation system based on an adaptive social structure. Background Art

[0002] With the development of artificial intelligence and machine learning, multi-agent systems have been widely used in many fields, including robotics, games, traffic management, and intelligent manufacturing. Some multiplayer games can be regarded as multi-agent environments, but they often do not support or are not suitable for existing researchers to develop algorithms for controlling agents. Traditional research on artificial intelligence and complex system modeling mainly focuses on the learning and optimization of a single agent, but this approach faces limitations when dealing with complex, dynamic, and uncertain environments. In nature, intelligent behavior is often achieved through the collective behavior and collaboration of multiple interacting agents. In particular, human intelligence as a collective has the ability to continuously innovate through social interaction and learning. Therefore, studying how artificial intelligence can imitate this characteristic of human intelligence through social learning and cultural evolution has become an important research topic.

[0003] Therefore, the present invention provides a multi-agent environment operation system based on an adaptive social structure. Summary of the invention

[0004] The present invention provides a multi-agent environment operation system based on an adaptive social structure, which integrates the social structure with the existing environment, can simulate the expanding physical resource space and social structure in a virtual environment, and can promote continuous compound innovation of artificial intelligence systems by simulating multi-level interactions in human society and culture.

[0005] The present invention provides a multi-agent environment operation system based on an adaptive social structure, comprising:

[0006] An intelligent analysis module is used to obtain the intelligent attributes corresponding to each running intelligent agent and construct the interaction modes corresponding to different running intelligent agents;

[0007] A structure building module, used to build a social simulation environment model according to the social structure, and analyze the interactive resources corresponding to each of the interactive modes in the social simulation environment model;

[0008] A resource pricing module, used for analyzing the operation value characteristics of each interactive resource to different operation intelligent entities in the social simulation environment model;

[0009] The resource synthesis module is used to synthesize the operation value characteristics in the social simulation environment model according to the interaction mode, obtain the operation agent interaction results and display them.

[0010] In one practicable manner,

[0011] The intelligent analysis module comprises:

[0012] A pre-connection unit, used to connect each of the running agents included in the multi-agent environment to a specified algorithm through an API;

[0013] An attribute analysis unit, used to analyze the state information corresponding to each of the running intelligent agents at the current moment by using the specified algorithm, and determine the intelligent attribute corresponding to each of the running intelligent agents;

[0014] An operation analysis unit, used to determine the intelligent decision-making mode, communication exchange mode and competition and cooperation mode corresponding to the operation intelligent agent according to the intelligent attribute, and to construct an internal operation mode set for each of the operation intelligent agents;

[0015] An interaction construction unit is used to determine the reward sharing relationship characteristics and observation sharing characteristics of each operating agent in the multi-agent environment based on the internal operating mode set, and to construct an interaction mode between different operating agents.

[0016] In one practicable manner,

[0017] Also includes:

[0018] The activity visualization unit is used to obtain the environment state (s, t) corresponding to the running agent at time t, and determine the activity information corresponding to the running agent in the multi-agent environment based on the environment state (s, t) and display it.

[0019] In one practicable manner,

[0020] The attribute analysis unit comprises:

[0021] A synchronous detection subunit, used to respectively obtain the activity information corresponding to each of the running agents, and determine the agent observation information and agent reward information corresponding to the running agent at time t according to the activity information;

[0022] An algorithm conversion subunit, used to determine, according to the specified algorithm, the expression of the running agent observation information corresponding to each running agent is (o_{t, i}), the expression of the corresponding running agent reward information is (r_{t, i}), and the agent action (a_{t, i}) corresponding to each running agent is calculated using the specified algorithm;

[0023] The attribute confirmation subunit is used to obtain the agent actions (a_{t, i}) corresponding to each of the running agents at different times, construct the activity actions corresponding to each of the running agents, generate the state information corresponding to the running agents at the current time, and determine the intelligent attributes corresponding to each of the running agents.

[0024] In one practicable manner,

[0025] The structure building module comprises:

[0026] A structure building unit, used to obtain a social structure set in advance, build a social simulation environment model in combination with a model building technology, and input the operating agent into the social simulation environment model for operation;

[0027] An interaction analysis unit, used for obtaining the operation performance of each interaction mode in the social simulation environment model;

[0028] The deep interaction unit is used to set a corresponding resource library for each of the running intelligent agents, and determine the corresponding interactive resources when each of the running intelligent agents performs different interactive modes according to the corresponding running performance.

[0029] In one practicable manner,

[0030] The resource pricing module includes:

[0031] A pricing trigger unit, used to obtain resource activity characteristics corresponding to each of the interaction modes in the social simulation environment model, and set resource trigger conditions for the corresponding operating agent based on each of the interaction modes;

[0032] A trigger supervision unit, configured to determine trigger information of the running agent when the running agent is running in the social simulation environment model;

[0033] The pricing operation unit is used to generate the operation value characteristics of the corresponding interactive resources to the operation intelligent body by using the information amount corresponding to the trigger information.

[0034] In one practicable manner,

[0035] The resource synthesis module includes:

[0036] A first synthesis unit, configured to construct a multi-layer directed graph of the social simulation model environment according to the interaction mode of the running intelligent agent in the social simulation environment model;

[0037] A second synthesis unit is used to determine the cooperation relationship and the subordinate relationship between different operating agents in the social simulation environment model according to the multi-layer directed graph;

[0038] A third synthesis unit, used for determining resource synthesis characteristics and resource extension characteristics when different operating agents interact based on the cooperation relationship and the subordinate relationship;

[0039] The fourth synthesis unit is used to synthesize the corresponding operation value characteristics based on the resource synthesis characteristics and the resource extension characteristics, obtain and display the interaction results of several operation agents in the social simulation environment model.

[0040] In one practicable manner,

[0041] Also includes:

[0042] The high-level intelligent entities to which each of the operating intelligent entities is subordinate are determined based on the subordinate relationships, and the same-frequency intelligent entities that each of the operating intelligent entities cooperates with are determined based on the cooperative relationships, and the social structure of the operating intelligent entities is established and displayed.

[0043] The achievable beneficial effects of the above technical solution are: in order to integrate the activities of intelligent agents into society, social analysis is conducted on the operating intelligent agents in the multi-agent environment, and the interactive resources of the operating intelligent agents during different interactions are determined by constructing a social simulation model, and then the operating value characteristics of each interactive resource to the operating intelligent agent and the results of the operating intelligent agent's interaction are further determined, thereby demonstrating the working conditions of the operating intelligent agent in society, achieving the purpose of synchronization with society, and realizing the leap of imitating human intelligence.

[0044] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0045] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0047] Figure 1 A schematic diagram of the composition of a multi-agent environment operation system based on an adaptive social structure in an embodiment of the present invention;

[0048] Figure 2 Schematic diagram of the composition of an intelligent analysis module of a multi-agent environment operation system based on an adaptive social structure in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0050] Example 1

[0051] This embodiment provides a multi-agent environment operation system based on an adaptive social structure, such as Figure 1 As shown, including:

[0052] An intelligent analysis module is used to obtain the intelligent attributes corresponding to each running intelligent agent and construct the interaction modes corresponding to different running intelligent agents;

[0053] A structure building module, used to build a social simulation environment model according to the social structure, and analyze the interactive resources corresponding to each of the interactive modes in the social simulation environment model;

[0054] A resource pricing module, used for analyzing the operation value characteristics of each interactive resource to different operation intelligent entities in the social simulation environment model;

[0055] The resource synthesis module is used to synthesize the operation value characteristics in the social simulation environment model according to the interaction mode, obtain the operation agent interaction results and display them.

[0056] In this case, an agent is an entity that can perceive the environment and take actions based on the perception results. It can be a software program, a robot, or other form of autonomous system.

[0057] In this example, the interaction mode refers to the mode in which two or more running agents perform interactive activities;

[0058] In this instance, a multi-agent environment refers to a scenario where multiple agents interact, collaborate, or compete in the same environment;

[0059] In this instance, social structure refers to the structure of relations in actual society;

[0060] In this instance, interactive resources represent resources that change when interactive work is performed;

[0061] In this example, resources represent the materials consumed by the agent when performing activities;

[0062] In this instance, the operational value feature represents the use value of the interactive resource to the operational agent;

[0063] In this example, the running agent interaction result refers to the result produced after the running agent interacts.

[0064] The working principle and beneficial effects of the above technical solution: In order to integrate the activities of intelligent agents into society, social analysis is conducted on the operating intelligent agents in the multi-agent environment, and the interactive resources of the operating intelligent agents during different interactions are determined by constructing a social simulation model. Then, the operating value characteristics of each interactive resource to the operating intelligent agent and the results of the operating intelligent agent interaction are further determined, thereby demonstrating the working conditions of the operating intelligent agent in society, achieving the purpose of synchronization with society, and realizing the leap of imitating human intelligence.

[0065] Example 2

[0066] Based on Example 1, the multi-agent environment operation system based on adaptive social structure is as follows: Figure 2 As shown, the intelligent analysis module includes:

[0067] A pre-connection unit, used to connect each of the running agents included in the multi-agent environment to a specified algorithm through an API;

[0068] An attribute analysis unit, used to analyze the state information corresponding to each of the running intelligent agents at the current moment by using the specified algorithm, and determine the intelligent attribute corresponding to each of the running intelligent agents;

[0069] An operation analysis unit, used to determine the intelligent decision-making mode, communication exchange mode and competition and cooperation mode corresponding to the operation intelligent agent according to the intelligent attribute, and to construct an internal operation mode set for each of the operation intelligent agents;

[0070] An interaction construction unit is used to determine the reward sharing relationship characteristics and observation sharing characteristics of each operating agent in the multi-agent environment based on the internal operating mode set, and to construct an interaction mode between different operating agents.

[0071] In this instance, API means programming software interface;

[0072] In this example, the specified algorithm represents the algorithm used to construct the virtual environment;

[0073] In this instance, intelligent attributes represent the attributes displayed by the running agent when performing intelligent activities;

[0074] In this example, intelligent decision-making means the way in which the operating agent makes actions and choices through perceptual information;

[0075] In this example, the communication exchange mode refers to the way in which the running agents communicate;

[0076] In this example, the competition-cooperation mode refers to the mode in which two or more operating agents perform competitive work and cooperative work;

[0077] In this example, the reward sharing relationship feature indicates the presence or absence of reward sharing activities;

[0078] In this instance, the observation sharing feature indicates the presence or absence of observation sharing activities;

[0079] In this example, the internal operation mode includes the intelligent decision-making mode, communication exchange mode, and competition and cooperation mode of the operating agent.

[0080] The working principle and beneficial effects of the above technical solution: In order to explore the interaction mode between different running intelligent agents, an algorithm is used to determine the intelligent attributes of the running intelligent agents in the multi-agent environment, and then a set of internal operating modes that can only be filled in by the running intelligent agents is constructed, and the interaction mode of the running intelligent agents in the multi-agent environment is further determined. The interaction mode between different running intelligent agents is constructed, and in-depth analysis is achieved to determine the interaction mode of each running intelligent agent.

[0081] Example 3

[0082] On the basis of Example 2, the multi-agent environment operation system based on the adaptive social structure further includes:

[0083] The activity visualization unit is used to obtain the environment state (s, t) corresponding to the running agent at time t, and determine the activity information corresponding to the running agent in the multi-agent environment based on the environment state (s, t) and display it.

[0084] In this example, s represents the environment information of the running agent.

[0085] The working principle and beneficial effects of the above technical solution are as follows: all information in a unique state environment is determined.

[0086] Example 4

[0087] On the basis of Embodiment 2, in the multi-agent environment operation system based on adaptive social structure, the attribute analysis unit comprises:

[0088] A synchronous detection subunit, used to respectively obtain the activity information corresponding to each of the running agents, and determine the agent observation information and agent reward information corresponding to the running agent at time t according to the activity information;

[0089] An algorithm conversion subunit, used to determine, according to the specified algorithm, the expression of the running agent observation information corresponding to each running agent is (o_{t, i}), the expression of the corresponding running agent reward information is (r_{t, i}), and the agent action (a_{t, i}) corresponding to each running agent is calculated using the specified algorithm;

[0090] The attribute confirmation subunit is used to obtain the agent actions (a_{t, i}) corresponding to each of the running agents at different times, construct the activity actions corresponding to each of the running agents, generate the state information corresponding to the running agents at the current time, and determine the intelligent attributes corresponding to each of the running agents.

[0091] In this example, the running agent observation information represents the environmental information obtained by the agent through sensors or other means when the agent is performing tasks or making decisions, and the running agent reward information represents the feedback obtained by the agent during the interaction with the environment;

[0092] In this instance, the agent action represents the specific behavior or operation selected by the agent based on its decision-making process.

[0093] The working principle and beneficial effects of the above technical solution are as follows: by analyzing the activity information of the running intelligent agent, the agent observation information and the agent reward information of the running intelligent agent at a moment are determined, and then the specified algorithm is used to calculate the agent action of the running intelligent agent, and the state information of the running intelligent agent at different moments is determined by combining the actions, thereby constructing the intelligent attributes of the running intelligent agent. In this way, the activities of the running intelligent agent can be continuously supervised and analyzed, and the intelligent attributes of the intelligent agent can be determined, which lays the foundation for subsequent activity work.

[0094] Example 5

[0095] On the basis of Example 1, the multi-agent environment operation system based on the adaptive social structure, the structure building module includes:

[0096] A structure building unit, used to obtain a social structure set in advance, build a social simulation environment model in combination with a model building technology, and input the operating agent into the social simulation environment model for operation;

[0097] An interaction analysis unit, used for obtaining the operation performance of each interaction mode in the social simulation environment model;

[0098] The deep interaction unit is used to set a corresponding resource library for each of the running intelligent agents, and determine the corresponding interactive resources when each of the running intelligent agents performs different interactive modes according to the corresponding running performance.

[0099] In this example, each running agent corresponds to a resource library, and resource capacities of resource libraries of different running agents are different.

[0100] The working principle and beneficial effects of the above technical solution are as follows: the social structure is used to build a social simulation environment model, and then the operating performance of the running intelligent agent is analyzed and a resource library is set up for it. The interactive resources used by the running intelligent agent in different interactions are analyzed, and the interactive process of each interactive mode is determined.

[0101] Example 6

[0102] On the basis of Example 1, the multi-agent environment operation system based on adaptive social structure, the resource pricing module includes:

[0103] A pricing trigger unit, used to obtain resource activity characteristics corresponding to each of the interaction modes in the social simulation environment model, and set resource trigger conditions for the corresponding operating agent based on each of the interaction modes;

[0104] A trigger supervision unit, configured to determine trigger information of the running agent when the running agent is running in the social simulation environment model;

[0105] The pricing operation unit is used to generate the operation value characteristics of the corresponding interactive resources to the operation intelligent body by using the information amount corresponding to the trigger information.

[0106] The working principle and beneficial effects of the above technical solution: Since different resources have different values ​​to different intelligent agents, the operating value of each operating intelligent agent is analyzed according to its interaction mode.

[0107] Example 7

[0108] On the basis of Example 1, the multi-agent environment operation system based on adaptive social structure, the resource synthesis module includes:

[0109] A first synthesis unit, configured to construct a multi-layer directed graph of the social simulation model environment according to the interaction mode of the running intelligent agent in the social simulation environment model;

[0110] A second synthesis unit is used to determine the cooperation relationship and the subordinate relationship between different operating agents in the social simulation environment model according to the multi-layer directed graph;

[0111] A third synthesis unit, used for determining resource synthesis characteristics and resource extension characteristics when different operating agents interact based on the cooperation relationship and the subordinate relationship;

[0112] The fourth synthesis unit is used to synthesize the corresponding operation value characteristics based on the resource synthesis characteristics and the resource extension characteristics, obtain and display the interaction results of several operation agents in the social simulation environment model.

[0113] In this example, a multi-layer directed graph represents a graph structure used to show the relationship between different running agents;

[0114] In this example, the resource synthesis feature represents the process of combining the corresponding interactive resource presentations when the running agent performs resource combination activities;

[0115] In this example, the resource expansion feature indicates the process of generating new resources by expanding the size of the corresponding interactive resources when the running agent performs resource combination activities;

[0116] In this example, there are two types of results of running agent interactions: reward sharing and observation sharing.

[0117] The working principle and beneficial effects of the above technical solution are as follows: by constructing a multi-layer directed graph to analyze the cooperation and subordinate relationships between different operating agents, and then determining the resource synthesis characteristics and resource extension characteristics of the operating agents when interacting, so as to construct and display the corresponding operating agent interaction results, which is convenient for understanding the interactive relationship between different operating agents.

[0118] Example 8

[0119] On the basis of Example 7, the multi-agent environment operation system based on the adaptive social structure further includes:

[0120] The high-level intelligent entities to which each of the operating intelligent entities is subordinate are determined based on the subordinate relationships, and the same-frequency intelligent entities that each of the operating intelligent entities cooperates with are determined based on the cooperative relationships, and the social structure of the operating intelligent entities is established and displayed.

[0121] In this example, a running agent can be subordinate to several different high-level agents at the same time. The high-level agents influence the value functions and strategies of the lower-level agents through strategies such as division of labor.

[0122] The working principle and beneficial effects of the above technical solution are as follows: the in-depth relationship between different operating intelligent agents is determined based on the subordinate and cooperative relationships, thereby establishing a social structure and realizing the emergence and evolution of the social structure. The intelligent agent can establish a higher-level intelligent agent to realize the hierarchical and complex social structure.

[0123] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A multi-agent environment operation system based on an adaptive social structure, characterized in that: include: An intelligent analysis module is used to obtain the intelligent attributes corresponding to each running intelligent agent and construct the interaction modes corresponding to different running intelligent agents; A structure building module, used to build a social simulation environment model according to the social structure, and analyze the interactive resources corresponding to each of the interactive modes in the social simulation environment model; A resource pricing module, used for analyzing the operation value characteristics of each interactive resource to different operation intelligent entities in the social simulation environment model; A resource synthesis module, used to synthesize the operation value characteristics in the social simulation environment model according to the interaction mode, obtain the operation agent interaction results and display them; The intelligent analysis module comprises: A pre-connection unit, used to connect each of the running agents included in the multi-agent environment to a specified algorithm through an API; An attribute analysis unit, used to analyze the state information corresponding to each of the running intelligent agents at the current moment by using the specified algorithm, and determine the intelligent attribute corresponding to each of the running intelligent agents; An operation analysis unit, used to determine the intelligent decision-making mode, communication exchange mode and competition and cooperation mode corresponding to the operation intelligent agent according to the intelligent attribute, and to construct an internal operation mode set for each of the operation intelligent agents; An interaction construction unit, configured to determine the reward sharing relationship characteristics and observation sharing characteristics of each of the operating agents in the multi-agent environment based on the internal operating mode set, and to construct an interaction mode between different operating agents; The structure building module comprises: A structure building unit, used to obtain a social structure set in advance, build a social simulation environment model in combination with a model building technology, and input the operating agent into the social simulation environment model for operation; An interaction analysis unit, used for obtaining the operation performance of each interaction mode in the social simulation environment model; A deep interaction unit, used to respectively set a corresponding resource library for each of the running intelligent agents, and determine the corresponding interactive resources when each of the running intelligent agents performs different interactive modes according to the corresponding running performance; The resource pricing module includes: A pricing trigger unit, used to obtain resource activity characteristics corresponding to each of the interaction modes in the social simulation environment model, and set resource trigger conditions for the corresponding operating agent based on each of the interaction modes; A trigger supervision unit, configured to determine trigger information of the running agent when the running agent is running in the social simulation environment model; A pricing operation unit, configured to generate operation value characteristics of the corresponding interactive resources to the operation intelligent entity by using the information amount corresponding to the trigger information; The resource synthesis module includes: A first synthesis unit, configured to construct a multi-layer directed graph of the social simulation model environment according to the interaction mode of the running intelligent agent in the social simulation environment model; A second synthesis unit is used to determine the cooperation relationship and the subordinate relationship between different operating agents in the social simulation environment model according to the multi-layer directed graph; A third synthesis unit, used for determining resource synthesis characteristics and resource extension characteristics when different operating agents interact based on the cooperation relationship and the subordinate relationship; The fourth synthesis unit is used to synthesize the corresponding operation value characteristics based on the resource synthesis characteristics and the resource extension characteristics, obtain and display the interaction results of several operation agents in the social simulation environment model.

2. A multi-agent environment operation system based on an adaptive social structure as claimed in claim 1, characterized in that: Also includes: The activity visualization unit is used to obtain the environment state (s, t) corresponding to the running agent at time t, and determine the activity information corresponding to the running agent in the multi-agent environment based on the environment state (s, t) and display it.

3. A multi-agent environment operation system based on an adaptive social structure as claimed in claim 1, characterized in that: The attribute analysis unit comprises: A synchronous detection subunit, used to respectively obtain the activity information corresponding to each of the running agents, and determine the agent observation information and agent reward information corresponding to the running agent at time t according to the activity information; An algorithm conversion subunit, used to determine, according to the specified algorithm, the expression of the running agent observation information corresponding to each running agent is (o_{t, i}), the expression of the corresponding running agent reward information is (r_{t, i}), and the agent action (a_{t, i}) corresponding to each running agent is calculated using the specified algorithm; The attribute confirmation subunit is used to obtain the agent actions (a_{t, i}) corresponding to each of the running agents at different times, construct the activity actions corresponding to each of the running agents, generate the state information corresponding to the running agents at the current time, and determine the intelligent attributes corresponding to each of the running agents.

4. A multi-agent environment operation system based on an adaptive social structure as claimed in claim 1, characterized in that: Also includes: The high-level intelligent entities to which each of the operating intelligent entities is subordinate are determined based on the subordinate relationships, and the same-frequency intelligent entities that each of the operating intelligent entities cooperates with are determined based on the cooperative relationships, and the social structure of the operating intelligent entities is established and displayed.

Images