Man-machine-object fusion service network construction method and system based on multi-agent collaboration

Abstract

The invention discloses a man-machine-object fusion service network construction method and system based on multi-agent collaboration. For a man-machine object fusion scene, when the human element is not considered or human modeling is too simplified, more stereoscopic and more dimensional analysis is carried out on the ability of a person, a corresponding relation between the person and the ability is constructed, and a more optimized human modeling method is realized; when a task and a scene which can be processed by a system are solidified and rigidified, self-organized execution of system capability is pulled by defining a resource-capability framework, constructing an environment time sequence data twinborn system, a resource library and a capability library, deploying a plurality of intelligent agents, generating environment events, disassembling the environment events into executable workflows, and constructing a resource-capability framework. And the flexibility and the universality of the man-machine-object service network are improved, so that the system can adapt to different scene requirements. According to the method, human-machine-object heterogeneous resources in the real world can be effectively integrated, and the problems of human modeling simplification and system solidification of the human-machine-object service network are effectively solved.

Description

Technical Field

[0001] The present invention belongs to the field of group intelligence computing technology, and in particular relates to a method and system for constructing a human-machine-object fusion service network based on multi-agent collaboration. Background Art

[0002] The integrated development of the Internet of Things and artificial intelligence has broken the information isolation between the social, cyber and physical spaces. Human groups, information spaces and all physical entities are being extensively and deeply connected to each other, and humans have leapt from the machine-object era into the intelligent interconnected era of the three-way integration of man, machine and object.

[0003] The existing machine-object system architecture mainly focuses on the network level, while ignoring the elements at the social and physical space levels. This makes it less applicable in human-machine-object fusion scenario applications and it is difficult to achieve resource integration across the boundaries of virtual and real space.

[0004] Although a variety of methods for building human-machine-object fusion service networks have been proposed in recent years, they generally have two major defects: (1) The modeling of humans is too simplistic and idealized, losing sight of the multidimensionality of humans and their adaptability to the environment.

[0005] (2) Most of the existing human-machine-object fusion service network processes are based on manual pre-settings and are only applicable to fixed scenarios and characters. They are not suitable for general scenarios and lack flexibility. Summary of the invention

[0006] To solve the above problems, the purpose of the present invention is to provide a method and system for constructing a human-machine-object fusion service network based on multi-agent collaboration, which establishes a resource library and a capability library, deploys multiple agents, obtains events occurring in the environment through heterogeneous resource perception data, decomposes them into executable workflows, drives corresponding capability units to execute functions, realizes self-generation and execution of system workflows, and copes with different scenarios and tasks.

[0007] To achieve the above object, the solution of the present invention is: In a first aspect, the present invention provides a method for constructing a human-machine-object fusion service network based on multi-agent collaboration, the method comprising the following steps: (1) Resource modeling: Register the resources in the environment in the resource library and capability library, abstract the three types of resources: human, machine, and material, and manage them in a unified manner; (2) Crowd-sensing: Resources report data to the system, enabling the environment analyzer to obtain and identify events occurring in the environment and hand these events over to the workflow decomposer; (3) Workflow decomposition: Use intelligent agents and knowledge bases to decompose events into executable workflows; (4) Task execution: The tasks within the workflow are executed by calling the corresponding capability units, thereby advancing the workflow within the task manager.

[0008] In a second aspect, the present invention provides a system for constructing a human-machine-object fusion service network based on multi-agent collaboration, comprising: Resource modeling unit: registers resources in the environment in the resource library and capability library, abstracts and uniformly manages the three types of resources: human, machine, and object; Crowd-sensing unit: Resources report data to the system, allowing the environment analyzer to obtain and identify events that occur in the environment and hand these events over to the workflow decomposer; Workflow decomposition unit: Uses agents and knowledge bases to decompose events into executable workflows; Task execution unit: It executes tasks within the workflow by calling corresponding capability units, thereby advancing the workflow within the task manager.

[0009] In the third aspect, the present invention provides the application of the human-machine-object fusion service network construction system based on multi-agent collaboration in the fire rescue scenario of a smart building.

[0010] Beneficial effects of the invention: The invention defines a resource-capability framework, builds an environmental time series data twin system, a resource library and a capability library, deploys multiple agents, generates environmental events, and disassembles them into executable workflows, driving the self-organizing execution of system capabilities, improving the flexibility and versatility of the human-machine-object service network, and enabling the system to adapt to different scenario requirements. The invention can effectively integrate heterogeneous resources of humans, machines and objects in the real world, and effectively solves the problems of simplified human modeling and system solidification in the human-machine-object service network. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 It is an overview of the scene-driven human-machine-object service network architecture based on multi-agent collaboration; Figure 2 It is a resource library and capability library for fire rescue scenarios in smart buildings. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with the accompanying drawings and implementation examples. It should be understood that the specific implementation methods described herein are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0013] like Figure 1 As shown, the embodiment of the present application provides a method for constructing a human-machine-object fusion service network based on multi-agent collaboration, which involves a resource library, a capability library, an agent, a task manager, and an environment analyzer to work together and respond to events in the environment, which mainly includes: Resource modeling: Register the resources in the environment in the resource library and capability library. After the resources in the scene are registered in the resource library and capability library, these resources will be abstracted and managed in a unified manner.

[0014] Crowd-sensing: Resources within the scene report data to the system, enabling the environment analyzer to obtain events occurring in the environment, which are then handed over to the task decomposer.

[0015] Workflow decomposition: By using intelligent agents and knowledge bases, events within the system are decomposed into executable workflows.

[0016] Task execution: The workflow within the task manager is advanced by calling the execution of the corresponding capability unit.

[0017] Furthermore, each resource corresponds to a specific entity in the scenario, while a function represents a function shared by a set of heterogeneous resources. A resource can be associated with multiple functions, and conversely, a function can also be associated with a set of resources with the same function.

[0018] Furthermore, the system does not focus on specific resource entities, but operates at the granularity of capability units when implementing task status changes. The system only needs to notify the responsible capability unit to perform the corresponding function without specifying specific resource entities.

[0019] Furthermore, the three types of entities in the system, namely people, network services, and physical devices, are all considered resources. Since the functions of network services and physical devices are predetermined, their capabilities can be explicitly listed during resource registration. This includes the specific API interfaces provided, the types of data processed, the required inputs, outputs, and other relevant details. Therefore, a direct relationship can be established between network services, physical devices, and their corresponding capabilities.

[0020] Furthermore, modeling of human resources refers to specifying or simply listing all capabilities of a person during resource registration. This approach is inappropriate. In view of the uniqueness of people, this application proposes a method for dynamically constructing the relationship between individuals and capabilities during workflow execution, as follows: During the execution of the workflow, the corresponding capability unit is called, and the system traverses and evaluates the human resources in the scene, inputting the basic information and other details of the human, such as name, gender, age, occupation, experience, certification, etc., into the deployed resource selection agent. This enables the agent to analyze whether the individual has a specific capability. If so, a relationship is established between the capability and the individual.

[0021] This application example also provides a human-machine-object fusion service network construction system based on multi-agent collaboration, including: Resource modeling unit: registers resources in the environment in the resource library and capability library, abstracts and uniformly manages the three types of resources: human, machine, and object; Crowd-sensing unit: Resources report data to the system, allowing the environment analyzer to obtain and identify events that occur in the environment and hand these events over to the workflow decomposer; Workflow decomposition unit: Uses agents and knowledge bases to decompose events into executable workflows; Task execution unit: It executes tasks within the workflow by calling corresponding capability units, thereby advancing the workflow within the task manager.

[0022] The embodiment of the present application also provides an application of a human-machine-object service fusion network construction system based on multi-agent collaboration in a fire rescue scenario in a smart building.

[0023] like Figure 2 As shown in the figure, the fire rescue scenario of the smart building is one of the classic human-machine-object scenarios, which includes fire extinguishers, humans in the room, smoke alarms, fire analysis services, cameras, etc. For this scenario, this example establishes the corresponding resource library and capability library, deploys intelligent agents, and then realizes the system's arrangement of scenarios and resources.

[0024] The specific process is: Step 1: Modeling the resource capability library (1) For the fire fighting capacity unit, the resource list includes a number of fire extinguishers.

[0025] (2) For the Smoke Alarm Capability Unit, the resource list includes smoke alarms.

[0026] (3) For the Fire Analysis Capability Unit, the resource list includes fire analysis services.

[0027] (4) For the observation capability unit, the resource list includes cameras.

[0028] (5) However, it is not associated with any specific capabilities for specific individuals.

[0029] Step 2: Crowd-sensing (1) The smoke alarm detects smoke and sends a smoke signal, while the camera observes a fire. The coordinates and events are uploaded, the system updates the environment information and hands it over to the environment analyzer, which identifies the event as a fire event.

[0030] Step 3: Workflow decomposition (1) The incident was analyzed and broken down into the following workflow: Find resources with the ability to extinguish the fire -> Find resources with the ability to move and transport items.

[0031] Step 4: Task execution (1) The sub-workflow "Find resources with fire extinguishing capabilities" is executed. By evaluating resource capabilities, the resource list contains fire extinguisher A and fire extinguisher B. Using the scheduling strategy, fire extinguisher A is selected because it is closer to the fire source. The indicator light on fire extinguisher A lights up, and the execution is completed, providing information about fire extinguisher A as the output of the next sub-workflow.

[0032] (2) The sub-workflow "Find resources with the ability to move and carry items" is executed. Based on the resource capability assessment, the resource list includes personnel C, personnel D, and personnel E. All personnel information is input to the deployed resource selection agent, which outputs the selection result, and personnel E is selected to perform the task. The information of fire extinguisher A is sent to personnel E.

[0033] (3) After personnel E completes the fire-fighting task, an event indicating that the execution is completed is triggered.

[0034] During the entire workflow execution process, the system effectively selected appropriate resources, especially human resources. After completing the execution process, the environment returned to normal and the fire incident was resolved.

[0035] Although the specific embodiments of the present invention have been described in detail, those skilled in the art will understand that various modifications and substitutions may be made to those details based on the disclosed teachings and inspirations, and these changes are within the scope of protection of the present invention. The scope of protection of the present invention is given by the attached claims and any equivalents thereof.

Claims

1. A method for constructing a human-machine-object fusion service network based on multi-agent collaboration, characterized in that The method comprises the following steps: (1) Resource modeling: Register the resources in the environment in the resource library and capability library, abstract the three types of resources: human, machine, and material, and manage them in a unified manner; (2) Crowd-sensing: Resources report data to the system, enabling the environment analyzer to obtain and identify events occurring in the environment and hand these events over to the workflow decomposer; (3) Workflow decomposition: Use intelligent agents and knowledge bases to decompose events into executable workflows; (4) Task execution: The tasks within the workflow are executed by calling the corresponding capability units, thereby advancing the workflow within the task manager.

2. The method according to claim 1, characterized in that: Each resource corresponds to a specific entity in the scenario and the functionality is shared by heterogeneous resources.

3. The method according to claim 1, characterized in that: The system operation capability unit realizes the execution of capabilities and feeds back to the system to realize task status changes.

4. The method according to claim 1, characterized in that: During workflow execution, basic personal information and capability assessment are used to connect human resources with set capabilities, dynamically build relationships between individuals and capabilities, and improve the modeling of human resources.

5. The method according to claim 4, characterized in that: The system traverses and evaluates human resources in the scene, inputs basic information and other details of the humans into the deployed resource selection agent, enabling the agent to analyze whether the individual has the set capabilities, and if so, establishes a relationship between the capabilities and the individual; the other details include name, gender, age, occupation, experience and certification.

6. The method according to claim 1 or 4, characterized in that: Using an environment information updater and an intelligent agent, the workflow is dynamically adjusted and updated based on changes in the environment.

7. The method according to claim 1, characterized in that: The design of the resource library and capability library takes into account the direct functional mapping of network services and physical devices, and is able to explicitly list its functions, API interfaces, and processed data types during resource registration.

8. A human-machine-object fusion service network construction system based on multi-agent collaboration, characterized by include: Resource modeling unit: registers resources in the environment in the resource library and capability library, abstracts and uniformly manages the three types of resources: human, machine, and object; Crowd-sensing unit: Resources report data to the system, allowing the environment analyzer to obtain and identify events that occur in the environment and hand these events over to the workflow decomposer; Workflow decomposition unit: Uses agents and knowledge bases to decompose events into executable workflows; Task execution unit: It executes tasks within the workflow by calling corresponding capability units, thereby advancing the workflow within the task manager.

9. Application of the human-machine-object fusion service network construction system based on multi-agent collaboration as described in claim 8 in fire rescue scenarios in smart buildings.

Images