Cyber-physical system

Abstract

[Problem] To provide a mechanism that enables a "thing" in the real world to autonomously operate in a cyber space and to support decision making by the "thing" that is autonomously managed. [Solution] A cyber-physical system according to an embodiment of the present invention is provided with: a virtual decision management unit and a support field management unit that prepare, in a cyber space, thought models corresponding to a thing in the real world and a support field model forming a support area in which a plurality of thought models can participate. On the basis of status information of each of the participating thought models, the support field model generates an action plan constituted of actions to be executed by the thought model and provides the action plan to the thought model. Each of the thought models receives the action plan and selects an action to be executed on the basis of the result of evaluation processing based on autonomous thought characteristics.

Description

Cyber-Physical Systems

[0001] TECHNICAL FIELD Embodiments of the present invention relate to cyber-physical system technology.

[0002] One way to utilize cyberspace (virtual space) is through CPS (cyber-physical systems). CPS is a system that creates added value by feeding back the results of analysis and accumulated information in cyberspace to the physical side (the real world). Another example of CPS is digital twins. Digital twins are a technology that "reproduces" the same manufacturing equipment and environments as in the real world in cyberspace, allowing them to operate in the same way as in the real world.

[0003] Japanese Patent Application Laid-Open No. 2020-173598

[0004] In recent years, various efforts have been made to protect global resources and aim for a sustainable society and economy. For example, efforts are being made to use "things" appropriately, for as long as possible, and with care. In order to use "things" appropriately, for as long as possible, and with care, these "things" need to be managed appropriately by people. However, it is not realistic for people to manage all "things," and it would require a great deal of effort and burden (cost, etc.).

[0005] The present invention aims to provide a mechanism for real-world "things" to operate autonomously in cyberspace, and to realize a cyber-physical system that can support decision-making for autonomously managed "things."

[0006] According to an embodiment, the cyber-physical system includes a virtual intention management unit that models a thought model corresponding to a real-world one in cyberspace and sets autonomous thinking characteristics of the thought model based on predetermined thought factors, and a support field management unit that models a support field model in cyberspace that forms a support area in which multiple thought models can participate. The support field model includes a situation understanding unit that collects situation information of each participating thought model, an action plan generation unit that generates action plans consisting of actions to be executed by the thought model based on the situation information, and an action proposing unit that provides the action plans to the corresponding thought model. The thought model includes an action plan receiving unit that receives action plans from the support field model and adds them to a candidate action list, an action evaluation unit that evaluates candidate actions based on the autonomous thinking characteristics and outputs the evaluation results, and an action selection unit that selects an action to be executed based on the evaluation results.

[0007] 1 is a conceptual diagram of a cyber-physical system according to a first embodiment. FIG. 2 is a system configuration diagram of a server-physical system according to the first embodiment. FIG. 3 is a functional block diagram of the server-physical system according to the first embodiment. FIG. 4 is a diagram illustrating a processing flow of the cyber-physical system according to the first embodiment. FIG. 5 is a diagram illustrating a configuration example of a virtual intention module that operates the thought model according to the first embodiment. FIG. 6 is a diagram illustrating a configuration example of a support field module that operates the support field model according to the first embodiment. FIG. 7 is a diagram for explaining thought factors according to the first embodiment. FIG. 8 is a diagram for explaining autonomous thinking characteristics based on thought factors according to the first embodiment. FIG. 9 is a diagram illustrating a processing flow of virtual thinking processing in the thought model according to the first embodiment. FIG. 10 is a diagram illustrating support processing of the support field model and virtual thinking processing of the thought model according to the first embodiment. FIG. 11 is a diagram illustrating virtual thinking processing of the thought model including a cooperation request for the support field model according to the first embodiment. FIG. 12 is a diagram illustrating a second embodiment, and is a functional block diagram of a server-physical system in which a support field management unit has autonomous thinking control. FIG. 13 is a diagram illustrating a processing flow of the cyber-physical system according to the second embodiment.

[0008] Hereinafter, embodiments will be described with reference to the drawings.

[0009] First Embodiment FIGS. 1 to 11 are diagrams for explaining a first embodiment.

[0010] A major feature of this embodiment is that it has a "support field model (place)" (described later) as a mechanism for making a thought model (described later) expressed in cyberspace execute virtual thought processing and operate effectively as a virtual will model. There are multiple types of thought models expressed in cyberspace. For example, there are thought models that correspond to "things" in the real world, thought models that correspond to "services" in the real world, thought models that correspond to people or organizations that provide services in the real world, and thought models that correspond to objects that are not linked to "things" in the real world.

[0011] First, we will explain the term "things" in this embodiment. "Things" refer to real-world objects, including tangible objects whose existence can be sensed (Kojien, 7th edition). In the following description, "things" such as facilities, machines, devices, products, and vehicles operating in the real world are given as examples, but are not limited to these. For example, artifacts (mobile equipment / moving bodies such as facilities, machines, devices, and vehicles, robots, structures such as buildings, bridges, and roads, products distributed or used, processed products using raw materials, products and parts manufactured by manufacturing facilities or manufacturing machines, or products and parts manufactured by hand), agricultural and livestock products (e.g., agricultural products or processed agricultural products, livestock products or processed livestock products), natural products such as plants obtained from nature (e.g., natural resources such as trees and flowers, and agricultural products grown as agricultural products), and animals (including livestock) that exist in nature can also be included as targets that can be applied to the cyber-physical system of this embodiment.

[0012] Furthermore, the "thing" described in this embodiment can include not only "things" that have processing mechanisms, connection functions, etc. in themselves, such as the above-mentioned facilities, machines, devices, and products, but also "things" that do not have processing mechanisms, connection functions, etc. In other words, for example, parts, raw materials, fluids, or aggregates thereof that constitute facilities, machines, devices, products, etc. can also be applied as "things" in this embodiment.

[0013] As mentioned above, things need to be managed in order to be used appropriately, for a long time, and with care. However, there are many "things." For example, there may be many different "things," or many identical "things." For this reason, managing many "things" individually is extremely time-consuming.

[0014] On the other hand, known CPSs can analyze collected sensor data in cyberspace and perform operations similar to those in the real world. Therefore, although known CPSs have the ability to manage "things," they are merely "reproduced" in cyberspace from real-world "things," and the reproduced "things" still need to be monitored and managed. Therefore, even if you try to use existing CPSs to manage "things" appropriately, for a long time, and with care, you still need to monitor and manage each of the many "things" reproduced in cyberspace.

[0015] In response to these challenges, the applicant came up with the idea that "Things are made by people for people to use, but by looking at the problem from the perspective of the things rather than from the perspective of people, a different solution than conventional solutions can be achieved."

[0016] In other words, from the perspective of the "thing," it can be seen as if the "thing" has a will of its own, such as "ensuring that the "thing" continues to be used for the purpose for which it was created, with its original functions and performance, for a long period of time."

[0017] Therefore, we provide a system (virtual will model) that uses software to represent "things" in cyberspace as if they were entities with will, allowing "things" to operate autonomously in cyberspace, deciding on their own what actions to take and carrying out those actions. This makes it possible to express the virtual will of "things" in cyberspace, such as continuing to be used for a long period of time (longevity for the "thing"), continuing to demonstrate the original functions and performance of the "thing" (health as a "thing"), and continuing to be useful when used (contribution as a "thing").

[0018] 1 is a conceptual diagram of a cyber-physical system according to this embodiment. In the real world, things do not actually have will, but in cyberspace, a virtual will model is created that acts according to the will of the things, as if the things had will.

[0019] For example, a thought model with goals common to many "things," such as "longevity," "health (maintaining a specified state)," and "maintaining appropriate functional performance (maintenance, repair, replenishment, etc.)," ​​can be expressed in software, and the thought model can be made to operate as a virtual will model by executing virtual thought processes.This makes it possible to provide a system in which "things" themselves can think autonomously, reducing resource consumption and waste due to consumption of "things."

[0020] The cyber-physical system of this embodiment expresses and operates a virtual will model in cyberspace as software to ensure that "things" are used appropriately, for a long time, and with care, enabling the "things" to manage themselves autonomously.

[0021] Next, the above-mentioned "service" will be explained. In addition to real-world "things," the cyber-physical system of this embodiment can also represent and operate real-world "services" provided to real-world "things" as virtual will models in cyberspace. In other words, the provision of services by people or organizations to real-world "things" in real-world social and economic activities can be represented in cyberspace as virtual will models. Not limited to this, the provision of services by people or organizations that are not provided to real-world "things" can also be represented in cyberspace as virtual will models. For example, there are services provided to people or organizations, such as management consulting. These services, rather than "things," also fall under the category of "services" in this embodiment and can be represented in cyberspace as virtual thought models. Note that the services of this embodiment are, for example, real-world services, services, benefits, etc. provided by people, organizations, things, machines, devices, products, etc., regardless of whether they are paid or free.

[0022] In addition, the thinking model that is modeled in cyberspace in response to a "thing" can be configured to hold various information associated with the "thing," such as basic information (product name, serial number, ID, manufacturing classification, etc.), configuration information (component parts, etc.), specification information (ratings and performance, etc.), status information (operating time, performance values, fuel consumption, etc.), surrounding environment information (location, temperature, humidity, wind direction, wind speed, etc.), maintenance history information (inspection results, maintenance results, part replacement, consumables, etc.), history (manufacturing location, manufacturer, user, manager, etc.), action selection as a result of autonomous thinking (described below), and execution history.

[0023] Furthermore, the thinking model that is modeled in cyberspace in response to the service provided to a "thing" can be configured to hold various information related to the service provided, such as basic information (service name, content provided, etc.), service target information (designation of the target to whom the service is provided, etc.), service specifications (service level, time and period when it can be provided, conditions for provision, location of provision, etc.), service fee information (service conditions and corresponding fees, etc.), service history information, action selection that is the result of autonomous thinking (described below), execution history, etc.

[0024] Next, we will explain other things besides "things" and "services." There are other modeling objects in cyberspace besides the above-mentioned "things" and "services." These are modeling objects that are not linked to "things" in the real world, and we will explain these. The cyber-physical system 100 of this embodiment can also use modeling objects that are not linked to "things" in the real world. The virtual intention management unit 122 can model in cyberspace a non-associative thought model that has a mutual relationship with the thought model corresponding to a "thing" and in which virtual thought processing is executed without being linked to a "thing" in the real world.

[0025] Examples of non-collaborative thinking models include an agent function that mediates between thinking models of "things," a matching function that matches with collaborators, a scheduling function that encourages the execution of actions at appropriate times and times, an optimization calculation processing function that optimizes energy consumption and action execution, and an AI function.

[0026] The non-associative thinking model is configured in the same way as the above-mentioned thinking model that is linked to real-world "things," except that it is a thinking model that is not linked to "things" in the real world. The non-associative thinking model also has autonomous thinking characteristics determined, and operates as a virtual will model that performs virtual thinking processing, which will be described below, by the cyberspace execution unit 124.

[0027] The cyber-physical system of this embodiment models a support field model in cyberspace that supports the decision-making of a virtual will model (thinking model) that acts according to the will of the "thing." The support field model provides a function to support the "thing" in determining what actions to take autonomously and how to act autonomously.

[0028] <System Configuration> Fig. 2 is a system configuration diagram of the server physical system 100 of this embodiment. Fig. 3 is a functional block diagram of the server physical system 100.

[0029] 3, the cyber-physical system 100 includes a communication device 110, a control device 120, and a storage device 130. The control device 120 includes a physical data interface unit 121, a virtual intention management unit 122, an autonomous thinking control unit 122A, a support field management unit 123, a cyberspace execution unit 124, and a cyber data interface unit 125.

[0030] 2, the physical data interface unit 121 provides information collected in the real world, for example, information collected through a predetermined data collection system (such as a sensor system), to cyberspace. The physical data interface unit 121 can acquire physical data such as sensor data from the data collection system via the communication device 110. In addition to the data collection system, the physical data interface unit 121 can also accept physical data from, for example, predetermined observation devices and equipment (including sensors, equipment signals, imaging devices, etc.) and provide the data to cyberspace. The physical data interface unit 121 can also associate the modeled thought models with the physical data, allowing each thought model to use the corresponding physical data.

[0031] In the real world (physical world), there are various "things" such as facilities, machines, devices, products, and moving objects (vehicles, aircraft, etc.), and artifacts and social infrastructure are constructed by the collection and combination of these "things." The same is true for services, which are formed by the provision of services by people or organizations for "things" in the real world, such as repair services.

[0032] The virtual intention management unit 122 models thought models corresponding to these real-world "things" and "services" in cyberspace. Specifically, it accepts input of the modeling target and generates a thought model corresponding to the input target. The virtual intention management unit 122 also includes an autonomous thought control unit 122A, which determines the autonomous thought characteristics of the generated thought model based on predetermined thought factors. The virtual intention management unit 122 may be configured as a single unit or multiple units.

[0033] The support field management unit 123 models a support field model in cyberspace that forms a support area in which multiple thought models can participate. Specifically, it accepts input for each support area to be modeled and generates a support field model corresponding to the input support area. A support area is a "field" in which multiple thought models modeled in cyberspace participate, and does not have to be a real area formed in cyberspace. In other words, the support field is the scope of application of the decision-making support function provided by the support field model, and one or more different support areas can each provide individual decision-making support functions for "things," "services," and "people or organizations" in cyberspace. For example, a thought model corresponding to a "thing" can participate in one or more different support areas. From another perspective, a "field" is formed by the aggregation of multiple thought models, and the support field model provides a mechanism for operating that "field" as a support area (decision-making support function). The same applies to thought models corresponding to "services" and "people or organizations" and unlinked thought models not linked to "things."

[0034] The cyberspace execution unit 124 is a functional unit that operates each thought model modeled in cyberspace and realizes the autonomous operation of the thought models through software control (program control). The cyberspace execution unit 124 causes the thought models to execute virtual thought processes based on autonomous thinking characteristics, so that the thought models operate as virtual will models (see Figure 1) that have virtual will in cyberspace and behave autonomously.

[0035] The cyberspace execution unit 124 is a functional unit that operates the support field model modeled in cyberspace and realizes the operation of the support field model through software control (program control). The cyberspace execution unit 124 causes the support field model to execute each process of the decision-making support function.

[0036] The cyber data interface unit 125 outputs the results of the virtual thought processing performed by each thought model in cyberspace. The results of this virtual thought processing are the results of autonomous thinking by the thought model through the virtual thought processing, or information based on the results of autonomous thinking. The cyber data interface unit 125 can, for example, output the results of the virtual thought processing to a specified display device, transmit them to a specified device via a network, or store them in a specified memory area. It can also notify specified people or organizations, which can lead to the execution of services, etc. in the real world.

[0037] FIG. 4 illustrates a processing flow of the cyber-physical system 100 according to this embodiment. A modeling target is input to the cyber-physical system 100 (S1). The modeling target is a "thing" in the real world. Input of the modeling target to the cyber-physical system 100 can be performed without using an input device such as a terminal. For example, in a manufacturing facility, each manufactured product can be detected and the detected product can be automatically input as the modeling target. In this way, a mechanism for linking with "things" in the real world can be introduced in advance, so that when a "thing" appears in the real world or is recognized in the real world, a corresponding mental model can be automatically modeled in cyberspace. As described above, there are many "things" in the real world, both the same and different. Therefore, a mechanism for automatically modeling "things" that appear in the real world or are recognized in the real world in cyberspace significantly contributes to reducing effort and costs. The system may also be configured to include a function for inputting the modeling target using an input device such as a terminal.

[0038] In addition to real-world "things," it is also possible to input modeling objects that are not linked to real-world "things" (S2).

[0039] The virtual intention management unit 122 generates a thought model corresponding to the input real-world modeling target (S101). The autonomous thought control unit 122A controls one or more different thought factors that make up the autonomous thought characteristics so that they can be set, and sets the thought factors for the generated thought model (S3). The autonomous thought control unit 122A determines the autonomous thought characteristics of the generated thought model based on the set thought factors (S102). The determined autonomous thought characteristics are applied to the corresponding thought model.

[0040] The mode of setting the thought factors can be, for example, determined in advance so that the autonomous thinking characteristics are automatically determined when generating a thought model (presetting processing of autonomous thinking characteristics).Also, it may be configured so that input means such as a terminal can be used to control input of the thought factors to be set for each thought model, and the autonomous thinking characteristics are determined.

[0041] Furthermore, a modeling target of the support area can be input to the cyber-physical system 100 (S4). The support field management unit 123 generates a support field model corresponding to the modeling target of the input support area (S103).

[0042] The cyber-physical system 100 receives input of physical data through the physical data interface unit 121 (S5).

[0043] The cyber-physical system 100 operates each thought model in the cyberspace execution domain through the cyberspace execution unit 124, and executes virtual thought processing based on autonomous thought characteristics (S104). The cyber-physical system 100 also operates the support field model in the cyberspace execution domain through the cyberspace execution unit 124, and executes processing related to decision-making support for the thought model (S105). The cyber-physical system 100 stores the results of the virtual thought processing performed by each thought model in cyberspace in the storage device 130 for each thought model (S106). Furthermore, the results of the autonomous thought of the thought model through the virtual thought processing, or information based on the results of the autonomous thought, are output through the cyber data interface unit 125 (S107).

[0044] FIG. 5 is a diagram showing an example of the configuration of the virtual intention module 200 that operates the thought model.

[0045] As described above, the thought model is configured as software (program) executed by the cyberspace execution unit 124, and one virtual thought module 200 is applied to one thought model. In the example of Fig. 5, the virtual thought module 200A and the virtual thought module 200B are applied individually to thought model A and thought model B, respectively. Note that the functions of the virtual thought modules applied to each thought model are the same.

[0046] The virtual intention module 200 includes a self-situation understanding function 201, an action evaluation function 202, an action selection function 203, an action execution function 204, an action candidate list 205, an action generation function 206, an action proposal reception function 207, and a mutual cooperation control function 208. The action proposal reception function 207 includes a notification unit 207a.

[0047] The self situation assessment function 201 performs situation assessment processing based on physical data, etc. The situation assessment processing is processing that uses the device's own physical data, etc. acquired through the physical data interface unit 121, to assess and analyze the device's own state and situation, and outputs the results. In other words, the situation assessment processing is processing that assesses the situation in order to take appropriate action depending on the situation. In this case, the situation assessment processing may be configured to analyze the influence of other thought models using, in addition to physical data related to the device itself, for example, physical data of other thought models and the results of the situation assessment processing, so as to assess the device's own state and situation.

[0048] Furthermore, self-situation understanding can also involve evaluating the reliability of a behavioral history based on, for example, an action execution history, which is the result of one's own thought process. This evaluation result can then be used to understand one's state. The evaluation result can then be quantified, such as points or virtual currency, based on predetermined standards or rules shared within a certain scope, such as the cyber-physical system, and used to understand and analyze one's own state and circumstances. Action execution history and points, which are one of the evaluation indicators, can be stored as information associated with "things" and "services" and used to understand one's state. Evaluation based on the action execution history can be based on, for example, the number of actions executed, the execution results (success, failure, completion, incompletion, etc.), and the number of action proposals accepted or rejected when received from the support field model (described below). Examples of high evaluations include a high number of actions executed or action proposals accepted, completion of execution as planned, and better-than-expected execution results.

[0049] The action evaluation function 202 performs an evaluation process of candidate actions based on the autonomous thinking characteristics and outputs the evaluation results.

[0050] The action selection function 203 selects an action to be executed based on the evaluation result output by the action evaluation function 202. In other words, the action selection function 203 is a process for extracting candidate actions that match the result of the situation understanding process based on the evaluation result.

[0051] The action execution function 204 executes the selected action. Here, the execution of the action is a process of outputting the selected execution action to the real world via the cyber data interface unit 125 or storing it in the storage device 130. The execution results of the determined execution action can also be reflected in the thought model and updated (e.g., the state of a "thing"). The results of autonomous thinking by the thought model or information based on the results of autonomous thinking are fed back to the real world via the cyber data interface unit 124 through virtual thinking processing. The results of autonomous thinking are, for example, the selected execution action described above. Furthermore, information based on the results of autonomous thinking is, for example, information related to the selected execution action that is notified to a "thing" or related people or organizations in the real world. The execution results of the selected execution action can also be fed back to the real world as information based on the results of autonomous thinking.

[0052] The action candidate list 205 holds multiple candidate actions. The candidate actions include various information such as the action type, the action execution date and time, the action content, and the execution status. The action type is information that identifies an independently executed action by the thought model itself, a mutually cooperative action that requires the cooperation of other thought models, or an action proposal (proposed action) provided by a support field model. The execution status is status information that indicates whether or not the action can be executed, and includes the statuses of candidate (not selected and not executed), confirmed execution (selected and not executed), and executed. This execution status is held as the action execution history of the thought model.

[0053] The candidate actions can be generated by the action generation function 206. For example, based on the results of the situation assessment process, a candidate action such as "Refuel at 12:00 on XX / XX / XX" can be generated. Furthermore, if the "thing" corresponding to the thought model is a moving object, an action of transporting luggage can be generated. The action generation function 206 grasps its own action schedule through the situation assessment process and generates a candidate action such as "Transport luggage from point a to point b at 10:00 on XX / XX / XX" in response to a request to transport luggage. The request to transport luggage is request information acquired from the physical data of the device itself acquired through the physical data interface unit 121 or from another thought model. Furthermore, although the candidate actions described above were generated in response to a request to transport luggage, it is also possible to directly acquire a candidate action such as "Transport luggage from point a to point b at 10:00 on XX / XX / XX" as request information from physical data or another thought model. The action candidate list 205 can hold candidate actions acquired through the physical data interface unit 121, candidate actions generated by the action generation function 206, candidate actions acquired from other thought models, and proposed actions provided by the support field model. The candidate actions can also include actions such as trading points or virtual currency circulated in the cyber-physical system with other virtual wills, and in such cases, the evaluation of the action can be influenced. This can be achieved, for example, by using a virtual will model that has as its element a thought factor that relatively highly values ​​the acquisition of predetermined points or virtual currency, and by imparting a thought characteristic that sets the weight of that thought factor high. Furthermore, when an action that includes a virtual currency transaction is selected as the execution target, the points or virtual currency transaction process is performed as part of the action execution.

[0054] The action proposal receiving function 207 receives an action proposal from the support field model and adds it to the action candidate list 205. The notification unit 207a notifies the support field model whether the provided action proposal can be executed.

[0055] The mutual cooperation control function 208 has a first function that, when the action selection function 203 selects a mutual cooperation type candidate action that requires the cooperation of another thinking model, evaluates the cooperation content presented by the other thinking model based on the autonomous thinking characteristics and performs mutual cooperation processing to form an agreement with the cooperating partner based on the results of the evaluation processing. In the first function, when an agreement is formed with the other thinking model that is the cooperating partner through the mutual cooperation control function 208, the selected candidate action is determined as the action to be executed.

[0056] The first function of the mutual cooperation control function 208 may include a mutual communication function for forming an agreement with a cooperation partner based on the results of the evaluation process and the terms of trade held by each of the thought models.

[0057] The mutual communication function may be configured to include the following transaction processing (transaction function). For example, the mutual cooperation control function 208 can communicate with each other between the thought models of itself and its partner. Therefore, for example, a known smart contract technology, in which software automatically concludes a contract, can be applied to define a transaction protocol that automatically executes a predetermined process, i.e., a transaction contract that concludes a mutual cooperation transaction, when pre-stored transaction conditions are met. The mutual communication function can execute a transaction processing (exchange to form an agreement based on each other's transaction conditions) with the mutual communication function of the partner through the transaction protocol method.

[0058] The transaction protocol method can, for example, define a series of mutual interactions in a transaction contract, and can be configured to include protocols for various processes aimed at reaching an agreement, such as negotiation, comparison, and decision processes leading up to the conclusion of the contract, as well as protocols for various processes aimed at reaching an agreement, such as accepting and holding requests and terminating the contract in each process. The cyberspace execution unit 124 provides the transaction protocol used in transaction processing by the mutual communication function, and the mutual communication function can perform transaction processing based on the transaction protocol. Note that, for example, in various processes aimed at reaching an agreement, negotiations leading up to the conclusion of the contract may not result in an agreement. In other words, the mutual communication function executes various processes aimed at reaching an agreement and engages in exchanges to reach an agreement based on the mutual transaction terms. If an agreement is not reached with the other party as a result, it determines that cooperation is not possible from that other party. The mutual cooperation control function 208 can then control the execution of the mutual communication function to reach an agreement based on the transaction terms with another cooperative partner evaluated through the above-mentioned evaluation process.

[0059] The cooperation content presented by other thought models can be obtained in advance or in real time. For example, a thought model can generate candidate actions including cooperation content that can be provided to other thought models in advance and store them in the action candidate list 205. The cyberspace execution unit 124 can perform a candidate action providing process that provides each thought model with candidate actions including cooperation content that can be provided to other thought models, which are stored in the action candidate list 205 of each thought model. Also, thought models can be configured to interactively obtain candidate actions including cooperation content from other thought models.

[0060] The mutual cooperation control function 208 can have a second function that outputs a cooperation request including the cooperation details to the support field model. While the above-mentioned first function allows each thinking model to individually find a cooperation partner, the second function outputs a cooperation request including the cooperation details to the support field model when the action selection function 203 selects a mutual cooperation-type candidate action that requires the cooperation of another thinking model in order to find a cooperation partner using the decision-making support function provided by the support field model.

[0061] FIG. 6 is a diagram showing an example of the configuration of the support field module 300 that operates the support field model.

[0062] As described above, the support field model is configured as software (program) executed by the cyberspace execution unit 124, and one support field module 300 is applied to one support field model. The functions of the support field module applied to each support field model are the same.

[0063] The support field module 300 is configured to include a situation assessment function 310 and a decision support function 320, and the decision support function 320 has a request receiving unit 321, an action plan generating unit 322, an action plan evaluating unit 323, and an action proposal control unit 324.

[0064] The situation assessment function 310 collects the self-situation information of each participating thought model. The self-situation information includes the unique information of the thought model as a "thing," the self-analysis results generated by the situation assessment process based on the physical data, and the action execution plan information.

[0065] The request receiving unit 321 receives requests input to the assistance area. Here, a request is request information that can be achieved by an action executed by one or more thought models.

[0066] The action plan generating unit 322 generates an action plan consisting of actions to be executed by the thought model in response to the received request, based on each piece of self-situation information of the thought model.

[0067] The action plan evaluation unit 323 performs an action plan evaluation process to evaluate each of the multiple action plans generated based on predetermined evaluation criteria, and extracts action plans to be proposed to the thinking model based on the action plan evaluation results.

[0068] The action proposal control unit 324 provides the action proposal extracted based on the action proposal evaluation result to the corresponding thought model.

[0069] <Explanation of Thought Factors and Autonomous Thinking Characteristics> Fig. 7 is a diagram for explaining thought factors of this embodiment. Fig. 8 is a diagram for explaining autonomous thinking characteristics based on thought factors.

[0070] As shown in Figure 7, there are five thought factors: the first thought factor governs "survival and longevity," the second thought factor governs "living and living," the third thought factor governs "health and good condition," the fourth thought factor governs "beauty and the best work," and the fifth thought factor governs "usefulness." The example in Figure 7 shows a hierarchical structure of thought factor units, with the first thought factor as the starting point (center) and the second to fifth thought factors each forming a layer.

[0071] Taking equipment as an example, the first thought factor, "Survival / Longevity," is a factor that generates, for example, the thought tendency (goal) of wanting to continue existing. The second thought factor, "Living / Life," is a factor that generates, for example, the thought tendency (goal) of wanting to continue operating. The third thought factor, "Health / Good Condition," is a factor that generates, for example, the thought tendency of wanting to maintain a specified state and operate in the best possible condition by maintaining appropriate functionality and performance (maintenance, repair, replenishment, etc.). The fourth thought factor, "Beauty / Excellent Work," is a factor that generates, for example, the thought tendency of wanting to perform at maximum performance (maximum functionality and performance). The fifth thought factor, "Usefulness," is a factor that generates, for example, the thought tendency of supporting others, providing services, and contributing to others. The fifth thought factor, "Usefulness," also includes factors similar to a person's desire for recognition from others and society by contributing to others.

[0072] Using these thought factors, the autonomous thinking characteristics of a thought model are determined based on one thought factor or a combination of multiple different thought factors. The autonomous thinking control unit 122A can hold table information that allows each thought factor to be specified, as shown in Figure 8, for example, and can control one or multiple different thought factors to be specified. Then, it can control the autonomous thinking characteristics of each thought model to be determinable based on the specified thought factors.

[0073] Furthermore, the autonomous thinking control unit 122A defines the autonomous thinking characteristics as a function composed of five thought factors, for example, as shown in Fig. 8. It can also be configured to set weight values ​​for multiple different thought factors and determine the autonomous thinking characteristics based on one or multiple different thought factors based on the set weight values.

[0074] Here, the autonomous thinking characteristics will be described with reference to the table in Figure 8. When the thought model is "equipment," the autonomous thinking characteristics designated by the second thought factor "living / living" have the characteristic of extending the operating time (allowing for long-term operation). Furthermore, the autonomous thinking characteristics designated by the third thought factor "health / good condition" have the characteristic of maintaining the condition of the equipment (device) in good condition. Furthermore, the autonomous thinking characteristics designated by the fifth thought factor "useful" have the characteristic of assisting other equipment.

[0075] An autonomous thinking characteristic designated by the third thought factor "health / good condition" and the fifth thought factor "useful" has the characteristic of supporting other equipment (devices) while maintaining one's own condition. An autonomous thinking characteristic designated by the second thought factor "life / living" and the third thought factor "health / good condition" has the characteristic of extending the operating time while maintaining the condition of equipment in good condition. Furthermore, an autonomous thinking characteristic designated by the fourth thought factor "beauty / best work" and the fifth thought factor "useful" has the characteristic of providing maximum support to other equipment.

[0076] The autonomous thinking control unit 122A can also control the thinking factors or combinations of multiple different thinking factors to dynamically change depending on the physical data corresponding to each thinking model or the results of the situation understanding process based on the physical data. It can also control the autonomous thinking characteristics once determined to be statically maintained (fixed) without dynamic change. The multiple thinking models modeled in cyberspace can also be configured to include those whose autonomous thinking characteristics, determined when the thinking model was generated, do not change, and those whose autonomous thinking characteristics change dynamically after the thinking model was generated.

[0077] Furthermore, as described above, the first thought factor ("longevity") and the third thought factor ("health (maintaining a specified state)" and "maintaining appropriate functional performance (maintenance, repair, replenishment, etc.)") can be commonly incorporated into the autonomous thinking characteristics of each thought model as thought factors (purposes) common to many "things." In other words, one or more predetermined different thought factors can be automatically set for each of the same type of "thing" or for each benefit or function provided by the "thing" to determine the autonomous thinking characteristics. This makes it possible to incorporate common thought factors (purposes) as common software or settings, rather than incorporating individual thought factors into a wide variety of "things." Conversely, even for the same type of "thing," even if the benefits and functions provided by the "thing" are the same, different thought factors or combinations of different thought factors can be set to determine the autonomous thinking characteristics.

[0078] In this way, the autonomous thinking characteristics of the thinking model of this embodiment can be determined based on one thinking factor or a combination of multiple different thinking factors. For example, it is possible to realize a virtual will model that enables autonomous decision-making, such as contributing to others, as long as it does not impair longevity or health.

[0079] The above-mentioned five thought factors are merely examples, and other thought factors may be applied to determine the autonomous thinking characteristics. Furthermore, the group of thought factors for determining the autonomous thinking characteristics may be set arbitrarily. For example, the autonomous thinking characteristics may be determined based on a group of thought factors consisting of three of the above-mentioned five thought factors, or on one or more combinations of the three thought factors.

[0080] <Virtual Thinking Processing of the Thinking Model> Figure 9 is a diagram showing the processing flow of the virtual thinking processing in the thinking model of this embodiment, which corresponds to step 104 in Figure 4. As shown in Figure 9, the virtual thinking processing consists of the above-mentioned processes: situation assessment processing (S201), generation of candidate actions based on the results of the situation assessment processing (S202), evaluation of the candidate actions based on autonomous thinking characteristics (S203), selection of candidate actions that match the results of the situation assessment processing based on the evaluation results (S204), and determination of the selected candidate action as the action to be executed (S205). In other words, the cyberspace execution unit 124 executes the virtual thinking processing through the virtual intention module 200, thereby realizing a mechanism for the thinking model to operate autonomously.

[0081] If a mutual cooperation type candidate action that requires the cooperation of another thinking model is selected, a mutual cooperation process is performed between steps S204 and S205 in Fig. 9, in which the first function of the above-mentioned mutual cooperation control function 208 evaluates the cooperation content presented by the other thinking model based on the autonomous thinking characteristics and forms an agreement with the cooperating partner based on the results of the evaluation process. Then, when an agreement is formed with the other cooperating thinking model through the mutual cooperation control function 208, the selected candidate action is determined as the action to be executed.

[0082] Here, a specific example of a virtual thinking process based on autonomous thinking characteristics will be described using the case where the "thing" corresponding to the thought model is a mobile object (hereinafter, the thought model "mobile object") as an example. Assume that the second thought factor "living / living" is specified as the autonomous thinking characteristic. In the virtual thinking process shown in FIG. 9 , assume that the result of the situation assessment process in step S201 indicates that the degree of deterioration of the drive battery is below a specified value. In this case, the thought model "mobile object" can assess that there is no problem in continuing operation without replacing the drive battery, since the degree of deterioration of the drive battery is below the specified battery replacement value. In step S203, the thought model "mobile object" determines that replacement of the drive battery is not necessary and evaluates candidate actions based on the autonomous thinking characteristics.

[0083] For example, the candidate actions "Since the degree of deterioration is below the battery replacement threshold, charge to full charge capacity" and "Although the degree of deterioration is below the battery replacement threshold, charge to 80% of full charge capacity" are extracted, and each candidate action is evaluated based on the autonomous thinking characteristic. Since the second thinking factor "living / living" is specified in the autonomous thinking characteristic, the candidate action of charging to full charge capacity is highly evaluated in order to enable long-term operation. In step S204, the thinking model "mobile body" selects a candidate action that matches the result of the situation understanding process based on the evaluation result.

[0084] The candidate actions "Since the degree of deterioration is less than the battery replacement standard value, charge to full charge capacity" and "Although the degree of deterioration is less than the battery replacement standard value, charge to 80% of full charge capacity" may be generated based on the results of the situation assessment process in step S202 of Figure 9, or may be candidate actions generated in advance.

[0085] The evaluation of actions regarding the need for battery replacement or charging is not limited to the above, and evaluation criteria vary depending on the situation. For example, the candidate action of "not replacing the battery even if it is below the specified value" may be highly evaluated based on the usage history and performance history of the user's own battery, as well as information from the information exchange community about the optimal timing for battery replacement specified for the battery.

[0086] Next, suppose the result of the situation assessment process in step S201 in Figure 9 indicates that the degree of deterioration of the drive battery is equal to or greater than a specified value. In this case, the "mobile body" thinking model can evaluate, based on its autonomous thinking characteristics, that the drive battery should be replaced because the degree of deterioration of the drive battery is equal to or greater than the battery replacement specified value. In step S203, the "mobile body" thinking model determines that the drive battery needs to be replaced and evaluates candidate actions based on its autonomous thinking characteristics.

[0087] For example, the candidate action "Vendor A: Same-day replacement possible, delivery in as little as two days" and the candidate action "Vendor B: One week delivery time to determine battery load status and select replacement battery" are extracted, and each candidate action is evaluated based on the autonomous thinking characteristics. Since the second thinking factor "Life / Living" is specified in the autonomous thinking characteristics, candidate actions that allow for early battery replacement work are highly evaluated because it is desired to keep the device running for a long time. The "Mobile" thinking model executes a mutual communication process with a cooperation partner that has proposed cooperation content that allows for early battery replacement work, and determines the candidate action corresponding to the cooperation partner based on the mutual communication process as the action to be executed. In other words, in step S204, the "Mobile" thinking model selects a candidate action corresponding to the cooperation partner based on the mutual communication process as a candidate action that matches the results of the situation understanding process.

[0088] The candidate action "Vendor A: Same-day replacement possible, delivery in as little as two days" and the candidate action "Vendor B: Delivery time is one week, as the battery to be replaced is selected based on the battery load condition" may be generated based on the results of the situation assessment process in step S202 of Figure 9, or may be a candidate action generated in advance.

[0089] Judgments and choices based on the above-mentioned autonomous thinking characteristics will differ if the autonomous thinking characteristics differ. For example, a thinking model "Mobile" with autonomous thinking characteristics that specify the second thinking factor "Living / Life" and the third thinking factor "Health / Good Condition" will have the characteristic of extending the operating time while maintaining the condition of the mobile in good condition. Therefore, even if the degree of deterioration of the drive battery is below a specified value, it will determine that the drive battery needs to be replaced and highly evaluate and select the candidate action for battery replacement. Judgments and choices will differ from autonomous thinking characteristics that specify only the second thinking factor "Living / Life."

[0090] In this way, by performing thought processing using each of the first to fifth thought factors, things can autonomously aim for longevity and health (maintenance of condition), which in turn can contribute to reducing resource consumption and waste, and contributing to the global environment, etc.

[0091] The cyber-physical system 100 of this embodiment can realize a cyberspace in which "things" in the real world, which do not actually have wills, virtually have wills in cyberspace and act according to the will of the "things."

[0092] Specifically, we provide a mechanism that allows a thinking model modeled in cyberspace to execute virtual thinking processing based on autonomous thinking characteristics in a software manner, making it operate according to the will of the "thing." By executing virtual thinking processing based on autonomous thinking characteristics, each thinking model can be made to operate as a virtual will model that behaves with autonomous will.

[0093] Therefore, it will become possible for real-world objects to be autonomously managed by themselves, so that they can be used appropriately, for a long time, and with care.

[0094] In the cyber-physical system 100 of this embodiment, "things" in the real world are modeled as thought models in cyberspace, enabling them to operate autonomously without human intervention, thereby achieving the following specific effects.

[0095] Even simple objects that do not have computing capabilities can have software that expresses virtual will, dramatically increasing the number of things participating in cyberspace and dramatically expanding the scope in which "things" can connect and cooperate with each other.

[0096] Furthermore, there is no need for people to manage all "things," and "things" will be able to autonomously manage their own lifespan (life) and condition (health), eliminating the burden of managing "things" on people. This will also enable new benefits to be provided, such as appropriate condition maintenance and maintenance for "things," and offerings for the next user and uses that suit the user's life stage.

[0097] The autonomous thinking characteristic is implemented as a virtual will of the "thing," with thought factors (purposes) common to many things, such as "longevity / survival" and "maintenance of health / condition," being implemented as common software, so there is no need to develop and implement software for each individual "thing."

[0098] In human society, people with wills cooperate with each other to accomplish things that cannot be done alone and to create new value. Similarly, the cyber-physical system 100 of this embodiment allows "things" represented as "virtual will model" software to build mutual cooperative relationships in cyberspace, thereby enabling the creation of new value that has not been possible before.

[0099] Based on the above, we can contribute to realizing an economy and society that contributes to the protection of global resources by using many "things" appropriately for as long as possible, rather than the traditional model of mass consumption and mass disposal.

[0100] In particular, by incorporating the concept of a "virtual intention model," things, people, services, etc. are placed on the same playing field in cyberspace, and virtualized so that they can be handled in the same way, which is a notable feature that differs from known CPS technologies.

[0101] <Decision-making support function of the support field model> As described above, each of the multiple thought models modeled in cyberspace can select and execute autonomous actions by executing virtual thought processing.

[0102] For example, if the "thing" is a mobile entity, it provides a service of "delivering luggage." In this case, when a delivery request for luggage is received, the "mobile entity" mental model attempts to carry out the service of delivering the luggage. Because the "mobile entity" mental model is linked to the "thing" in the real world, it can directly receive the candidate action of "delivering luggage to a specified location" as physical data, or can receive delivery requests from other mental models. Based on its autonomous thinking characteristics, the "mobile entity" mental model can select and execute the candidate action of "delivering luggage to a specified location" as an action that matches the results of the situation understanding process. Therefore, the remarkable effects of the cyber-physical system described above can be obtained.

[0103] However, each thinking model has difficulty selecting actions for areas (external) in which it is not directly involved. For example, it cannot select the action of delivering packages until a request is received. Furthermore, in order to receive requests from other thinking models, the thinking model itself must search for requests and demonstrate that it can undertake delivery services. In other words, a thinking model can only select and execute actions using its own means, i.e., only in areas in which it can be involved starting from itself.

[0104] A thinking model can take autonomous action in cyberspace through a virtual thinking process in which it "decides its own will = chooses an action." However, if the decision-making structure is limited to the scope that originates from the self, this will lead to restrictions on the actions that the thinking model can take.

[0105] Therefore, in this embodiment, a support field model that forms a support area in which multiple thinking models can participate is modeled in cyberspace, and an environment is realized that supports decision-making for ``things'' that are managed autonomously without being limited to the scope starting from the self.

[0106] 10 is a diagram illustrating the support process of the support field model and the virtual thinking process of the thinking model in this embodiment. The thinking model participates in the support area F1 formed by the support field model in advance, and the support field model holds a participant list (not shown).

[0107] The support field model collects self-status information of each participating thought model (S301). As described above, the self-status information includes unique information of the thought model as a "thing," self-analysis results generated by the situation understanding process based on physical data, and action execution schedule information. In the case of a mobile thought model, the unique information includes the movement speed, maximum load capacity, service life, actual operating time, etc., and the self-analysis results include the battery charge amount, battery deterioration level, and current travelable distance. The action execution schedule information is action schedule information for each selected candidate action, and is the action schedule information of the thought model including the date and time of action execution. The situation understanding function 310 acquires self-status information from each participating thought model at any timing.

[0108] Next, the support field model receives a "request" from a third party other than the thought model corresponding to the "item" (S302). For example, the request receiving unit 321 can receive the "request" as physical data through the physical data interface unit 121. For example, the request "Deliver package M from private residence p to private residence q" is received. The request includes the scheduled date and time of execution of the request.

[0109] In response to the received request, the support field model performs an action plan generation process (S303). Specifically, based on the self-situation information of each participating thought model, an action plan consisting of actions to be executed by the thought model in response to the received request is generated. For example, assume that the thought models participating in the support area F1 are drones A, B, and trucks P, Q, and R. In this case, the following action plans can be generated by referring to the self-situation information of each thought model. Action Plan 1: Drone A picks up package M at a private residence p and transports it to point a. Truck Q picks up package M at point a and transports it to collection and delivery point z. Action Plan 2: Drone B picks up package M at a private residence p and transports it to point a. Truck Q picks up package M at point a and transports it to collection and delivery point z. Action Plan 3: Drone A picks up package M at a private residence p and transports it to point a. Truck R picks up package M at point a and transports it to collection and delivery point z. Action Plan 4: Drone B picks up package M at a private residence p and transports it to point a. Truck R picks up cargo M at point a and transports it to collection and distribution point z.

[0110] The generation of the action plan can be performed by applying a known method such as AI. The action plan generation unit 322 can generate an action plan by performing optimization processing according to the self-situation information of each thought model in response to the request "Deliver package M from private home p to private home q."

[0111] An example of generating action plan 1 will be described. For example, drone A can ascertain from its own status information that, as a candidate action corresponding to the scheduled execution date and time of the request, there is a plan to transport luggage from point f to point a, followed by aerial photography. At this time, the action plan generation unit 322 confirms that drone A's payload capacity, battery charge, and the like are sufficient, and determines that drone A can "fly to private residence p, collect luggage M, and deliver it to point a" before transporting the luggage from point f to point a, and can generate an action plan for drone A.

[0112] On the other hand, truck Q can also ascertain from its own situation information that it has a plan to transport package K to loading point g at point a as a candidate action corresponding to the scheduled execution date and time of the request, and that it is scheduled to refuel at point g afterwards. At this time, the action plan generation unit 322 confirms that truck Q has sufficient payload, remaining fuel, etc., and determines that it is possible to load package M together with package k at point a and transport package M to collection and delivery point z before going to point g, and can generate an action plan for trunk Q that "pick up package M at point a and transport it to collection and delivery point z."

[0113] The action plan generation unit 322 can generate action plan 1 consisting of actions to be executed by multiple different thinking models (drone A, truck Q) in response to the request "Deliver package M from private home p to private home q."

[0114] In the above example, an action plan consisting of actions executed by multiple different thinking models is generated for the received request, but this is not limited to this. For example, there may be cases where the request can be achieved by the actions of a single thinking model. In this case, an action plan consisting of actions executed by a single thinking model is generated.

[0115] When multiple different action plans are generated, the action plan evaluation unit 323 performs an action plan evaluation process to evaluate each of the multiple action plans based on predetermined evaluation criteria (S304), and extracts an action plan to be proposed to the thinking model based on the action plan evaluation results (S305).

[0116] The predetermined evaluation criterion may be any criterion, for example, an evaluation criterion may be applied in which an action plan with a small difference between the planned action of the "thing" included in the action candidate list and the generated action plan is highly evaluated.

[0117] Taking drone A and truck Q as an example, distance, time, or cost can be considered as the difference from the planned actions of drone A and truck Q. That is, the difference in physical travel distance can be evaluated by how much the proposed action will increase, and the smaller the increase in travel distance due to the proposed action, the higher the evaluation. The same applies to time and cost, and the amount of time required by the proposed action and the amount of cost required by the proposed action can be applied as evaluation criteria, and the smaller the difference, the higher the evaluation.

[0118] For example, the differences in distance, time, and cost can be calculated for each action plan as follows. The fee is an amount according to the difference in distance based on the arbitrary transportation fees for drones and trucks. Action plan 1: Drone A's difference (deviation) = 300 m (flight time at 36 km / h: over 30 seconds) (fee: 600 yen) Truck Q's difference (deviation) = 400 m (travel time at 60 km / h: over 24 seconds) (fee: 300 yen) Action plan 2: Drone B's difference (deviation) = 700 m (flight time at 36 km / h: over 70 seconds) (fee: 1,200 yen) Truck Q's difference (deviation) = 400 m (travel time at 60 km / h: over 24 seconds) (fee: 300 yen) Action plan 3: Drone A's difference (deviation) = 300 m (flight time at 36 km / h: over 30 seconds) (fee: 600 yen) Truck P's difference (deviation) = 600 m (travel time at 60 km / h: over 36 seconds) (Fee: 300 yen) Action plan 4 Drone B's difference (deviation) = 700 m (flight time at 36 km / h: over 70 seconds) (Fee: 1,200 yen) Truck P's difference (deviation) = 600 m (travel time at 60 km / h: over 36 seconds) (Fee: 300 yen)

[0119] The action plan evaluation unit 323 selects action plan 1 with small differences in distance, time, and cost (fee). While the case where distance, time, and cost are all taken into account as evaluation criteria has been described as an example, the present invention is not limited to this. For example, any one of distance, time, and cost may be applied as the evaluation criterion, or any combination of these may be applied as the evaluation criterion. For example, an action plan with a small time difference even if it is costly may be highly rated, or an action plan with a small cost difference even if it takes a long time may be highly rated.

[0120] In the above example, the drone and truck had scheduled actions (candidate actions selected) around the scheduled execution date and time of the request "Deliver package M from private residence p to private residence q," but there may be cases where no scheduled actions exist around the scheduled execution date and time of the request. In this case, for example, in action plan 1, the distance, time, and cost required for drone A to travel from its waiting location to private residence p, pick up package M, and transport it to point a become the difference from the scheduled action.

[0121] The action plan proposal control unit 324 presents action plan 1 to each thought model of drone A and truck Q included in action plan 1 extracted based on the evaluation results (S306). The thought model of drone A is presented with an action plan of "picking up package M at private residence p on November 1st, 2000, at 10:10 and transporting it to point a." The thought model of truck Q is presented with an action plan of "picking up package M at point a on November 1st, 2000, at 10:25 and transporting it to collection and delivery point z." Note that the action plan evaluation unit 323 performs the evaluation process after understanding the status of drone A and truck Q, such as their planned activities, and therefore can present action plans that include the date and time for executing the action.

[0122] As shown in Figure 6, the situation understanding function 310 can collect not only the self-situation information of each participating thought model, but also physical data related to the environment of the "things" in the real world corresponding to the thought model. For example, it can collect natural environment information such as weather information and disaster information, and social environment information such as traffic congestion information. The action proposal generation unit 322 can be configured to generate action proposals based on the self-situation information of the thought model and physical data related to the environment in the real world.

[0123] Next, the function of the thought model side where action plans are presented will be described.

[0124] A plurality of different thought models participate in the support area F1 formed by the support field model. Each thought model of drone A and truck Q corresponding to action plan 1 receives the above-mentioned action plan in their respective action plan receiving functions 207 (YES in S210). The action plan receiving function 207 adds the action plan to the action candidate list 205 (S211).

[0125] Upon receiving the proposed action, the thinking model performs a situation assessment process based on the physical data, etc. (S212). As a result of the situation assessment process, the thinking model considers new candidate actions and generates candidate actions by itself (S213). The processes of steps S212 and S213 take into consideration the possibility that adding the proposed action to the action candidate list 205 may cause a change in the candidate actions that the thinking model itself can select.

[0126] The thinking model then evaluates candidate actions, including proposed actions, based on the autonomous thinking characteristics. Specifically, the action evaluation function 202 extracts candidate actions around the scheduled execution date and time of the proposed action presented from the support field model, and performs an evaluation process for the extracted candidate actions. In other words, the action evaluation function 202 extracts one or more candidate actions around the scheduled execution date and time of the proposed action as evaluation targets, based on the user's own unique information, self-analysis results, and candidate actions that are scheduled to be performed, as grasped in the situation grasping process. The proposed action is included in the evaluation targets.

[0127] For example, the thinking model of drone A can extract the following candidate actions as evaluation targets. (1) Candidate Action 1 Action type: proposed action, execution status: candidate. Action content: "On November 1st, 2000, at 10:10, pick up package M from private residence p and transport it to point a." (2) Candidate Action 2 Action type: independent execution action, execution status: candidate. Action content: "On November 1st, 2000, at 10:00, transport package M from point f to point a." (3) Candidate Action 3 Action type: independent execution action, execution status: candidate. Action content: "On November 1st, 2000, at 10:30, take aerial photographs of point h." (4) Candidate Action 4 Action type: independent execution action, execution status: candidate. Action content: "On November 1st, 2000, at 11:00." (5) Candidate action 5. Action type: Mutual cooperation action, execution status: Candidate. Action content: “Refill fuel at 12:00 on November 1st, year XX.”

[0128] Candidate action 1 is an action proposal presented by the support field model, candidate actions 2 to 4 are actions generated by the user, which are actions scheduled to be executed and stored in the action candidate list 205 before the action proposals are presented, and candidate action 5 is a newly generated action in response to a change that has occurred in the candidate actions that the user can select by accepting an action proposal from the support field model.

[0129] Similarly, the thought model of truck Q also performs steps S210 to S213, and the following candidate actions can be extracted as evaluation targets. (a) Candidate action 1-1 Action type: proposed action, execution status: candidate. Action content "On November 1st, 2000, at 10:25, pick up package M at point a and transport it to collection and delivery point z." (b) Candidate action 1-2 Action type: independent execution action, execution status: candidate. Action content "On November 1st, 2000, at 10:30, transport package K to loading point g at point a." (c) Candidate action 1-3 Action type: independent execution action, execution status: candidate. Action content "On November 1st, 2000, at 10:00, transport package J from point d to point a." (d) Candidate action 1-4 Action type: independent execution action, execution status: candidate. Action content: "Refuel at point g at 11:00 on November 1st, XXX."

[0130] Candidate action 1-1 is an action proposal presented by the support field model, candidate actions 1-2 and 1-3 are actions generated by the user, and are actions scheduled for execution that were stored in the action candidate list 205 before the proposed action was presented, and candidate action 1-4 is a newly generated action in response to a change that occurred in the candidate actions that the user could select by accepting the action proposal from the support field model.

[0131] Next, the action evaluation function 202 of drone A performs evaluation processing based on the autonomous thinking characteristics for the candidate actions to be evaluated (including the proposed action) (S214). For example, by comparing the candidate actions to be evaluated, if there is no overlap in time (start time, execution time, etc.), evaluation processing can be performed based on evaluation criteria that highly evaluate the proposed action presented by the support field model.

[0132] Drone A's candidate action 1 (the proposed action proposed by support area F1) is "November 1, 2000, 10:10 AM, pick up package M from residence p and deliver it to point a," and candidate action 2 is "November 1, 2000, 10:10 AM, pick up package M from residence p and deliver it to point a." The difference in start time between candidate action 1 and candidate action 2 is 10 minutes, and the package's destination is the same, point a. Therefore, even if the drone stops at residence p, it can be determined that this will not affect the subsequent scheduled candidate actions 3 and 4. For example, a threshold can be set for the difference in start time between candidate actions, and if there is a time allowance equal to or greater than the threshold, control can be exercised to incorporate the new candidate action into the execution schedule.

[0133] Then, as described above, based on the evaluation criteria of highly evaluating the proposed actions presented by the support field model, the action evaluation function 202 outputs an evaluation result in which candidate action 1 receives the highest evaluation.

[0134] Then, the action selection function 203 selects candidate action 1 (the proposed action presented from the support area F1) (S215). The action selection function 203 updates the status information indicating whether candidate action 1 in the action candidate list can be executed to execution confirmed (selected but not yet executed).

[0135] The notification unit 207a of the action proposal receiving function 207 notifies the support field model of whether the proposed action can be executed based on the action selection result (S216). Since the action selection function 203 selected candidate action 1 linked to the proposed action presented by the support field model, the notification unit 207a notifies the support field model of a response of "OK to execute proposed action 1." Note that the process of updating the status information indicating whether candidate action 1 can be executed to "confirmed execution" (selected but not yet executed) may be configured to be performed by the notification unit 207a. In other words, the status information indicating whether candidate action 1 can be executed may be updated to "confirmed execution" (selected but not yet executed) upon notification of whether the proposed action can be executed to the support field model.

[0136] As will be described later, if the presented action plan is not selected, the notification unit 207a of the thought model notifies the support field model of the answer "Execution of action plan 1 is NG."

[0137] Similarly, the action evaluation function 202 of the track Q performs evaluation processing based on the autonomous thinking characteristics for the candidate actions to be evaluated (including the proposed action) (S214). For example, by comparing the candidate actions to be evaluated, if there is no overlap in time (start time, execution time, etc.), evaluation processing can be performed based on evaluation criteria that highly evaluate the proposed action presented by the support field model.

[0138] Truck Q's candidate action 1-1 (a proposed action from support area F1) is "Pick up package M at point a and transport it to collection and distribution center z at 10:25 on November 1, 2000," and candidate action 1-2 is "Pick up package M at point a and transport it to collection and distribution center z at 10:30 on November 1, 2000." The difference in start time between candidate actions 1-1 and 1-2 is five minutes, and although the destinations of the packages are different, collection and distribution center z and point g are close in distance, so it can be determined that the impact on the execution of candidate action 1-2 is low. For example, a threshold can be set for the difference in start time between candidate actions, and if there is a time allowance equal to or greater than the threshold, control can be performed to incorporate a new candidate action into the execution schedule.

[0139] Then, based on the evaluation criteria of highly evaluating the proposed actions presented by the support field model, the action evaluation function 202 outputs an evaluation result in which the candidate action 1-1 receives the highest evaluation.

[0140] Then, the action selection function 203 of the track Q selects the candidate action 1-1 (the proposed action presented from the support area F1) (S215). The action selection function 203 updates the status information indicating whether the candidate action 1-1 in the action candidate list can be executed to execution confirmed (selected but not yet executed).

[0141] The notification unit 207a of truck Q notifies the support field model of whether the proposed action can be executed based on the action selection result (S216). Because the action selection function 203 selected candidate action 1-1 linked to the proposed action presented by the support field model, the notification unit 207a notifies the support field model of the answer "OK ​​to execute proposed action 1." Note that, similar to drone A, the notification unit 207a may be configured to update the status information indicating whether candidate action 1-1 can be executed to execution confirmed (selected but not yet executed).

[0142] Then, the thought models of drone A and truck Q execute candidate action 1 and candidate action 1-1, which have been confirmed for execution (S217). After execution, the status information indicating whether candidate action 1 and candidate action 1-1 can be executed is updated to "executed."

[0143] In this way, the cyber-physical system of this embodiment realizes a mechanism in which the support field model (support area F) generates and provides action proposals after understanding the situation of the "thing." This allows actions outside of the area in which one can be involved to be taken from one's own perspective, and supports the expansion of decision-making in the thinking model corresponding to the "thing." In particular, for actions that must be carried out in cooperation with multiple "things," the "things" can enjoy the action without being aware of the situation of the other "things," and action generation, evaluation, and selection can be efficiently performed in cyberspace.

[0144] In particular, each thinking model can create an environment in which people can choose actions that will enable them to perform their roles in creating value outside of their own scope (external), without being limited by the information that comes to them or the information that they find themselves.

[0145] Furthermore, different evaluation criteria for each support field model can be applied. In other words, it is possible to set for each support field model the evaluation criteria that are emphasized when evaluating action plans and extracting action plans based on the evaluation results. This allows for action plans to be provided by other support field models even if no action plans are provided by one support field model, thereby expanding the scope of the services that can be performed. Furthermore, it is also possible to build a support field model that generates action plans specialized for requests in a certain field, and to assign different roles to each support field model, and to configure it so that action plans corresponding to that role are generated and provided to the thinking model.

[0146] 10, the support field model receives a response from the corresponding thinking model to the proposed action plan (S307). The action proposal control unit 324 receives the feasibility of the proposed action notified from the thinking model, and records the feasibility of one or more actions constituting the proposed action (S308).

[0147] The action proposal control unit 324 determines whether or not there is a response of "no execution" to the proposed action plan (S309). If the response of the proposed action plan does not include "no execution" (NO in S309), a confirmation process for the action plan for the request is performed (S310). This confirmation process can be configured to include a notification to the requester who input the request into the support field model (support area F1).

[0148] On the other hand, in step S309, if the answer to the proposed action plan includes "no execution" (YES in S309), a cancellation process is performed on the request. In the cancellation process, when the action plan is composed of actions from multiple different thought models, the action proposal control unit 324 notifies the thought model that answered "yes" of the cancellation of the action plan. In other words, by determining whether other thought models other than the thought model that notified "no execution" are involved in (included in) the action plan, and notifying the other involved thought models that the proposed action plan is invalid, the thought model that answered "yes" to the presented action plan can cancel the candidate action that has been confirmed for execution.

[0149] Next, when the action plan generation unit 322 receives a notification from a thinking model that the provided action plan will not be implemented, the action plan generation unit 322 performs a re-proposal process to regenerate an action plan that excludes the action plan for the thinking model that notified the "no implementation." In the following, when a notification of "no implementation" is received, an action plan is regenerated that excludes the thinking model itself (hereinafter referred to as "track Q") that notified the "no implementation." However, this is not limited to this. That is, the action plan generation unit 322 may perform a re-proposal process to regenerate an action plan that includes the thinking model (track Q) that notified the "no implementation" by changing the conditions for the thinking model (track Q) that were used to generate the action plan that was not implemented. Furthermore, after a notification of "no implementation" is received, the situation assessment function 310 may assess changes in the situations of all thinking models participating in the support field model that includes the thinking model that notified the "no implementation" and regenerate the action plan that was not implemented. In this way, the action plan generation unit 322 generates various action plans even in response to a notification of "no implementation."

[0150] For example, in the case of the above-mentioned action plan 1, the thinking models of both drone A and truck Q compare the candidate actions to be evaluated, and if there is no overlap in time (start time, execution time, etc.), they perform an evaluation process based on the autonomous thinking characteristic evaluation criteria that highly evaluates the action plan presented by the support field model. At this time, truck Q is assumed to have evaluation criteria that further emphasize the weight of thinking factors.

[0151] In this case, truck Q's candidate action 1-1 (the proposed action presented by support area F1) is "Pick up package M at point a and transport it to collection and distribution center z at 10:25 on November 1, 2000," and candidate action 1-2 is "Pick up package M at point a and transport it to collection and distribution center z at 10:30 on November 1, 2000." The difference in start time between candidate action 1-1 and candidate action 1-2 is five minutes. Although the destinations of the packages are different, collection and distribution center z and point g are close in distance, so it can be determined that the impact on the execution of candidate action 1-2 is low. On the other hand, truck Q has a high weighting for the "health / good condition" thinking factor in its autonomous thinking traits. Therefore, if truck Q selects candidate action 1-1, the travel distance will be longer in relation to the execution of other actions, and candidate action 1-1 (the proposed action) will be evaluated lower than the other candidate actions.

[0152] Therefore, by combining the first evaluation criterion that highly evaluates the action proposal presented by the support field model with the second evaluation criterion of the thinking factor "health / good condition" of the autonomous thinking characteristics, the action evaluation function 202 of track Q performs evaluation processing so that the evaluation of candidate action 1-1 presented by the support field model is lower than the evaluation of candidate action 1-2. In this case, the notification unit 207a of track Q notifies the support field model of the answer "no execution" in step S216 of FIG.

[0153] The first evaluation criterion, which highly evaluates the proposed action from the support field model, can also be considered a thinking factor of the autonomous thinking characteristics. While the thinking factors are illustrated in Figures 7 and 8, for example, a thinking factor of "listening to external opinions" can also be set. In this case, virtual thinking processing is performed to highly evaluate the candidate action proposed from the outside. Therefore, the first evaluation criterion and the second evaluation criterion can be considered as autonomous thinking characteristics of the thinking model, and can also be configured to include evaluation criteria separate from the autonomous thinking characteristics of the thinking model.

[0154] The action plan generation unit 322 receives a notification from the thought model of track Q that the provided action plan 1 is not to be carried out, and so generates an action plan again excluding the action plan for the thought model of track Q that has been notified of "not to be carried out."

[0155] The action plan generator 322 generates action plans 3 and 4 as proposal candidates from among the above-mentioned action plans 1 to 4, excluding action plans that include the action of track Q.

[0156] Action plan 3 Drone A picks up package M at a private residence p and delivers it to point a. Truck R picks up package M at point a and delivers it to collection and delivery point z. Action plan 4 Drone B picks up package M at a private residence p and delivers it to point a. Truck R picks up package M at point a and delivers it to collection and delivery point z.

[0157] Note that, before the re-proposal process of the action plan generation unit 322, the situation assessment function 310 may be configured to refer to the latest situation information of all the thinking models (drones A, B, trucks P, Q, R) participating in the support field model and generate an action plan for the "request." In this case, the "request" remains unchanged and is "Deliver package M from private home p to private home q."

[0158] Similar to the evaluation method described above, the action plan evaluation unit 323 can consider distance, time, or cost as the difference from the planned actions of drone A and truck R. In other words, it can be configured to evaluate how much the difference in physical travel distance will increase due to the action plan, and to give a higher evaluation the smaller the increase in travel distance due to the action plan.

[0159] For example, the differences in distance, time, and cost can be calculated for action plans 3 and 4. Action plan 3 Drone A's difference (deviation) = 300 m (flight time at 36 km / h: over 30 seconds) (Fee: 600 yen) Truck P's difference (deviation) = 600 m (travel time at 60 km / h: over 36 seconds) (Fee: 300 yen) Action plan 4 Drone B's difference (deviation) = 700 m (flight time at 36 km / h: over 70 seconds) (Fee: 1,200 yen) Truck P's difference (deviation) = 600 m (travel time at 60 km / h: over 36 seconds) (Fee: 300 yen)

[0160] The action plan evaluation unit 323 selects action plan 3, which has small differences in distance, time, and cost (fee).

[0161] The action plan proposal control unit 324 presents action plan 3 to each of the thought models of drone A and truck P included in action plan 3 extracted based on the evaluation results (S306). The action plan presented to drone A's thought model is "Pick up package M at private residence p at 10:10 on November 1st, 2000, and transport it to point a." The action plan presented to truck P's thought model is "Pick up package M at point a at 10:25 on November 1st, 2000, and transport it to collection and delivery point z." Note that the action plan evaluation unit 323 performs the evaluation process after understanding the status of drone A and truck P, such as their planned activities, and therefore can present action plans that include the date and time for executing the action.

[0162] When presented with action plan 3, drone A performs the same evaluation process as it did for action plan 1, selecting the candidate action for action plan 3, "Pick up package M at private residence p and transport it to point a at 10:10 on November 1st, 2000," and updates the status information to confirm execution, while notifying the support field model of the answer "possible to execute" (S210 to S217 in Figure 10).

[0163] The thought model of truck P also performs steps S210 to S213, and the following candidate actions can be extracted as evaluation targets. (a) Candidate action 3-1 Action type: proposed action, execution status: candidate. Action content "On November 1st, 2000, at 10:25, pick up package M at point a and transport it to collection and delivery point z." (b) Candidate action 3-2 Action type: proposed action, execution status: confirmed. Action content "On November 1st, 2000, at 10:00, transport package P from point c to private residence r." (c) Candidate action 3-3 Action type: independently executed action, execution status: candidate. Action content "On November 1st, 2000, at 11:30, perform maintenance at point f."

[0164] Candidate action 3-1 is the proposed action currently presented by the support field model, candidate action 3-2 is another action presented by the support field model whose execution has already been confirmed, and candidate action 3-3 is an action scheduled for execution that was stored in the candidate action list 205 before the proposed action was presented.

[0165] The action evaluation function 202 of the track P performs evaluation processing based on the autonomous thinking characteristics for the candidate actions to be evaluated (including the proposed action) (S214). For example, when the candidate actions to be evaluated are compared and there is no overlap in time (start time, execution time, etc.), evaluation processing can be performed based on evaluation criteria that highly evaluate the proposed action presented by the support field model. In this case, it is assumed that the track P also has evaluation criteria that emphasize the weight of thinking factors.

[0166] Candidate action 3-1 is "Pick up package M at point a at 10:25 on November 1, 2000, and deliver it to collection and distribution center z." Candidate action 3-2 is "Deliver package P from point c to residence r at 10:00 on November 1, 2000," which is presented by the support field model and confirmed for execution with an "executable" response (OK response). Candidate action 3-2 is an action recorded in the action candidate list 205 as confirmed for execution. The difference in start time between candidate action 3-1 and candidate action 3-2 is 25 minutes, and the destinations of the packages are different. However, it can be determined that there is little time impact even if truck P departs from point c, picks up package M at point a, delivers package M to collection and distribution center z, and then delivers package P to residence r. On the other hand, truck P evaluates candidate action 3-1 higher than the other candidate actions because the weight of the "useful" thinking factor in the autonomous thinking trait is set high.

[0167] Therefore, by combining the first evaluation criterion of highly evaluating the action proposals presented by the support field model with the second evaluation criterion of the thinking factor "useful" of the autonomous thinking characteristic, the action evaluation function 202 of track P performs evaluation processing so that the evaluation of candidate action 3-1 presented by the support field model is higher than candidate actions 3-2 and 3-3.

[0168] Then, the action selection function 203 of the track P selects the candidate action 3-1 (the proposed action presented from the support area F1) (S215). The action selection function 203 updates the status information indicating whether the candidate action 3-1 in the action candidate list can be executed to execution confirmed (selected but not yet executed).

[0169] The notification unit 207a of the truck P notifies the support field model whether the proposed action can be executed based on the action selection result (S216). Since the action selection function 203 selected the candidate action 3-1 linked to the proposed action presented by the support field model, the notification unit 207a notifies the support field model of the answer "Proposed action 3 can be executed."

[0170] Then, each of the thought models of drone A and truck P executes the candidate action that has been confirmed for execution (S217). After execution, the status information indicating whether each candidate action can be executed is updated to "executed."

[0171] (Request for cooperation using the decision support function of the support field model) Fig. 11 is a diagram showing the virtual thinking process of the thinking model including a request for cooperation to the support field model. Fig. 11 corresponds to Fig. 9, and the same processes are assigned the same reference numerals and the description thereof will be omitted.

[0172] As described above, the support field model receives a "request" in the support area F1, generates an action plan for the input "request," and provides it to the participating thinking models. Meanwhile, the thinking model can select and execute a mutual cooperation action to request cooperation from others in the virtual thinking process. When selecting a mutual cooperation action, the thinking model uses the mutual cooperation control function 208 to find other thinking models and obtain their cooperation. However, a mechanism for obtaining cooperation from other thinking models can also be realized by utilizing the decision-making support function of the support field model.

[0173] Specifically, the mutual cooperation control function 208 of the thinking model (virtual intention module 200) can have a second function. When the action selection function 203 selects a mutual cooperation candidate action that requires the cooperation of another thinking model (S2041) in order to find a cooperation partner in the decision-making support function provided by the support field model, the second function outputs a cooperation request including the cooperation details to the support field model (S2042).

[0174] In the support processing shown in Figure 10, the support field model (support field module 300) is such that the "request" input into the support area F1 in step S302 becomes a request for cooperation in the thinking model, and the other processing from step S301 to step S311 is the same as described above.

[0175] The request receiving unit 321 of the support field model (support field module) receives a cooperation request from a thought model as a request input to the support area F1 (S302 in FIG. 10). Then, the action plan generating unit 322 generates an action plan consisting of actions to be executed by other thought models corresponding to the cooperation content based on the self-situation information of other thought models other than the thought model that requested cooperation (S303 in FIG. 10). After that, an evaluation process (S304 in FIG. 10) and selection of an action plan based on the evaluation result (S305 in FIG. 10) are performed, and the action plan is provided to the other thought model that is the cooperation partner. The other thought model that is the cooperation partner performs steps S210 to S217 in FIG. 10 and notifies the support field model of a response regarding whether or not to execute the action plan presented in step S216.

[0176] If the response from the other thought model is "possible to execute (OK response)", the support field model sends a notification to the thought model that made the cooperation request, that it can execute the action in response to the cooperation request, in step S310. This notification includes information about the other thought model that is the cooperation partner.

[0177] 11, the thinking model receives a notification regarding whether or not the action in response to the cooperation request can be executed (S2043). If the notification indicates that the action in response to the cooperation request can be executed (YES in S2044), the thinking model updates the status of the mutual cooperation type candidate action to execution confirmed. The thinking model executes the mutual cooperation type candidate action (S205).

[0178] 11, i.e., before updating the status of the mutual cooperation-type candidate action to execution confirmation, a mutual cooperation process may be performed with another thought model as a cooperation partner. The mutual cooperation process is a process performed by the first function of the mutual cooperation control function 208 described above, but the process of evaluating the cooperation content presented by the other thought model based on the autonomous thinking characteristics can be omitted, and a process of forming an agreement with the cooperation partner can be performed. Then, if an agreement is formed with the other thought model as a cooperation partner through the mutual cooperation control function 208, the selected candidate action is determined as the action to be executed. If an agreement is not formed, the process transitions to the process when a notification is received indicating that the action in response to the cooperation request can be executed, but an agreement was not reached with the cooperation partner, and therefore a notification indicating that the action in response to the cooperation request cannot be executed (NO in S2044).

[0179] If a notification of whether or not to execute the action in response to the cooperation request is received in step S2043, the thinking model selects other candidate actions other than the corresponding mutual cooperation candidate action based on the evaluation result (S204). The subsequent processing is as described above.

[0180] In the example of Figure 11, the thinking model itself does not need to find other thinking models, that is, the support field model understands the situation of each participating thinking model, so it can generate many appropriate action plans and propose them to the thinking model that has requested mutual cooperation. Therefore, consensus building between thinking models can be efficiently achieved.

[0181] Second Embodiment FIGS. 12 and 13 are diagrams for explaining a second embodiment.

[0182] 12 is a functional block diagram of a server-physical system in which the support field management unit 123 has autonomous thinking control. As shown in FIG. 12, the support field management unit 123 includes an autonomous thinking control unit 123A, which sets the autonomous thinking characteristics of the support field model based on predetermined thinking factors. The autonomous thinking control unit 123A has the same configuration as the autonomous thinking control unit 122A of the virtual intention management unit 122.

[0183] The cyberspace execution unit 124 operates the support field model modeled in cyberspace and executes various processes of the decision support function described above based on the autonomous thinking characteristics. The support field model operates as a virtual will model that has a virtual will in cyberspace and behaves autonomously.

[0184] The action plan evaluation unit 323 can perform the above-mentioned action plan evaluation process based on the autonomous thinking characteristics of the support field model.

[0185] In the first embodiment, an example of evaluating action plans generated using predetermined evaluation criteria has been described. The evaluation criteria may be, for example, distance, time, or cost, and differences in physical travel distance, time, and cost are taken into consideration. In this case, each action plan can be evaluated using autonomous thinking characteristics based on thinking factors.

[0186] For example, if an action plan with a small time difference is highly rated even if it is costly, the support field model has autonomous thinking characteristics based on a thinking factor (evaluation standard) that emphasizes the importance of "beauty / excellent work." Also, if an action plan with a small cost difference is highly rated even if it takes time, the support field model has autonomous thinking characteristics based on a thinking factor (evaluation standard) that emphasizes the importance of "health / good condition."

[0187] In this way, the decision-making support function of the support field model can evaluate and judge according to different autonomous thinking characteristics for each corresponding support area and provide action plans to the participating thinking models. Therefore, even for the same request, different action plans can be proposed depending on the autonomous thinking characteristics of the participating support field models, and requests to be input to the support area can be selected according to the characteristics of the autonomous thinking characteristics of the support field models.

[0188] 13 is a diagram showing the processing flow of the cyber-physical system of this embodiment, and corresponds to Fig. 4 of the first embodiment. The same processes are denoted by the same reference numerals and the description thereof will be omitted.

[0189] As shown in FIG. 13 , a modeling target for the support area is input (S4). The support field management unit 123 generates a support field model corresponding to the modeling target for the input support area (S103). At this time, the autonomous thinking control unit 123A controls one or more different thinking factors constituting the autonomous thinking characteristics so that they can be set, and sets the thinking factors for the generated support field model (S4a). The autonomous thinking control unit 123A determines the autonomous thinking characteristics of the generated support field model based on the set thinking factors (S103a). The determined autonomous thinking characteristics are applied to the corresponding support field model.

[0190] The above describes the embodiments. In the decision-making support function of the first and second embodiments, a "request (a request from a third party or a request from a thought model)" is input into the support area, and action proposals are generated and presented in response to the input of the "request."

[0191] Meanwhile, the support field model collects the self-situation information of each participating thinking model using the situation assessment function 310. Therefore, even if a "request" is not input to the support area, the support field model can generate an action plan based on the collected self-situation information of each thinking model and provide it to the corresponding thinking model.

[0192] For example, if the support field model determines, based on the self-situation information of a participating thought model, that the conditions necessary for the thought model to execute its planned action may not be met, the support field model can generate a proposed action to fulfill the conditions based on the situations of the other participating thought models and reach a consensus with the other thought models. Specifically, for example, suppose the thought models participating in support area F1 are drones A, B, and trucks P, Q, and R. Among the action candidates selected by drone A, "arriving at point P by a certain time" has a high priority. However, the situation of drone A as understood by the support field model indicates that the battery capacity may be insufficient to execute that action. To enable the execution of the high-priority action, even though drone A has not requested it, the support field model can generate a proposed action to "transport drone A to point P" for truck P, which is near drone A and has ample fuel, based on the situations of the other participating thought models. This action can then be presented to drone A and truck P, leading to a consensus.

[0193] In this way, the cyber-physical system can generate and provide action plans that match the self-situation information of each thinking model on the support field model side, along with the virtual thinking processing of the thinking model. In other words, the support field model does not include a request receiving unit 321, and the following cyber-physical system is configured.

[0194] That is, a cyber-physical system is configured that includes a virtual intention management unit 122 that models, in cyberspace, thought models corresponding to "things" in the real world and sets the autonomous thinking characteristics of the thought models based on predetermined thought factors, and a support field management unit 123 that models, in cyberspace, a support field model that forms a support area in which multiple thought models can participate, wherein the support field model has a situation understanding unit 310 that collects self-situation information of each participating thought model, an action plan generation unit 322 that generates action plans based on the self-situation information, and an action proposing unit 324 that provides the action plans to the corresponding thought model, and wherein the thought model has an action plan receiving unit 206 that receives action plans from the support field model and adds them to an action candidate list 205, an action evaluation unit 202 that evaluates candidate actions based on the autonomous thinking characteristics and outputs the evaluation results, and an action selection unit 203 that selects an action to be executed based on the evaluation results.

[0195] In addition, each function that constitutes a cyber-physical system can be realized by a program, and computer programs prepared in advance to realize each function are stored in an auxiliary memory device.A control unit such as a CPU reads the program stored in the auxiliary memory device into a main memory device, and the control unit executes the program read into the main memory device, thereby operating the functions of each unit.

[0196] The program can also be provided to a computer in a state recorded on a computer-readable recording medium. Examples of computer-readable recording media include optical discs such as CD-ROMs and Blu-ray® Disc Rewritables, phase-change optical discs such as DVD-ROMs, magneto-optical discs such as MO (Magneto Optical), magnetic discs such as floppy disks and hard disks, and memory cards such as SD memory cards and USB flash drives. Also included as recording media are hardware devices such as integrated circuits (e.g., IC chips such as ROMs and RAMs) specially designed and configured for the purposes of the present invention. Furthermore, the present invention, including the program, is not limited to being executed on a von Neumann computer architecture, but may also be executed on so-called non-von Neumann computer architectures, such as neurocomputers based on the mechanisms of neural circuits in the brain and quantum computers that apply quantum mechanics to information processing.

[0197] Although the embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. This novel embodiment can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, and are also included in the inventions and their equivalents as set forth in the claims.

[0198] 100 Cyber-physical system 110 Communication device 120 Control device 121 Physical data interface unit 122 Virtual intention management unit 122A Autonomous thinking control unit 123 Support field management unit 123A Autonomous thinking control unit 124 Cyberspace execution unit 125 Cyber ​​data interface unit 130 Storage device 200 Virtual intention module 201 Self-situation understanding function 202 Action evaluation function 203 Action selection function 204 Action execution function 205 Action candidate list 206 Action generation function 207 Action plan reception function 207a Notification unit 208 Mutual cooperation control function 300 Support field module 310 Situation understanding function 320 Decision-making support function 321 Request reception unit 322 Action plan generation unit 323 Action plan evaluation unit 324 Action proposal control unit

Claims

1. A cyber-physical system comprising: a virtual intention management unit that models a thought model corresponding to one in the real world in cyberspace and sets the autonomous thinking characteristics of the thought model based on predetermined thought factors; and a support field management unit that models a support field model in cyberspace that forms a support area in which a plurality of thought models participate, wherein the support field model has: a situation understanding unit that collects situation information of each participating thought model; an action plan generation unit that generates action plans consisting of actions to be executed by the thought model based on the situation information; and an action proposal unit that provides the action plans to the corresponding thought model, wherein the thought model has: an action plan receiving unit that receives action plans from the support field model and adds them to a candidate action list; an action evaluation unit that performs an evaluation process for candidate actions based on the autonomous thinking characteristics and outputs the evaluation results; and an action selection unit that selects an action to be executed based on the evaluation results.

2. The cyber-physical system described in claim 1, characterized in that the virtual intention management unit models a thought model corresponding to a real-world service in cyberspace and sets the autonomous thinking characteristics of the thought model based on predetermined thought factors.

3. The cyber-physical system described in claim 1, characterized in that the virtual intention management unit models in cyberspace a thought model corresponding to a person or organization that provides a service in the real world, and sets the autonomous thinking characteristics of the thought model based on predetermined thought factors.

4. The cyber-physical system described in claim 1, characterized in that the virtual intention management unit models a thought model in cyberspace that is not linked to anything in the real world, and sets the autonomous thinking characteristics of the thought model based on specified thought factors.

5. A cyber-physical system as described in any one of claims 1 to 4, characterized in that it comprises a request receiving unit that receives requests input into the support area, and the action proposal generation unit generates action proposals for requests received by the request receiving unit.

6. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the action plan generation unit generates a plurality of different action plans consisting of actions to be executed by one or more thinking models, the support field model has an action plan evaluation unit that performs an action plan evaluation process to evaluate the generated action plans based on predetermined evaluation criteria and extracts action plans to propose to the thinking model based on the action plan evaluation results, and the action proposal unit provides the action plans extracted based on the action plan evaluation results to the corresponding thinking model.

7. The cyber-physical system described in claim 6, characterized in that the support field management unit sets the autonomous thinking characteristics of the support field model based on the specified thinking factors, and the action plan evaluation unit performs the action plan evaluation process based on the autonomous thinking characteristics of the support field model.

8. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the situation understanding unit collects situation information of each participating thought model and physical data regarding the environment of real-world objects corresponding to the thought model, and the action proposal generation unit generates action proposals based on the situation information and the physical data.

9. The cyber-physical system described in claim 8, characterized in that the situation information of each thought model includes an action execution history of the thought model, and the action plan generation unit generates the action plan based on the action execution history.

10. The cyber-physical system described in any one of claims 1 to 4, characterized in that the thought model is provided with a notification unit that notifies the support field model whether or not the proposed action can be executed based on the action selection result by the action selection unit, the action proposal unit receives a notification of whether or not the proposed action can be executed from the thought model, determines whether or not other thought models other than the thought model that notified the execution or non-execution are involved in the proposed action, and notifies the other involved thought models that the proposed action is invalid, and the action proposal generation unit, when receiving a notification of whether or not the proposed action can be executed from the thought model, performs a second process to regenerate the action proposal.

11. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the situation information includes unique information of each participating thought model, self-analysis results generated by situation understanding processing based on physical data, and action execution plan information.

12. A cyber-physical system as described in any one of claims 1 to 4, characterized in that it comprises a physical data interface unit that provides collected physical data to cyberspace, the thinking model has an action generation unit that generates candidate actions based on the results of situation understanding processing based on the physical data and adds the generated candidate actions to an action candidate list, the action evaluation unit performs the evaluation processing of candidate actions included in the action candidate list based on the autonomous thinking characteristics, and the action selection unit selects an action to be executed autonomously based on the evaluation result.

13. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the support field model has a request receiving unit that receives a request input into the support area, the thinking model has a mutual cooperation control unit that outputs a cooperation request including cooperation details to the support field model when the selected action is a mutual cooperation action that requires cooperation from other thinking models, the request receiving unit receives the cooperation request as a request input into the support area, and the action plan generation unit generates an action plan consisting of actions to be executed by other thinking models corresponding to the cooperation details based on situation information of other thinking models other than the thinking model that made the cooperation request.

14. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the virtual intention management unit has an autonomous thinking control unit that determines the autonomous thinking characteristics of the thinking model based on one thinking factor or a combination of multiple different thinking factors, and the autonomous thinking control unit controls one or multiple different thinking factors in a specifiable manner and controls the autonomous thinking characteristics of each of the thinking models in a determinable manner.

15. The cyber-physical system described in claim 7, characterized in that the support field management unit has an autonomous thinking control unit that determines the autonomous thinking characteristics of the support field model based on one thinking factor or a combination of multiple different thinking factors, and the autonomous thinking control unit controls one or multiple different thinking factors in a specifiable manner and controls the autonomous thinking characteristics of each of the thinking models in a determinable manner.

16. A cyber-physical system as claimed in any one of claims 1 to 4, characterized in that the thought factors include a first thought factor governing survival and longevity, a second thought factor governing living and lifestyle, a third thought factor governing health and good condition, a fourth thought factor governing beauty and the best work, and / or a fifth thought factor governing usefulness.

17. A cyber-physical system as described in any one of claims 1 to 4, characterized in that the counterpart of the thought model is at least one of equipment, machinery, devices, products or moving objects operating in the real world.

18. A method for executing a cyber-physical system in which thought models corresponding to those in the real world are modeled in cyberspace, the autonomous thinking characteristics of the thought models are set based on predetermined thinking factors, and a support field model that forms a support area in which multiple thought models participate is modeled in cyberspace, wherein the support field model performs the following steps: collecting situation information of each participating thought model; generating action plans consisting of actions to be executed by the thought models based on the situation information; and providing the action plans to the corresponding thought model; and the thought model performs the following steps: accepting the action plans from the support field model and adding them to a candidate action list; evaluating the candidate actions based on the autonomous thinking characteristics and outputting the evaluation results; and selecting an action to be executed based on the evaluation results.

Images