Measure formulation method and measure formulation system

Abstract

To provide a measure formulation method and a measure formulation system that can formulate more diverse measures quickly and efficiently by more accurately grasping needs of a local resident in a public space.SOLUTION: A measure formulation method executed by a measure formulation system that uses a virtual space that simulates a real space to evaluate a measure related to the use of the real space, generates, on the basis of user input, using a generative AI, a measure goal and constraint. Then, the method generates, on the basis of the goal and the constraint, using a generative AI, a simulation task for executing a human-flow simulation that simulates a flow of a human for the measure. The method then executes a simulation task on a virtual space to perform the human-flow simulation, and displays the result of the human-flow simulation on a display screen.SELECTED DRAWING: Figure 14

Description

[Technical Field]

[0001] The present invention relates to a policy formulation method and a policy formulation system. [Background technology]

[0002] If vacant spaces in public spaces such as inside train stations are not utilized efficiently, it will result in lost opportunities for businesses that provide or use the public spaces. To utilize public spaces efficiently, it is important to understand the needs of users of the public spaces and to formulate and verify measures that meet those needs.

[0003] For example, Patent Document 1 discloses a technology for determining which stores to attract when formulating measures to attract businesses to create a city that maximizes the satisfaction of visitors and other mobile entities, by evaluating the effectiveness of attracting businesses based on people flow patterns and similarity scores between industries. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-250948 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the above-mentioned conventional technology evaluates the effectiveness of attracting businesses based on people's route information and business types, and does not provide a system that understands the needs of local residents.Furthermore, it only determines the business types of businesses to attract.

[0006] The present invention was made in consideration of the above circumstances, and aims to formulate a wider variety of policies quickly and efficiently by more accurately understanding the needs of local residents in public spaces. [Means for solving the problem]

[0007] In one aspect of the present invention, there is provided a policy formulation method executed by a policy formulation system that uses a virtual space that mimics a real space to evaluate policies regarding the use of space in the real space, the policy formulation method having a data management unit that manages local resident information including the needs of users, including local residents of the real space, for the space, and space information regarding the attributes of the space, and is characterized by having the following steps: a goal / constraint input step that uses a generation AI (artificial intelligence) to generate goals and constraints for the policy based on input from a user; a task generation step that uses the generation AI to generate simulation tasks for executing a people flow simulation that simulates the flow of people based on the goals and constraints generated by the goal / constraint input step, the local resident information, and the space information; and a simulation execution / evaluation step that executes the simulation task generated by the task generation step in the virtual space to execute the people flow simulation, and displays the results of the people flow simulation on a display screen. [Effects of the Invention]

[0008] According to the present invention, it is possible to more accurately grasp the needs of local residents in public spaces and then quickly and efficiently formulate a wider variety of policies. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is an explanatory diagram of an outline of an embodiment. [Figure 2] FIG. 1 is a diagram showing the configuration of a policy formulation system according to an embodiment. [Figure 3] FIG. 2 is a diagram showing the configuration of three-dimensional space information according to the embodiment. [Figure 4] FIG. 2 is a diagram showing the configuration of local resident information according to the embodiment. [Figure 5] FIG. 2 is a diagram showing the configuration of space information according to the embodiment. [Figure 6] FIG. 2 is a diagram showing the configuration of goal / constraint information according to the embodiment. [Figure 7] FIG. 10 is a diagram showing a list of simulation items and KPIs that can be verified by the simulator according to the embodiment. [Figure 8] FIG. 10 is a diagram showing a list of constraints that can be set in the simulator according to the embodiment. [Figure 9] FIG. 4 is a diagram showing policy proposal data according to the embodiment. [Figure 10] FIG. 10 is a diagram showing a learning model management table according to the embodiment. [Figure 11] 10 is a flowchart showing a data management process according to the embodiment. [Figure 12] 10 is a flowchart showing a goal and constraint input process according to the embodiment. [Figure 13] 10 is a flowchart showing a task generation process according to the embodiment. [Figure 14] 4 is a flowchart showing a simulation execution and evaluation process according to the embodiment. [Figure 15] FIG. 10 is a diagram showing a free space management screen according to the embodiment. [Figure 16] FIG. 10 is a diagram showing a policy proposal generation screen according to the embodiment. [Figure 17] FIG. 10 is a diagram showing a generated AI operation log screen according to the embodiment. [Figure 18] FIG. 10 is a diagram showing a screen of a proposed policy and an evaluation result according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the following description and drawings are examples for explaining the present invention, and appropriate omissions and simplifications have been made for clarity of explanation. Furthermore, not all combinations of features described in the embodiments are necessarily essential to the solution of the invention. The present invention is not limited to the embodiments, and all application examples consistent with the concept of the present invention are included within the technical scope of the present invention. Those skilled in the art can make various additions and modifications to the present invention within the scope of the present invention. The present invention can also be implemented in various other forms. Unless otherwise specified, each component may be plural or singular.

[0011] In the following description, a "CPU (Central Processing Unit)" is an example of one or more processor devices. The at least one processor device is typically not limited to a CPU, but may be another type of processor device such as a GPU (Graphics Processing Unit). The at least one processor device may be a single-core or multi-core. The at least one processor device may be a processor core.

[0012] At least one processor device may be a circuit that is a collection of gate arrays written in a hardware description language that performs some or all of the processing. The circuit is a processor device in the broad sense, such as a field-programmable gate array (FPGA), a complex programmable logic device (CPLD), or an application-specific integrated circuit (ASIC).

[0013] In the following description, a program is executed by a CPU to realize a processing function called "XXX unit" and become the executing entity of the process. The processing function may be realized by one or more computer programs being executed by a processor, or may be realized by one or more hardware circuits (e.g., FPGA or ASIC), or may be realized by a combination of these.

[0014] When a function is realized by executing a program by a processor, the specified processing is performed using a storage device and / or an interface device, etc., so the function may be considered to be at least a part of the processor. Processing described using a functional unit as the subject may also be processing performed by a processor or a device having that processor.

[0015] The program may be installed from a program source. The program source may be, for example, a program distribution computer or a computer-readable recording medium (e.g., a non-transitory recording medium). The description of each function is an example, and multiple functions may be combined into one function, or one function may be divided into multiple functions.

[0016] In the following explanation, various information may be explained in table format. A "YYY table" may be called "YYY information." The data format of the information may be a format other than a table format (for example, CSV (Comma Separated Values) format). Furthermore, various information may be stored in a memory unit as a table, or may be embedded as logic in a program.

[0017] In addition, in the following description, when describing elements of the same type without distinguishing between them, common reference symbols will be used, and when describing elements of the same type with distinction between them, reference symbols will be used.

[0018] In the following embodiment, the real space for which measures are to be formulated is the public space of a station (the so-called "inside the station"), and the virtual space that simulates the real space is the digital twin of the inside of the station. Using artificial intelligence (AI), we automatically generate measures to effectively utilize the vacant space inside the station, and evaluate the effectiveness of the measures by simulating people flow in the virtual space.

[0019] (Outline of the embodiment) 1 is an explanatory diagram of an overview of an embodiment. In this embodiment, a space management 2A mechanism capable of managing space information 2 (spaces such as advertising spaces, signage display spaces, tenant store spaces, gallery exhibition spaces, etc.) of a real space 1 is constructed in a virtual space 1A (a digital twin space inside a station).

[0020] First, in step S1, the generation AI 4 (AI agent) sequentially learns the needs of the space information 2 (each space). The sequential learning of the needs of the space information 2 continues in parallel with subsequent steps each time the space information 2 is updated. Next, in step S2, the generation AI 4 learns the simulation specifications (simulation procedure, simulation tasks, method of interpreting simulation results, etc.) of the simulator 15 in the virtual space 1A, and a learning model M is generated.

[0021] Specifically, in steps S1 and S2, large-scale language models (LLMs) are fine-tuned by utilizing LangChain (registered trademark) or the Lora model, etc., to learn what can be executed by simulation, the execution procedures, etc., and generate a learning model M so that the LLM can handle the simulator 15. The generation AI 4 detects updates to the simulator 15, and if there is an update, also updates the learning model M.

[0022] The learning model M has learned how to use the multiple simulators 15, and when generating a simulation task (described later), the simulation task is generated based on the usage of one or more simulators selected from the multiple simulators 15 according to the goal and constraints. The generation AI 4 also generates measures and simulation tasks by referring to external information such as the Internet.

[0023] Next, in step S3, the generation AI 4 accepts input of objectives and constraints from the policy verifier 5 (user). Next, in step S4, the generation AI 4 generates a policy proposal and a simulation task for this policy proposal based on the objectives and constraints input in step S3, based on the learning model M generated in steps S1 and S2. Next, in step S5, the generation AI 4 transmits the simulation task generated in step S4 to the virtual space 1A.

[0024] Next, in step S6, the simulator 15 executes a simulation task of quantitatively evaluating the proposed measures in the virtual space 1A using the space managed by the space management 2A and models such as a human behavior model, a traffic flow model, a facility model, and an evaluation model.

[0025] Next, in step S7, the generation AI 4 obtains the results of the evaluation simulation of the measures in the virtual space 1 A (predicted values of the measures' effects). Next, in step S8, the generation AI 4 outputs the measures formulated in step S4 and the predicted values of the measures' effects to the measure verifier 5.

[0026] (Configuration of policy formulation system S according to the embodiment) 2 is a diagram showing the configuration of a policy formulation system S according to an embodiment. The policy formulation system S is connected to an external system 200 and an external server 300 via a network N so as to be able to communicate with each other.

[0027] The external system 200 is a system that manages the inside of each station, acquires various data related to the inside of each station, generates and updates three-dimensional space information 171, local resident information 172, and space information 173, and transmits them to the policy formulation system S.

[0028] The external server 300 is a server on which a generation AI4 (AI agent) runs. The generation AI4 has the function of generating and outputting measures related to the use of in-station facilities in response to an inquiry about measures related to the use of in-station facilities written in natural language input by a user from an external system such as the policy formulation system S. Well-known technology is used for the generation AI4. The generation AI4 is not limited to being provided by the external server 300, but may also be provided by the policy formulation system S.

[0029] The policy formulation system S is configured to include a CPU 11, an input device 12 such as a keyboard, an output device 13 such as a display, a communication device 14 which is a network interface, a simulator 15, a main storage device 16 such as a memory, and an auxiliary storage device 17 such as a storage. The simulator 15 may be provided outside the policy formulation system S and may be linked to the policy formulation system S via a network N.

[0030] The main memory device 16 has a data management unit 161, a goal and constraint input unit 162, a task generation unit 163, and a simulation execution and evaluation unit 164, which are realized by executing a predetermined program. The auxiliary memory device 17 stores three-dimensional space information 171, local resident information 172, space information 173, goal and constraint information 174, policy proposal data 175, a learning model management table 176, a simulation procedure 177, and simulation results 178.

[0031] The data management unit 161 executes a data management process for accepting and updating new data of three-dimensional space information 171, local resident information 172, and space information 173, which will be described later. Details of the data management process will be described later with reference to FIG.

[0032] The goal / constraint input unit 162 executes a goal / constraint input process for setting goals and constraints (goal / constraint information 174) or generating a policy plan (policy plan data 175) in response to a user input. Details of the goal / constraint input process will be described later with reference to FIG. 12.

[0033] The task generation unit 163 executes a task generation process for creating a simulation procedure 177 for evaluating a policy proposal using a generation AI based on the goal and constraints set by the goal and constraint input unit 162. Details of the task generation process will be described later with reference to FIG.

[0034] The simulation execution and evaluation unit 164 executes the simulator 15 based on the three-dimensional space information 171, local resident information 172, space information 173, and policy proposal data 175, and obtains simulation results 178 including an evaluation of the policy proposal.

[0035] (Three-dimensional space information 171 according to the embodiment) 3 is a diagram showing the configuration of three-dimensional space information 171 according to the embodiment. The three-dimensional space information 171 includes spatial measurement data such as point cloud data 1711 and CAD (Computer Aided Design) data 1712. The three-dimensional space information 171 may include mesh data as the spatial measurement data.

[0036] The point cloud data 1711 is a point cloud that represents each station interior, including objects within the space, for which measures are being formulated, and is data in which color data is associated with each coordinate in the three-dimensional space of the station interior. The CAD data 1712 manages the names of objects that exist within the space of each station for which measures are being formulated, the type of CAD data, and the CAD file name in association with each other. The CAD data 1712 manages CAD files that are stored in a specified memory area and represent the space of each station for which measures are being formulated and the structure of objects that exist within this space.

[0037] (Local resident information 172 according to the embodiment) 4 is a diagram showing the configuration of local resident information 172 according to the embodiment. The local resident information 172 includes POS (Point Of Sale) data 1721, station data 1722, travel section data 1723, and questionnaire results 1724.

[0038] The POS data 1721 includes the area to which each station for which a policy is being formulated belongs, the nearest station to each area, sales format, type of product sold, number of sales per day, number of purchasers by age group, and number of purchasers by gender. For example, the first line of the POS data 1721 indicates that the nearest station to area X is A, the number of bento boxes sold per day at the station sales space of station A is 500, the number of teenage purchasers is 50, ..., and the number of male purchasers is 300.

[0039] The station data 1722 includes the area to which each station for which a policy is being formulated belongs, the nearest station in each area, the number of visitors per day, the number of visitors by type of use, the number of visitors by age group, and the number of visitors by gender. For example, the first line of the station data 1722 indicates that the nearest station to area X is A, the number of visitors per day to station A is 1,000, the number of commuter visitors (commuter commuter pass) is 200, the number of teenagers visitors is 50, ..., and the number of male visitors is 500.

[0040] The boarding section data 1723 includes the number of entrants for each entry time period for each boarding section of the entry station and exit station for each passenger, the number of entrants by usage type, the number of entrants by age group, and the number of entrants by gender. For example, the first row of the boarding section data 1723 indicates that the number of entrants in the 5 o'clock hour for the boarding section from entry station A to exit station B is 10, the number of commuter entrants (commuter commuter pass) is 5, the number of teenagers entrants is 1, ..., and the number of male entrants is 5.

[0041] The survey result 1724 includes the area to which each station within which the policy is being formulated belongs, the nearest station in each area, the question, the answer to the question, the respondent's attributes, and the respondent's gender. For example, the first line of the survey result 1724 indicates that the nearest station to area X is A, the question is "What kind of store would you like to see around station A?", the answer is "I want a delicious bakery," the respondent's attributes are "workers (around station A)," and the respondent's gender is female.

[0042] (Space information 173 according to the embodiment) 5 is a diagram showing the configuration of space information 173 according to an embodiment. The space information 173 includes information on the location of each space identified by a space ID, the type of space, and the area indicated in the local resident information 172 in which the space is located. The space information 173 also includes information on coordinates indicating the location in the real space 1, coordinates indicating the location in the virtual space 1A, the size of the space, the price of the space, and the usage status of the space. In other words, the space includes advertising space for displaying advertisements and tenant spaces, which are rooms or sections for locating tenants.

[0043] For example, the first line of space information 173 indicates that a space with a space ID of 1 is a merchandise space and is located in area X. It also indicates that the space with a space ID of 1 has real-world coordinates (x1, x2, x3), virtual-world coordinates (x5, x5, x5), a space size of 5 x 3 x 3 m, a price of 50,000 yen / day, and a usage status of "vacant."

[0044] (Goal and constraint information 174 according to the embodiment) FIG. 6 is a diagram showing the configuration of the goal and constraint information 174 according to the embodiment. The goal and constraint information 174 includes a goal and constraint ID, a user request (goal), a simulation item, a KPI (Key Performance Indicator), and constraints. The goal and constraint ID is information for identifying the goal and constraint information of each record. The user request (goal) is a description in natural language of the content (goals, conditions, etc.) that the policy verifier 5 (user) requests of each station for which a policy is to be formulated. The simulation item includes information for specifying the type of simulation to the simulator 15 to be executed when a simulation task is executed. The KPI is an index for evaluating the formulated policy. The constraint is a constraint condition when calculating the KPI.

[0045] 7 is a diagram showing a list 174a of simulation items and KPIs that can be verified by the simulator 15 according to the embodiment. The list 174a shows candidate simulation items and corresponding KPIs in the goal and constraint information 174 shown in FIG. 6. For example, the first line of the list 174a indicates that when "measurement of advertising effectiveness" is selected as the simulation item, "GRP (Gross Rating Point)," "OTS (Opportunity to See Base)," and "OOH (Out Of Home)" are adopted as KPIs.

[0046] Fig. 8 is a diagram showing a list 174b of constraints that can be set in the simulator 15 according to the embodiment. The list 174b shows candidate constraints of the goal and constraint information 174 shown in Fig. 6. For example, the first line of the list 174b indicates that "specify a time range" can be set as a constraint.

[0047] The above-mentioned "List of items and KPIs that can be verified by the simulator" and "Constraints that can be set by the simulator" are inputs to the simulator 15. For example, a change in this input / output corresponds to an update of the simulator 15, and therefore triggers an update of the learning model M of the generation AI 4.

[0048] (Measure proposal data 175 according to the embodiment) 9 is a diagram showing the measure proposal data 175 according to the embodiment. The measure proposal data 175 manages the measure proposal generated by the generation AI 4 and a simulation execution flag indicating whether or not to execute a simulation by the simulator 15. The measure proposal data 175 has columns for the measure proposal ID, the goal constraint ID, the space ID, the type, the location, the measure proposal, the measure proposal details, and the simulation execution flag.

[0049] The policy proposal ID is information that identifies the policy proposal for each record. The goal constraint ID is identification information in the goal and constraint information 174 of the goals and constraints that are the premise of the policy proposal for the corresponding record. The space ID is identification information in the space information 173 of the space that is the target of the policy proposal for the corresponding record. The type and location are the same as the type and location in the space information 173 of the space that is the target of the policy proposal for the corresponding record. The policy proposal is a policy generated by the generation AI 4 for the combination of the goal constraint ID and space ID of the corresponding record. The policy proposal details are detailed information of the policy proposal for the corresponding record. The policy proposal (policy) may include any of the method of using the space, the selling price of the space, and potential buyers of the space. The simulation execution flag is information that indicates whether a simulation of the policy proposal for the corresponding record will be executed by the simulator 15.

[0050] (Learning model management table 176 according to the embodiment) FIG. 10 is a diagram showing a learning model management table 176 according to an embodiment. The learning model management table 176 manages the learning model M of the simulation specifications of the simulator 15 learned in steps S1 and S2 (FIG. 1). The learning model management table 176 manages the learning model M stored in a predetermined storage area. For example, the first row of the learning model management table 176 indicates that the model name of the learning model M learned using "Lora" to learn "simulation procedure" is "model1.ggml". Furthermore, the second row of the learning model management table 176 indicates that the model name of the learning model M learned using "LangChain (registered trademark)" to learn "simulation procedure" is "model2.ggml".

[0051] (Data Management Process According to the Embodiment) 11 is a flowchart showing the data management process according to the embodiment. The data management process is executed every time new data is received from the external system 200.

[0052] First, in step S11, the data management unit 161 accepts new data for the space information 173, local resident information 172, and three-dimensional space information 171. Next, in step S12, the data management unit 161 updates the space information 173 with the new data. Next, in step S13, the data management unit 161 updates the local resident information 172 with the new data. Next, in step S14, the data management unit 161 updates the three-dimensional space information 171 with the new data.

[0053] Next, in step S15, the data management unit 161 develops the space information 173 and the three-dimensional space information 171 on the simulator 15 to construct the virtual space 1A (FIG. 1).

[0054] (Goal and constraint input process according to the embodiment) 12 is a flowchart showing a target and constraint input process according to the embodiment. The target and constraint input process is executed each time a target and constraint for effective space utilization or a request for policy proposal generation is input to the generation AI 4.

[0055] First, in step S21, the goal and constraint input unit 162 receives input from the policy verifier 5 (user). In step S21, the goal and constraint input unit 162 acquires information for narrowing down the policy formulation and simulation targets from the user input.

[0056] Next, in step S22, the goal / constraint input unit 162 determines whether the user input received in step S21 is a goal / constraint setting or a request for generating a policy plan. If the user input is a goal / constraint setting, the goal / constraint input unit 162 proceeds to step S23, and if the user input is a request for generating a policy plan, the goal / constraint input unit 162 proceeds to step S26.

[0057] In step S23, the goal / constraint input unit 162 creates a prompt for querying goal / constraint information. The goal / constraint input unit 162 uses a generation AI to find out what the user wants to do during a conversation such as a chat. At this time, if the goal / constraint information required to satisfy a specific termination condition is obtained, a prompt is set to end the conversation. For example, the unit may ask the user to fill in a table of goal / constraint information, and may request that the user end the conversation or issue an instruction to continue the conversation until the table is filled. One example of a termination condition is when all of the following are met: information on the target area, station, space, etc. is specified; the content and purpose of the measure to be implemented are specified; and constraints that can be set in the simulator are input.

[0058] Next, in step S24, the goal and constraint input unit 162 uses the prompt created in step S23 to have a dialogue with the user, and obtains goal and constraint information 174 through sentence understanding by the generation AI. More specifically, the goal and constraint input unit 162 extracts target space and area information from the text information entered by the user, and verifies that the data matches the space and area information registered by the data management unit 161. The goal and constraint input unit 162 also extracts the content of the measures the user wants to take from the text information entered by the user, and verifies that the content matches the list of items that can be verified by the simulator. The goal and constraint input unit 162 also extracts categories and numerical values related to constraints from the text information entered by the user, and verifies whether the categories and numerical values can be set in the simulator.

[0059] In step S24, the goal / constraint input unit 162 determines whether the user's input satisfies the amount of information necessary for generating the goal and constraints. If the required amount of information is not met, the goal / constraint input unit 162 may accept input of additional information from the user via a user prompt. Alternatively, if the required amount of information is not met, the goal / constraint input unit 162 may inquire of the generation AI 4 to obtain additional information that satisfies the required amount of information. Then, the goal / constraint input unit 162 may obtain goal / constraint information 174 using the generation AI based on the user's input and additional information.

[0060] Next, in step S25, the goal / constraint input unit 162 stores the goal / constraint information 174 obtained in step S25 in the auxiliary storage device 17.

[0061] On the other hand, in step S26, the goal / constraint input unit 162 acquires the goal / constraint information 174, the space information 173, and the local resident information 172 from the auxiliary storage device 17.

[0062] Next, in step S27, the goal / constraint input unit 162 passes the user input, interpretation information interpreting the user input, and information related to the interpretation information among the goal / constraint information 174, space information 173, and local resident information 172 acquired in step S26 to the generation AI 4. The generation AI 4 inputs information by providing prompts, and generates and outputs a policy proposal. Note that in this embodiment, the generation AI 4 generates a policy proposal in step S27, but this is not limiting, and the policy proposal may be manually generated and determined by the user.

[0063] For example, a prompt might be, "Think of an advertisement that you think would be effective at Station A. The users and needs of Station A are "XXX (enter information obtained in step S26)." For the majority of users, "(enter user attributes)," please search the Internet and tell us what they are currently most interested in. Based on the search results, please tell us the top three industries that would be most effective if advertised at Station A, as well as the type and content of the advertisement." Another prompt might be, "We are considering displaying real estate industry advertisements at Station A. The users and needs of Station A are "YYY (enter information obtained in step S26)." Considering user attributes and needs, please tell us what kind of real estate advertisements would attract interest."

[0064] The goal and constraint input unit 162 stores the measure proposal data 175 generated by the generation AI4 in step S27 in the auxiliary storage device 17.

[0065] Next, in step S28, the goal and constraint input unit 162 determines whether or not to execute a simulation based on the dialogue with the user and the proposed policy. The goal and constraint input unit 162 uses a generation AI to find out whether the user considers the proposed policy to be an adoption candidate during the dialogue with the user, and determines whether or not to execute a simulation of the proposed policy. The goal and constraint input unit 162 records a simulation execution flag in the proposed policy data 175 for the policy for which it has been determined that a simulation of the proposed policy should be executed.

[0066] (Task generation process according to the embodiment) 13 is a flowchart showing a task generation process according to the embodiment. In the task generation process, a generation AI 4 generates a simulation task based on the goal and constraint information 174, the measure proposal data 175, and the learning model management table 176. The simulation task generated here is a measure proposal determined to be simulated in step S28 of the goal and constraint input process (FIG. 12). The task generation process is executed at a predetermined interval or continuously after the goal and constraint input process is completed.

[0067] First, in step S31, the task generation unit 163 acquires the goal and constraint information 174 and the policy proposal data 175 from the auxiliary storage device 17. Next, in step S32, the task generation unit 163 uses the generation AI 4 having a learning model M that has learned how to use the simulator 15, etc., to generate a simulation task for the policy proposal for which the task is to be generated based on the policy proposal data 175. Specifically, the task generation unit 163 inputs a prompt such as "The constraints are AAA and BBB. Please output a simulation task for advertising effectiveness evaluation" into the large-scale language model to obtain a text-based simulation task. That is, the generation AI 4 has a learning model M that has learned how to use the simulator 15 that executes a people flow simulation, and the simulation task is generated based on the usage method of the simulator 15 that the learning model M has learned. When generating the simulation task, the task generation unit 163 uses the generation AI 4 to convert any natural language information into a data file to be used by the simulator 15.

[0068] Next, in step S33, the task generation unit 163 uses the generation AI to create a simulation execution procedure for each task. Specifically, the task generation unit 163 inputs a prompt, "Please also output the procedure for each task," into the large-scale language model, and obtains a text-based simulation procedure. The task generation unit 163 stores the simulation procedure 177 generated by the large-scale language model in the auxiliary storage device 17.

[0069] In step S33, for example, for tasks such as "Task: Place an α advertisement in space X within station A and simulate the advertising effect" and "Task: Place a β advertisement in space X within station A and simulate the advertising effect," the following procedure is generated: "Procedure: Set station A as the simulation target, generate pedestrian data within station A (including destinations, preferences, and behavior models), place the α advertisement (β advertisement) in space X, set a view counter, run a one-day simulation, and count the number of views."

[0070] (Simulation execution and evaluation process according to the embodiment) 14 is a flowchart showing a simulation execution and evaluation process according to an embodiment. In the simulation execution and evaluation process, a simulation task of a policy proposal is executed by the simulator 15 in the virtual space 1A, thereby evaluating the policy proposal. The simulation execution and evaluation process is executed at a predetermined cycle or continuously after the task generation process (FIG. 13) ends.

[0071] First, in step S 41 , the simulation execution and evaluation unit 164 acquires the simulation procedure 177 from the auxiliary storage device 17 .

[0072] Next, in step S42, the simulation execution and evaluation unit 164 generates a simulation setting file and constructs a simulation environment (virtual space 1A). Specifically, in step S42, the simulation execution and evaluation unit 164 generates input data for the simulator 15 as a simulation setting file generation procedure. For example, the procedure is "Procedure: Set station A premises as the simulation target, generate pedestrian data (including destinations, preferences, and behavior models) within station A premises, place advertisement A on Space X, set a view counter, run a one-day simulation, and count the number of views." The input data is, for example, in the form of a JSON file. The JSON file is created by the generation AI4 or by other methods. The simulation environment is set up based on this JSON file. It is assumed that the generation AI4 has learned the correct way to set up the simulation environment.

[0073] The JSON file that serves as input data to the simulator 15 is, for example, as follows: {”station”:”A”, “passenger”:2000, “OD”:”matrix A”, “behavior”:”model A”, “set_ad”:”area_X”,”view_count”:”area_X”, “time”:”04:00-23:00”, “KPI”:”GRP”}

[0074] Next, in step S43, the simulation execution / evaluation unit 164 executes a simulation in the simulation environment set up in step S42. The simulation is executed by operating the simulator 15 based on the simulation setting file generated in step S42. For example, based on the simulation setting file, 3D data is read from the three-dimensional space information 171, the target area is expanded in the simulation environment, and the simulator 15 is started. The simulation setting file can specify which of multiple models, such as a pedestrian model and a behavior model, included in the simulator 15 is to be used.

[0075] For example, in step S43, the people flow simulation may be performed using people flow measurement data including GPS (Global Positioning System) information of each station or railway user, history information of each user's IC (Integrated Circuit) ticket, and LiDAR (Light Detection And Ranging) measurement data measuring the distance to each user. Alternatively, the people flow simulation may be performed using a behavior model created based on this people flow measurement data.

[0076] Next, in step S44, the simulation execution and evaluation unit 164 evaluates the proposed measure based on the simulation executed in step S43. The simulator 15 can reproduce people's behavior in the virtual space 1A, so it can measure, for example, how many people will view an advertisement if one is placed there. It can also measure the level of interest based on a preference model. The preference model is the preference for each attribute that can be grasped from the survey results 1724 (FIG. 4), such as what kind of food people like and what their current interests are. The level of interest in the advertisement may be evaluated for each attribute of the survey respondents, such as employees in the area or tourists. The simulation execution and evaluation unit 164 may also evaluate changes in sales, congestion, customer satisfaction, etc.

[0077] The effectiveness of the measures can be evaluated based on the set KPIs. For example, GRP is the viewer rating, which indicates the percentage of people who saw the advertisement. It is calculated by detecting the gaze of people in virtual space 1A, counting 1 when their gaze is fixed on the advertisement, and calculating the percentage of viewers who saw the advertisement out of the total number of people.

[0078] Next, in step S45, the simulation execution and evaluation unit 164 outputs the evaluation result of the proposed measures by the simulation in step S44 from the simulator 15 in a text-based report format.

[0079] Next, in step S46, the simulation execution / evaluation unit 164 uses the generation AI4 to create interpretation information for the evaluation results of the proposed measures output in step S45. As described above, the generation AI4 has learned how to interpret the simulation results (evaluation results of the proposed measures), and can output interpretation information for the evaluation results of the proposed measures based on this learning result. Next, in step S47, the simulation execution / evaluation unit 164 outputs the proposed measures and the evaluation results to the user on a screen. The screen output may look, for example, like the proposed measures and evaluation result screen D4 (FIG. 18) described below.

[0080] (Free space management screen D1 according to the embodiment) 15 is a diagram showing a free space management screen D1 according to an embodiment. The free space management screen D1 is displayed on the display screen of the output device 13. The free space management screen D1 has, as GUI (Graphical User Interface) elements operated via the input device 12, an area selection menu D11, an area appearance display D12, a space list D13, a new policy proposal generation button D14, and a price change button D15.

[0081] The area selection menu D11 accepts the selection of an area from a drop-down list of area candidates according to a user operation. When an area is selected, the available space management screen D1 displays the corresponding area and the spaces within the area in the space information 173 in the area appearance display D12.

[0082] The space list D13 displays, for the spaces within the corresponding area displayed in the area appearance display D12, information based on the space information 173 in association with the proposed measures based on the proposed measure data 175. When the "Details" display in the proposed measure in the space list D13 is clicked, the corresponding "Proposed measure details" information is read from the proposed measure data 175 and displayed.

[0083] When a space record is selected in the space list D13 and the Create New Plan button D14 is pressed, a plan and general manager information for the space are generated. When a space record is selected in the space list D13 and the Change Price button D15 is pressed, the price of the space is edited.

[0084] (Measure proposal generation screen D2 according to the embodiment) 16 is a diagram showing a measure proposal generation screen D2 according to the embodiment. The measure proposal generation screen D2 is a chat screen that is displayed by screen transition on the display screen of the output device 13 when the new measure proposal generation button D14 is pressed on the free space management screen D1. When the measure proposal generation screen D2 is displayed, the goal and constraint input process (FIG. 12) is started.

[0085] The measure proposal generation screen D2 has, as GUI elements operated via the input device 12, a chat input area D21, a send button D22, a self-chat display area D23, a partner-side chat display area D24, and a prompt extraction button D25.

[0086] The policy verifier 5 (user) inputs an inquiry D231 related to a policy proposal for effective use of space into the chat input area D21 and presses the send button D22. The inquiry D231 input by the user is displayed in the self-chat display area D23 and is also sent to the generation AI4 of the external server 300. The generation AI4 sends a follow-up question D241 in response to the inquiry D231 from the user as necessary. The follow-up question D241 is displayed in the other-side chat display area D24.

[0087] A hyperlink related to one of the spaces is selected from the link display D2411 of the follow-up question D241, and the "View details in the metaverse space" button D2412 is pressed. Then, as shown in Figure 15, an overview of the corresponding space can be viewed in virtual space 1A (metaverse space).

[0088] The user inputs a response D232 to the follow-up question D241 in the chat input area D21 and presses the send button D22. The response D232 input by the user is displayed in the self-chat display area D23 and is also sent to the generation AI4. The generation AI4 further sends a follow-up question D242 in response to the user's response D232. The follow-up question D242 is displayed in the other-side chat display area D24.

[0089] The user inputs an answer D233 to the follow-up question D242 in the chat input area D21 and presses the send button D22. The answer D233 input by the user is displayed in the user's own chat display area D23 and is also sent to the generation AI4. The generation AI4 sends a notification D243 indicating that a policy proposal will be created in response to the user's answer D233. The notification D243 is displayed in the other party's chat display area D24. Finally, the generation AI4 creates and sends the policy proposal D244 and displays it in the other party's chat display area D24.

[0090] (Generated AI operation log screen D3 according to the embodiment) Figure 17 is a diagram showing the generated AI operation log screen D3 relating to the embodiment. The generated AI operation log screen D3 is a screen that displays the operation log of the internal prompts of the policy proposal generation screen D2. The generated AI operation log screen D3 is a screen that is displayed as a pop-up screen of the policy proposal generation screen D2 on the display screen of the output device 13 when the prompt extraction button D25 of the policy proposal generation screen D2 is pressed. The generated AI operation log screen D3 has an operation log display area D31 and a close button D32 as GUI elements that are operated via the input device 12.

[0091] The operation log display area D31 displays logs D311 and D312. Log D311 is an operation log of an internal prompt for receiving an inquiry D231 on the policy proposal generation screen D2. Log D312 is an operation log of a prompt called up by the internal prompt for receiving the inquiry D231. The user presses the close button D32 to finish checking the operation logs.

[0092] (Measure proposal and evaluation result screen D4 according to the embodiment) 18 is a diagram showing a proposed policy and evaluation result screen D4 according to an embodiment. The proposed policy and evaluation result screen D4 is displayed on the display screen of the output device 13 during execution of step S47 of the simulation execution and evaluation process (FIG. 14). The proposed policy and evaluation result screen D4 has, as GUI elements operated via the input device 12, a title display area D41, a proposed policy display area D42, and a basis and interpretation display area D43 for the proposed policy. The proposed policy and evaluation result screen D4 also has, as GUI elements operated via the input device 12, a sales forecast result display area D44, a resident information display area D45, and a simulation result display area D46.

[0093] The title display area D41 displays information about the area and space where the people flow simulation of the proposed measure is to be performed. The proposed measure display area D42 displays the proposed measure. The proposed measure's rationale and interpretation display area D43 includes a display of the rationale for the proposed measure that was generated together with the proposed measure by the generation AI4 in step S27 of the goal and constraint input process (Figure 12). The proposed measure's rationale and interpretation display area D43 also includes a display of the interpretation of the proposed measure that was generated by the generation AI4 in step S46 of the simulation execution and evaluation process (Figure 14).

[0094] The sales forecast result display area D44 displays the daily bread sales in the sales space 1 of area X predicted by running a people flow simulation of the proposed measure, broken down by attributes such as the type of bread, the age of the purchaser, and the gender of the purchaser.

[0095] The resident information display area D45 displays local resident information 172 (station data 1722, riding section data 1723, and survey results (needs)) (FIG. 4).

[0096] The simulation result display area D46 displays, as a video, the results of a people flow simulation of the proposed measure, which was performed by placing a person h along with the target area a (area X) and space s (merchandise space 1) in the virtual space 1A. The simulation result display area D46 includes a play button D461, a stop button 462, a time slider bar D463, and a video display D464. The user operates the play button D461, the stop button 462, and the time slider bar D463 to play the video at the desired time and check the results of the people flow simulation of the proposed measure.

[0097] (Effects of the embodiment) In the above-described embodiment, by using a generation AI and a simulation in a virtual space that mimics the real world, various measures can be generated and evaluated quickly and efficiently, based on a more accurate understanding of the attributes and needs of local residents with minimal user input.

[0098] Specifically, we will build a system that can manage real-world space information (advertising space, exhibition space, tenants, etc.) in a virtual space (for example, a digital twin space inside a station). We will then use generative AI to create effective measures and tasks for these spaces based on user input of objectives and constraints, and run simulations based on the automatically generated simulation tasks. This makes it possible to easily and quantitatively verify the effectiveness of measures.

[0099] Furthermore, even users who do not know how to run or analyze a simulator can run the simulator and analyze and evaluate measures.

[0100] Furthermore, by selecting and combining simulators that match the goals and constraints from multiple simulators and running them, it is expected that new simulation results that have never been seen before can be obtained.

[0101] Although the embodiments of the present disclosure have been described above in detail, the present disclosure is not limited to the above-described embodiments and can be modified in various ways without departing from the spirit of the present disclosure. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present disclosure is not necessarily limited to those having all of the described configurations. Furthermore, some of the configurations of the above-described embodiments can be added to, deleted from, or replaced with other configurations.

[0102] Furthermore, the above-described configurations, functional units, processing units, etc. may be partially or entirely implemented in hardware, for example, by designing them as integrated circuits. The above-described configurations, functions, etc. may also be implemented in software by a processor interpreting and executing a program that implements each function. Information such as the programs, tables, and files that implement each function can be stored in memory, storage devices such as HDDs and SSDs, or recording media such as IC cards, SD cards, and DVDs.

[0103] In addition, in the above-mentioned drawings, the control lines and information lines shown are those that are considered necessary for explanation, and do not necessarily show all the control lines and information lines that are actually implemented. For example, it may be considered that almost all components are actually connected to each other.

[0104] The above-described arrangement of the processing functions and data is merely an example, and the arrangement of the processing functions and data can be changed to an optimal arrangement in terms of the performance of the hardware and software, processing efficiency, communication efficiency, etc. [Explanation of symbols]

[0105] S: Policy formulation system, M: Learning model, 1: Real space, 1A: Virtual space, 11: CPU, 15: Simulator, 16: Main memory device, 161: Data management unit, 162: Goal and constraint input unit, 163: Task generation unit, 164: Simulation execution and evaluation unit, 171: 3D space information, 172: Local resident information, 173: Space information, 174: Constraint information, 175: Policy proposal data, 176: Learning model management table, 177: Simulation procedure, 178: Simulation results.

Claims

1. A policy formulation method executed by a policy formulation system that evaluates policies regarding space utilization in a real space using a virtual space that simulates the real space, comprising: The policy formulation system includes: a data management unit that manages local resident information including needs for the space of users including local residents of the real space, and space information regarding attributes of the space; a goal / constraint input step of generating goals and constraints of the measures based on input by a user using artificial intelligence (AI); a task generation step of generating, using the generation AI, a simulation task for executing a people flow simulation that simulates the flow of people based on the goals and constraints generated in the goal / constraint input step, the local resident information, and the space information, for the measures generated based on the goals and constraints; a simulation execution / evaluation step of executing the simulation task generated by the task generation step in the virtual space to execute the people flow simulation and displaying the execution result of the people flow simulation on a display screen; A policy formulation method comprising the steps of:

2. The policy formulation method according to claim 1, In the task generation step, The measures are generated using the generation AI based on the goals and constraints generated in the goal / constraint input step, the local resident information, and the space information. A policy formulation method characterized by:

3. The policy formulation method according to claim 1, In the simulation execution and evaluation step, Using the generating AI, interpretation information is generated that indicates the execution result of the simulation in an interpretable manner for the user, and an evaluation of the execution result including the interpretation information is displayed on the display screen. A policy formulation method characterized by:

4. The policy formulation method according to claim 1, In the goal and constraint input step, Determine whether the input by the user satisfies the amount of information required to generate the goal and the constraints, and if the amount of information required is not satisfied, accept input of additional information from the user via a predetermined user prompt, or obtain the additional information by querying the generation AI, and generate the goal and constraints of the measure using the generation AI based on the input by the user and the additional information. A policy formulation method characterized by:

5. The policy formulation method according to claim 1, The generating AI has a learning model that has learned how to use a simulator that executes the people flow simulation, In the task generation step, The simulation task is generated based on the usage of the simulator that the learning model has learned. A policy formulation method characterized by:

6. The policy formulation method according to claim 5, In the task generation step, Using the generating AI, a data file used in the simulator is generated by converting any natural language information into data. A policy formulation method characterized by:

7. The policy formulation method according to claim 5, The generation AI detects updates to the simulator and updates the learning model if an update is found. A policy formulation method characterized by:

8. The policy formulation method according to claim 5, the learning model has learned how to use a plurality of the simulators; In the task generation step, generating the simulation task based on a method of using one or more of the selected simulators according to the goal and the constraints; A policy formulation method characterized by:

9. The policy formulation method according to claim 5, The generation AI generates the measures and the simulation tasks by referring to external information. A policy formulation method characterized by:

10. The policy formulation method according to claim 1, The data management unit The virtual space is constructed using space measurement data including at least one of CAD (Computer Aided Design) data, point cloud data, and mesh data. A policy formulation method characterized by:

11. The policy formulation method according to claim 1, In the simulation execution and evaluation step, The people flow simulation is performed using people flow measurement data including GPS (Global Positioning System) information of the user in the real space, history information of the user's IC (Integrated Circuit) ticket, and LiDAR (Light Detection And Ranging) measurement data that measures the distance to the user, or a behavior model created based on the people flow measurement data. A policy formulation method characterized by:

12. The policy formulation method according to claim 1, The space is It is an advertising space for displaying advertisements and a tenant space, which is a room or section for installing tenants. A policy formulation method characterized by:

13. The policy formulation method according to claim 1, The above measures are: The space usage method, the selling price of the space, and the potential buyers of the space are included. A policy formulation method characterized by:

14. A policy formulation system that uses a virtual space that imitates a real space to evaluate policies related to the use of the real space, a data management unit that manages local resident information including needs for the space of users including local residents of the real space, and space information regarding attributes of the space; a goal and constraint input unit that generates goals and constraints of the measures based on input by a user using artificial intelligence (AI); a task generation unit that generates a simulation task for executing a people flow simulation that simulates the flow of people based on the goals and constraints generated by the goal / constraint input unit, the local resident information, and the space information, using the generation AI, for the measures that are generated based on the goals and constraints generated by the goal / constraint input unit, the local resident information, and the space information; a simulation execution and evaluation unit that executes the simulation task generated by the task generation unit in the virtual space to execute the people flow simulation and displays the execution results of the people flow simulation on a display screen; A policy formulation system comprising:

Images