Information processing methods, computers, and programs

Abstract

Further improvements will be made to the voice interaction system that allows for user interaction. [Solution] An information processing method according to one aspect of this disclosure is executed by a computer included in a system attached to a room. The AI ​​agents include a room AI agent that receives and responds to instructions from a user in the room via a voice interaction interface (VUI), and one or more personal AI agents that correspond to one or more users in the room. The information processing method connects each AI agent to a virtual space in which they can interact with each other. The room AI agent recognizes the user in the room and the topic of the suggestion request from the user via the VUI and notifies one or more personal AI agents. One or more personal AI agents generate one or more suggestions based on the topic and the user's profile. The computer applies settings or processing to the system according to the suggestion adopted by the user.

Description

【Technical Field】 【0001】 The present disclosure relates to an information processing method, a computer, and a program. For example, the present disclosure relates to an information processing method using an AI agent capable of voice interaction with a user in a computer that cooperates with a room. 【Background Art】 【0002】 Patent Document 1 discloses a system for calling a voice interaction agent. 【Prior Art Documents】 【Patent Documents】 【0003】 【Patent Document 1】 Japanese Patent Application Laid-Open No. 2021-117302 【Summary of the Invention】 【Problems to be Solved by the Invention】 【0004】 In the conventional voice interaction agent system, further improvement is required. 【0005】 One of the problems to be solved by the present disclosure is to achieve further improvement with respect to an agent capable of interactive dialogue associated with a user. [[ID=...]] 【Means for Solving the Problems】 【0006】 An information processing method according to one aspect of this disclosure is an information processing method for an AI agent executed by a computer included in a system associated with a room. The AI ​​agent includes a room AI agent and one or more personal AI agents. The room AI agent receives and responds to instructions from users in the room via a voice interaction interface (VUI). The one or more personal AI agents correspond to one or more users in the room. The information processing method includes connecting the one or more personal AI agents and the room AI agents to a virtual space in which they can interact with each other, and when the room AI agent connected to the virtual space receives a request for a suggestion on a topic from one or more users in the room, it recognizes the user who made the suggestion request and the accepted topic via the voice dialogue interface (VUI), notifies the one or more personal AI agents connected to the virtual space of the result of the recognition, the one or more personal AI agents connected to the virtual space generate one or more suggestions based on the recognized topic and the profile of the recognized user, and when a suggestion is adopted by the user from among the one or more suggestions generated, the computer applies settings or processing to the system associated with the room according to the adopted suggestion. The one or more personal AI agents include a first agent corresponding to a first user in the room and a second agent corresponding to a second user in the room. The virtual space includes a first cyber room, a first cyber room, and a second cyber room. The first cyber room is a cyber room in which each of the AI ​​agents can interact with each other. The first cyber room is a cyber room in which the room AI agent and the first agent can interact with each other. The second cyber room is a cyber room in which the room AI agent and the second agent can interact with each other. If the room AI agent receives a request for a proposal from the first user or the second user regarding a first topic unrelated to the user's privacy information, it will interact with at least the personal AI agent corresponding to that user in the first cyber room. When the room AI agent receives a request for a proposal from the first user or the second user regarding a second topic related to the user's privacy information, it interacts with the personal AI agent corresponding to the user in the first cyber private room or the second cyber private room, where information is exchanged between the two parties. [Effects of the Invention] 【0007】 According to one aspect of the information processing method described herein, further improvements can be achieved using an AI agent capable of dialogue linked to the user. [Brief explanation of the drawing] 【0008】 [Figure 1] Figure 1 shows an example of the overall configuration of the dialogue agent system according to this embodiment. [Figure 2A] Figure 2A shows an example of the hardware configuration of the dialogue agent system according to this embodiment. [Figure 2B]Figure 2B is a flowchart illustrating an example of the processing flow of the dialogue agent system according to this embodiment. [Figure 3] Figure 3 illustrates an example of connecting a user-used agent to the conversational agent system according to this embodiment. [Figure 4] Figure 4 is a sequence diagram showing an example of the process of connecting a user agent to the conversational agent system according to this embodiment. [Figure 5] Figure 5 illustrates an example of connecting a user-used agent to the conversational agent system according to this embodiment. [Figure 6] Figure 6 illustrates an example of adding the dialogue agent system according to this embodiment to a conversation group in which an agent used by a user participates. [Figure 7] Figure 7 is a sequence diagram showing an example of the process from adding the user's agent to a conversation group and then excluding it from that conversation group, according to this embodiment of the dialogue agent system. [Figure 8] Figure 8 is a sequence diagram showing an example of the processing flow in the dialogue agent system according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this decision is reflected in the in-vehicle system. [Figure 9] Figure 9 shows an example of how, in the dialogue agent system according to this embodiment, the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, allows the user to decide whether to accept or reject them, and reflects this decision in the in-vehicle system (navigation system). [Figure 10]Figure 10 is a sequence diagram showing an example of the processing flow in the dialogue agent system according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this decision is reflected in the in-vehicle system (navigation system). [Figure 11] Figure 11 is a diagram illustrating an example of the dialogue history between agents in the dialogue agent system (cyber vehicle interior) according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, allows the user to decide whether to accept or reject them, and reflects this decision in the in-vehicle system (navigation system). [Figure 12] Figure 12 is a flowchart illustrating an example of the process flow in the dialogue agent system according to this embodiment, where the vehicle agent uses each user's agent to narrow down the candidates recommended to the user, and the user decides whether to accept or reject them. [Figure 13] Figure 13 is a sequence diagram showing an example of the processing flow in the dialogue agent system according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this decision is reflected in the in-vehicle system (navigation system). [Figure 14] Figure 14 is a flowchart illustrating an example of the process flow in the dialogue agent system according to this embodiment, where the agent switches the recipient of information based on the nature of the information when sending information. [Figure 15] Figure 15 shows an example in the dialogue agent system according to this embodiment, in which a vehicle agent responds to a user's privacy inquiry by communicating individually with only the user's agent. [Figure 16]FIG. 16 is a sequence diagram showing an example of a process in which a vehicle agent responds to an inquiry regarding a user's privacy by communicating individually only with the user's agent in the dialogue agent system according to the present embodiment. [Figure 17] FIG. 17 is a diagram showing an example of using an API (HTTP request / response) when a vehicle agent communicates individually with a user's agent in the dialogue agent system according to the present embodiment. [Figure 18] FIG. 18 is a diagram showing an example of user data referred to by a user's agent in the dialogue agent system according to the present embodiment. [Figure 19] FIG. 19 is a diagram showing an example of cooperation between the software configuration and the hardware configuration of the dialogue agent system according to the present embodiment. 【Embodiments for Carrying Out the Invention】 【0009】 [Findings on which the present disclosure is based] In recent years, technological development of AI agents utilizing LLMs (Large-Scale Language Models) has been progressing. AI agents can directly or indirectly access short- and long-term memories (user attributes, usage logs, or parts thereof), and can autonomously communicate with external applications and web services via a network, as well as launch and operate other applications and web services. As a result, an AI agent is a computer system, or software program, that sets or updates goals through text or voice communication with the user (the content of instructions to the AI ​​is also called a prompt), autonomously generates a set of tasks necessary to achieve the goal, and sequentially executes the information processing of the generated tasks autonomously or while communicating with the user, thereby achieving the final goal. AI that can handle information not only in one modal (data format) such as text, but also by combining multiple different modals such as voice and images for input or output processing is called a multimodal type, and technological development of multimodal AI agents is also progressing. 【0010】 AI agents can acquire specialization and characteristics depending on the databases they refer to when processing information and the algorithms used for task generation. This allows AI agents to be implemented as highly specialized agents focused on specific functions. On the other hand, AI agents can be implemented as personalized agents that cater to individual users by learning the preferences, biometric information, and past behavioral history of the user they communicate with, and by accessing and analyzing (hereinafter referred to as "profiles") databases containing such individual user data. The former type of AI agent is sometimes called a specialized agent, and the latter a partner-type AI agent. 【0011】 The partner-type agent is considered to be able to function as a particularly effective partner in the mobility space. This is because when a user moves to a place different from their normal activity range by vehicle and has a new (or non-daily) experience, the partner-type agent can become a suitable navigator that conforms to the user's personality. For example, when navigating the driving route of a vehicle, the partner-type agent can make selections or proposals that reflect the user's preferences. Examples of user preferences include whether the user prefers the shortest route, a road with sidewalk separation or the like that is easy to drive on, or visiting tourist spots. 【0012】 Similarly, the partner-type agent can propose candidates for sightseeing and meals at the destination, reflecting the user's preferences and past behavior history. For example, if the user likes historical sites, it is conceivable to propose a stop at historical sites around the route, and if the user prefers a local dining experience, it is conceivable to propose a meal at a restaurant where famous local cuisine can be eaten. 【0013】 The inventor has examined a series of user experiences regarding the use of vehicles and AI agents. 【0014】 First, what was assumed was a setting method for connecting or disconnecting the vehicle agent that can be used in the vehicle and the agent of the user who gets on the vehicle. There has been no examination of a method for easily connecting the user's agent to the vehicle agent when the user gets on the vehicle, and a method for easily disconnecting the user's agent from the vehicle agent when the user gets off the vehicle, and it was considered that there are new issues from the perspectives of user experience (hereinafter abbreviated as UX) and implementation means. 【0015】 Furthermore, we considered that by connecting the vehicle agent with the agent of each passenger, it might be possible to support the selection of a single decision that satisfies all users from a vast number of options regarding the vehicle's route and destination, taking into account the ever-changing traffic conditions, the congestion levels of tourist spots and restaurants, and even the preferences and interests of each passenger. However, no solution has been shown that supports decision-making or consensus building among more than one user inside the vehicle, and it seemed that there were unresolved issues regarding UX and implementation methods. 【0016】 Furthermore, we believe that if the vehicle agent and each user's agent are connected and their interactions are shared in a single online meeting room-like environment, a new problem will arise: each user's preferences and interests may be shared with other users' agents. In other words, a new problem may arise: information leakage to agents used by other users. 【0017】 Furthermore, we considered cases where one or more users make decisions or change their behavior regarding travel, not limited to inside a vehicle. These included cases where users discuss travel-related topics with family members in a room at home before using a vehicle, or where they discuss customer acquisition topics with colleagues in a conference room at the office. For this reason, we considered car interiors, rooms at home, and office conference rooms as rooms in a broad sense and accumulated knowledge and insights. Common challenges included the fact that reaching a consensus takes longer the more users there are, and that support is needed to encourage users to change their travel behavior regarding the agreed-upon topic. 【0018】 Each aspect of this disclosure described below is based on the above findings; however, the invention described in the claims is not limited to the above findings. 【0019】 [Summary of the Embodiment] An information processing method according to one aspect of this disclosure is an information processing method for an AI agent executed by a computer included in a system associated with a room. The AI ​​agent includes a room AI agent and one or more personal AI agents. The room AI agent receives and responds to instructions from users in the room via a voice interaction interface (VUI). The one or more personal AI agents correspond to one or more users in the room. The information processing method includes connecting the one or more personal AI agents and the room AI agents to a virtual space in which they can interact with each other, and when the room AI agent connected to the virtual space receives a request for a suggestion on a topic from one or more users in the room, it recognizes the user who made the suggestion request and the accepted topic via the voice dialogue interface (VUI), notifies the one or more personal AI agents connected to the virtual space of the result of the recognition, the one or more personal AI agents connected to the virtual space generate one or more suggestions based on the recognized topic and the profile of the recognized user, and when a suggestion is adopted by the user from among the one or more suggestions generated, the computer applies settings or processing to the system associated with the room according to the adopted suggestion. 【0020】 According to this, when there are multiple users inside the vehicle, the vehicle's AI agent can generate suggestions that meet the needs of each user inside the vehicle, in consultation with each user's AI agent. For example, if each user's AI agent understands and analyzes the personality and preferences of the user it is responsible for (hereinafter referred to as a profile), it can communicate suggested proposals that reflect the personalities and preferences of the multiple users inside the vehicle to the vehicle's AI agent via the virtual space. The vehicle's AI agent can then adjust these suggestions to make a proposal that meets the needs of all the users inside the vehicle. 【0021】 Because the in-vehicle AI agent handles the interaction between the user and the AI ​​agent, rather than the users inside the vehicle, unnecessary communication between the user and the dialogue system is eliminated. As a result, the load on speech recognition processing can be reduced compared to a system where the in-vehicle AI agent and the user engage in step-by-step voice dialogue to coordinate opinions. Furthermore, for example, if one of the users is operating the vehicle, eliminating unnecessary communication reduces cognitive load and lowers the risk of accidents. 【0022】 For example, if the vehicle is an autonomous vehicle such as a robotaxi, it can perform autonomous driving towards a destination that meets the needs of multiple users inside the vehicle, along a route, or in a manner that suits the driving operation. In this disclosure, "destination" is understood to include not only the final destination but also intermediate stops, and furthermore, parking locations and stopping locations. 【0023】 The "system associated with the room" mentioned above refers to a system configured to control predetermined services provided to users entering a predetermined space from which users enter and exit, and includes at least one computer. Here, "predetermined services" may include services related to space provision such as reservation and entry / exit management, services related to the indoor environment such as seating, lighting, and air conditioning, services for projecting materials, maps, and content, and services for providing communication with remote locations via web conferencing, etc. These services may be provided via an AI agent. Furthermore, "the predetermined space from which users enter and exit" may include mobility spaces such as the passenger compartment of a vehicle, or indoor spaces such as conference rooms, rest rooms, or private rooms (booths) for online meetings. As an example, "the system associated with the room" may also be an in-vehicle system installed in a vehicle with a passenger compartment. If the "associated system" is an in-vehicle system, "predetermined services" may include driving assistance services such as navigation to a destination and autonomous driving. Note that "the computer included in the system associated with the room" may be, for example, only the first computer included in the in-vehicle system. In this case, the vehicle AI agent is implemented within the in-vehicle system. Alternatively, "a computer included in the system associated with the room" may refer to multiple computers, including a first computer included in the in-vehicle system and a second computer that communicates with the first computer via a network. In this case, the vehicle AI agent may be implemented within the in-vehicle system or within the second computer (e.g., a server), or it may be implemented in a distributed manner across both. 【0024】 The "virtual space" described above may be a virtual space managed by the first computer and / or the second computer. Alternatively, the "virtual space" may be a virtual space managed by a third computer (e.g., a server) on a wide-area network (e.g., the internet) operated by a different operator than the first computer and / or the second computer. If the virtual space is managed by the second or third computer, the vehicle AI agent's connection to the virtual space may be a connection using communication over a wide-area network. If the virtual space is managed by the first computer, the vehicle AI agent's connection to the virtual space may be a connection within the in-vehicle system, and the first and second AI agents' connections to the virtual space may be, for example, a connection using P2P communication between each user's information terminal and the in-vehicle system. 【0025】 [Embodiment] Hereinafter, with reference to the drawings, an example of an information processing method, computer, program, communication terminal, and dialogue agent system according to this embodiment will be described. 【0026】 First, let's define what an AI agent (Artificial Intelligence Agent) is. An AI agent is software or a computer system designed to achieve predefined goals. Based on external information obtained via a network regarding communication with the user, the environment surrounding the user, and interactions with the user, the AI ​​agent is designed to autonomously generate, select, and execute actions to achieve its goals. Communication with the user includes all means of conveying the user's emotions, will, and thoughts. For example, this includes one or more means such as conversation, timing before speaking, tone of voice, expression of intent through visual information such as letters and symbols (GUI), expression of intent through physical operations such as buttons and switches, facial expressions, gaze, posture, or physical actions such as gestures. In this embodiment, an AI agent will also be simply referred to as an agent. The agent may be represented as a unique character (avatar) as a visual and auditory bodily representation when interacting with the user, and in that case, it will have attribute information for that avatar. The attribute information of an avatar representing an agent's character includes, for example, information about one or more of the following: the avatar's appearance, body, clothing, accessories, gestures, facial expressions, voice, personality, preferences, habits, knowledge, or experience (records of past interactions with users). 【0027】 In other words, the "AI agent" as disclosed herein is a computer system that, in accordance with the intentions, instructions, or past behavioral data of a human user, acts on behalf of that user to autonomously complete processes intended or instructed by the user through API communication to the computer system and operation of applications. In other words, "the AI ​​agent connecting to a virtual space" or "the AI ​​agent performing dialogue processing within a virtual space" is equivalent to the AI ​​agent participating in an online space such as an online meeting room or chat room on behalf of the user. Furthermore, the "AI agent" uses mechanisms that enable information sharing within that online space to interact with other people or other AI agents and exchange information. For example, broadcasting information to an unspecified number of people on social media is also included in "connecting to a virtual space." 【0028】 Specifically, by inputting the output of one AI agent into another AI agent, and then repeating the reverse process, a state in which two AI agents are virtually conversing can be created. In the conversational agent system disclosed herein, each of the two or more AI agents understands the user it corresponds to and engages in conversation with the other AI agent. Here, inputting the output of one AI agent into another AI agent can be achieved by the AI ​​agent operating a browser via voice / GUI (automatically through software processing), by using a chat room in an online meeting room, or by communicating information via an API. 【0029】 Therefore, the "AI agent" relating to this disclosure is a computer system that can autonomously receive a highly abstract task requested by a user and perform a series of information processing necessary for task completion, such as subdividing the task, processing it, evaluating it, and prioritizing the next action, in order to complete the task. 【0030】 In the following, the agent used by a user will be described as a program or computer system designed and configured to autonomously communicate with vehicle agents, other users' agents, or web services, etc., on behalf of the user, based on the user's preferences and past behavioral history. 【0031】 [Configuration of the conversational agent system] Figure 1 shows an example of the overall configuration of the dialogue agent system 100 according to this embodiment. For the sake of explanation, in this disclosure, communication using text data other than dialogue, image data, voice data, haptic data, user biometric information, information combining one or more of these, or information processed from one or more of these may be collectively referred to as dialogue. This communication includes communication between users, between users and agents, and between agents. 【0032】 The dialogue agent system 100 according to this embodiment is an example of a dialogue agent system that can interact (communicate) with a user. 【0033】 As shown in Figure 1, the dialogue agent system 100 according to this embodiment consists of an in-vehicle system 102 mounted on a vehicle, a vehicle agent 101 which operates on the in-vehicle system 102 (or on a computer on a network that operates in conjunction with the in-vehicle system 102), an information terminal A104 used by user A riding in the vehicle, agent A103 which operates on information terminal A104 (or on a computer on a network that operates in conjunction with information terminal A104), an information terminal B107 used by user B108 riding in the vehicle, agent B106 which operates on information terminal B107 (or on a computer on a network that operates in conjunction with information terminal B107), and a computer network that enables communication between the in-vehicle system 102, information terminal A104, and information terminal B107. The dialogue agent system 100 not only includes these devices and agents as components, but is also a system in which at least two or more agents connected via the network 109 can exchange information with each other. 【0034】 Here, the vehicle agent 101 is an example of a room AI agent in this embodiment. 【0035】 Here, we assume that each user has an agent who knows them well (who can access data including their profile or user privacy information). 【0036】 Here, an agent that knows each user well is an example of a personal AI agent in this embodiment. 【0037】 The vehicle's interior space (real interior) contains the in-vehicle system 102, user A, information terminal A104, user B108, and information terminal B107. On the other hand, the online virtual space (e.g., an online meeting room) where the vehicle agent 101 and the agent of the passenger communicate is called the cyber interior. As will be explained in more detail later, when a user boards (enters) the real interior, it can be thought of as the agent used (or acting as a proxy) for that user virtually boarding (entering) the cyber interior. Similarly, when a user alights (exits) the real interior, the user's agent is also considered to have alighted (exited) the cyber interior (e.g., left the online meeting room), and the connection with other agents is severed. 【0038】 The vehicle agent 101, which is an AI agent in the vehicle, works in conjunction with the in-vehicle system 102 to acquire, control, or modify various information about the vehicle. For example, it can acquire, control, or modify the vehicle's current location, driving-related information (such as speed and driving operation information), vehicle sensor data (such as camera images), fuel level, battery level, settings for the vehicle's air conditioning and lighting, the route being set, and information about the content being played. 【0039】 Agent A103, a personal AI agent, works in conjunction with information terminal A104 to acquire, modify, and record various information about the user (hereinafter referred to as user data). For example, it can acquire, modify, or record the user's name, username used for service use, address, gender, age, facial image, health status, preferences, knowledge, behavioral history, payment information such as credit cards and electronic money, and biometric information, including real-time data. This user data may be securely stored in the memory of information terminal A104, or it may be stored in the memory of an external computer connected via a network that Agent A103 can access when needed. Similarly, Agent B106 works in conjunction with information terminal B107. 【0040】 By connecting the in-vehicle system 102, information terminal A104, and information terminal B107 via a network in this way, agents that correspond to users in the real vehicle can be virtually connected and interacted with in a cyber vehicle on the network. 【0041】 In this embodiment, when a user makes a request to or responds to the in-vehicle system 102 or the vehicle agent 101, the description may not explicitly distinguish which system the user is speaking to. This is because the user's intention at the time is what they are addressing, and it is difficult to identify from the outside. However, the in-vehicle system 102 located in the real vehicle cabin is the physical communication interface for the user's requests and responses, and depending on the content of the communication, the vehicle agent 101's avatar may act as a person-to-person communication interface with the user via the UI unit 215 (described later). When it is important to emphasize that the communication is more person-to-person, the description may state that the user is communicating with the vehicle agent 101 rather than the in-vehicle system 102. 【0042】 While this description focuses on a configuration where a user's agent, who is in a physical vehicle, is connected to a cyber vehicle, this disclosure is not limited to this configuration. The agent connected to the cyber vehicle may be that of a user who is not in a physical vehicle. For example, by having the agent of a user who is in a remote location enter the cyber vehicle, it is possible to share audio from inside the physical vehicle and camera footage from the vehicle with the computer used by the remote user, thereby providing a simulated experience as if the user were traveling together in the vehicle. 【0043】 Furthermore, the audio from the real vehicle interior, the vehicle's current location, and camera footage may be shared with the in-vehicle system (or vehicle agent 101) of another vehicle or with the agent of a user riding in another vehicle. In this case, smooth communication across vehicles is possible, especially when multiple vehicles are traveling together. 【0044】 In this way, by recreating the vehicle's interior space in cyberspace and connecting a vehicle agent 101 representing the vehicle and an agent representing the user to it, it becomes possible to collect and analyze real-time information from both inside and outside the vehicle within the cyber cabin, create recommendations for users in the real vehicle cabin, and support user decision-making. Previously, there were challenges such as a lack of information about the vicinity of the route and difficulties in multiple people making a single decision smoothly, but with the above configuration, it is possible to make the user's travel experience in the real vehicle cabin richer, more efficient, and more valuable. 【0045】 Figure 2A shows an example of the hardware configuration of the dialogue agent system 100 according to this embodiment. 【0046】 Information terminals A104 and B107 both have the same hardware configuration. They include a detection unit 203 for acquiring video information, audio information, and / or physical quantities of the surrounding environment; a UI unit 205 that provides the user with information via video, audio, and vibration, and accepts button presses and touch operations; a processing unit 202 that performs various calculations, including AI model learning and inference processing, as well as information processing such as image display and audio playback; a memory 204 that holds data and files used by the processing unit; and a communication unit 201 for communicating with other computers on the communication network. 【0047】 The UI unit 205 includes a display for displaying a GUI (Graphical User Interface) and a speaker and microphone for inputting and outputting a VUI (Voice User Interface). The arithmetic unit 202 is an example of an AI processor for causing the agent to execute generation processing. The memory 204 is an example of a memory that stores access information, including addresses (endpoints) for accessing the agent, and agent attribute information about the agent. The memory 204 is also an example of a memory that stores user usage logs of the agent. Furthermore, the memory 204 is an example of a memory that stores usage logs for one or more agents available to the user. 【0048】 In this embodiment, information terminal A104 or information terminal B107 is described as a smartphone, but is not limited to that. It may take the form of a smartwatch, smart glasses, smart earphones worn on the ear, smart ring, voice-controlled smart speaker, or even a robot with movable parts, as long as it is a device that the user can wear or carry. 【0049】 The in-vehicle system 102 includes a communication unit 211 for communicating with other computers on a communication network, a memory 214 that stores information about the vehicle and its management programs, and a processing unit 212 that performs various data processing related to the vehicle's core systems and vehicle functions, including the car navigation system, content playback, and learning and inference processing for the vehicle agent 101. 【0050】 Furthermore, the in-vehicle system 102 includes a control unit 216 that controls and drives the vehicle's driver assistance and autonomous driving functions, as well as the actual in-vehicle equipment (seats, lighting, air conditioning, etc.), a detection unit 213 for detecting the space and objects around the vehicle, as well as the position and state of people and objects inside the vehicle, and a UI unit 215 that provides video and audio information and accepts input from passengers such as touch and voice operations. Here, vehicle driver assistance may include driver assistance linked to a navigation system with a set destination, automatic detection of deviations from the route, alarms, recalculations, or pre-downloading maps from the cloud. 【0051】 The UI unit 215 includes a display for displaying a GUI, and a speaker and microphone for inputting and outputting VUI. The detection unit 213 includes at least one of the following: a camera or LiDAR for detecting the space and objects around the vehicle, an interior camera or interior monitoring system for imaging the real interior of the vehicle, a GPS sensor for measuring the current position of the vehicle, sensors provided at each seat in the interior of the vehicle, and sensors for detecting the vehicle's driving state or control state or driving operations by the driver. The calculation unit 212 is an example of an AI processor for causing the vehicle agent to perform generation processing. The memory 214 is an example of a memory that stores access information including an address for accessing the vehicle agent, and agent attribute information related to the vehicle agent. The memory 214 may also record, for example, information for identifying users or users' agents who have used the vehicle in the past. 【0052】 Furthermore, information terminals A104 and B107, and the in-vehicle system 102 may communicate using means other than computer networks such as the Internet of a wide-area communication network. For example, short-range wireless communication may be used for communication processing between the in-vehicle system 102 and information terminal A104. 【0053】 Figure 2B is a flowchart showing an example of the information processing overview in the dialogue agent system 100 according to this embodiment. 【0054】 The information processing in the dialogue agent system 100 includes at least connection processing (S2801), consensus building processing (S2802-S2806), operation processing (S2807), and disconnection processing (S2808). 【0055】 During the connection process, the AI ​​agents are connected to the virtual space (S2801). The connected AI agents include one or more personal AI agents corresponding to one or more users in the room, and a room AI agent corresponding to the room. 【0056】 A personal AI agent, in this context, is an agent that works closely with the user, capable of accessing and analyzing (profiling) data such as the user's personal information, past behavior, and preferences. Personal AI agents are typically used by users on their own information devices, such as smartphones. 【0057】 Here, the room AI agent is an agent corresponding to the room the user is entering or exiting, and can configure and modify the hardware attached to or linked to the room, and the software that controls it. The hardware and the software that controls it have a user interface, which includes a GUI provided via a display, etc., and a VUI provided via a microphone, speaker, etc. It also includes a projector for projecting content, room lighting, and air conditioning. Furthermore, if the room is a vehicle, it includes hardware and software that makes the vehicle move as a vehicle, as well as hardware and software that controls driving assistance such as route guidance, navigation, and autonomous driving. Here, autonomous driving is not limited to fully autonomous driving, but also includes limited autonomous driving such as cruising mode where autonomous driving is deactivated by braking. 【0058】 Here, the virtual space corresponds to the room the user enters and exits, and one or more connected personal AI agents and room AI agents collaborate on processing. This collaborative processing includes autonomously aiming for a goal (such as a proposal) in response to requested topics, using generative AI technology. 【0059】 Following the connection process, a consensus-building process takes place. This consensus-building process utilizes generative AI technology to autonomously aim for a Goal (proposal, etc.). 【0060】 In the consensus-building process, if a user in the room requests a topic, the room's AI agent, which is responsible for the room, recognizes the requester (user) and understands the content of the request (recognition) through voice dialogue processing (S2802). 【0061】 Here, user recognition is typically performed by analyzing the user's utterances, but the topic requester may also be analyzed by analyzing the video footage acquired by the indoor camera. 【0062】 The room AI agent notifies the personal AI agents connected to the virtual space of the recognition results from the voice dialogue processing, namely the recognized requester (user) and the recognized (understood) topics (S2803). For example, if there are two users entering the room, user A and user B, and user A requests a recommendation for a ramen restaurant as a topic, the room AI agent notifies the personal AI agents of both user A and user B, who are connected to the virtual space, that user A has requested a recommendation for a ramen restaurant as a topic. 【0063】 A personal AI agent, notified of suggestions related to a topic, generates one or more suggestions, utilizing the corresponding user's profile. These suggestions may include evaluations and recommendations. The AI ​​agent may also judge the effectiveness and prioritize the generated suggestions. The AI ​​agent understands local and map information of the region to which the destination belongs, such as the room's location on a map and the current time, and uses this information to judge effectiveness and prioritize. 【0064】 For example, if user A's profile indicates they dislike spicy food, and user B's profile indicates they like spicy food, then the system will suggest two restaurants: Restaurant A (destination) that serves both spicy and non-spicy ramen, and Restaurant B (destination) that serves only non-spicy ramen. The system will also provide information indicating that Restaurant A is recommended at level 4 out of 5, and Restaurant B is recommended at level 3. Restaurant A is rated higher because it satisfies the preferences of both users. 【0065】 The AI ​​agent in the room considers the distance from the room to restaurants A and B, as well as the estimated arrival time. If restaurant B cannot be reached during lunchtime or if there is no parking nearby, the agent may determine that restaurant B is not a viable option and remove it from the suggestions. Alternatively, it may add restaurant C, which it is aware of, to the suggestions. Restaurant C can be identified, for example, from emails containing recommended lunchtime restaurants that are electronically delivered to the area where the room is located. 【0066】 While it is possible for a personal AI agent, rather than the room AI agent, to access local information and map information, accessing such information may require third-party charges. Therefore, it is more practical for the room AI agent, which holds and stores information centered on the room's location, to access local information and map information, rather than for a personal AI agent to respond to users moving to various locations. 【0067】 If the AI ​​agent generates a proposal, it will at least present the proposal to the client (S2805). This presentation may be done via a voice user interface (VUI) for voice dialogue processing, or via a graphical user interface (GUI) on a display or similar device. 【0068】 If the user selects one of the presented proposals, the room AI agent requests the computer attached to or linked to the room to configure and modify the room's system (S2806). 【0069】 The consensus-building process ends here, and then the action processing takes place. 【0070】 In the operation process, a computer attached to or linked to the room executes the requested settings and changes to the room using hardware and the software that controls them (S2806). For example, this includes the room's lighting, air conditioning, and content playback projector. Furthermore, if the room is a vehicle, it includes route guidance, navigation, and autonomous driving to reach the destination. For example, if the user has adopted the suggestion of a delicious ramen restaurant A, content introducing ramen restaurant A and access information will be provided to the user using a projector or display. If it is a vehicle, route guidance, navigation, and autonomous driving will be performed with ramen restaurant A set as the destination. Lighting and air conditioning will be controlled in conjunction with the content being played, or, if the vehicle is moving, will be controlled according to environmental changes due to movement. 【0071】 Disconnection typically follows operation, but can occur at any time after connection. For example, it can be performed as a forced termination during consensus building or operation. 【0072】 The disconnection process disconnects the AI ​​agent's connection to the virtual space (S2808). Of the AI ​​agents, at least the personal AI agent is disconnected, but the room AI agent may be left resident without being disconnected. 【0073】 Disconnection typically occurs when a user in the room requests termination, but it can also occur automatically when the user leaves the room. In the case of a conference room, it can occur when the conference room use is completed. In the case of a vehicle, it can occur when the user arrives at their destination or parks in their home's parking lot. If the vehicle is a rental car, it can occur when the rental contract period ends or when the rental car is returned to the rental company. In addition, if there are multiple users, only the personal AI agent of the user who leaves the room may be disconnected. 【0074】 The following describes in detail the embodiment outlined in Figure 2B. 【0075】 [Connecting / Disconnecting agents to the cyber bay] Next, using Figures 3 to 8, we will explain an example of a connection process in an embodiment in which the user's agent enters and exits (connects / disconnects) the cyber vehicle in conjunction with the user entering and exiting the real vehicle. 【0076】 Figure 3 illustrates an example of connecting a user agent to the dialogue agent system 100 according to this embodiment. In this example, user A105 uses information terminal A104 to read a QR code (registered trademark) to access vehicle agent 101 or the cyber cabin. The following is a chronological explanation. 【0077】 First, in (I), the avatar of the vehicle agent 101 is displayed on the UI section 215 of the in-vehicle system 102. There, the vehicle agent 101 asks user A, "Do you want to connect with the agent?", confirming whether user A intends to connect user A's agent with the vehicle agent 101 (or cyber cabin, hereafter the same). In response to this question, user A answers "yes", indicating that they intend to do so. 【0078】 Next, in (II), the vehicle agent 101 displays a QR code containing access information for connecting the user's agent to the vehicle agent 101 on the UI unit 215 of the in-vehicle system 102, based on the user's intent. Furthermore, the vehicle agent 101 instructs user A105 to scan the QR code with their smartphone. User A105 launches the application installed on the information terminal A104 and scans the QR code. 【0079】 Next, in (III), the app reads the access information for the connection destination contained in the QR code (for example, connection information for an online meeting, connection information indicating a specific chat channel, etc.) and displays "Who do you want to connect to the Cyber ​​Cabin?" on the screen of information terminal A104, allowing user A105 to select which of the multiple agents the user is using to connect to. Here, the screen allows user A105 to choose from two avatars: an avatar of agent A103 that can respond autonomously using user A105's preferences and past history, and an avatar of another agent used for a specific purpose. User A105 selects the avatar of agent A103 and designates agent A103 as the agent to connect to the Cyber ​​Cabin. Note that agents used for other specific purposes are, for example, agents used to support user A's studies or work. 【0080】 Next, in (IV), the app attempts to connect to the user's agent A103 according to the access information for the cyber room contained in the QR code. Here, the UI section 205 of the information terminal A104 displays to user A105 that agent A103 is attempting to connect to the cyber room of this vehicle, making it easier for them to understand the status. 【0081】 Next, in (V), the applications of the vehicle agent 101 and the information terminal A104 each detect that agent A103 has connected to the cyber room and notify the user accordingly. The vehicle agent 101 says, "Connected to agent A," and the UI section 205 of the information terminal A104 displays the words "Connection complete." User A then understands that the vehicle agent 101 (or cyber room) and agent A103 have connected and that direct communication is now possible. 【0082】 Next, in (VI), based on the fact that vehicle agent 101 has been able to connect with agent A103, it says, "From now on, I will consult with agent A and guide you." This means that vehicle agent 101 will make recommendations, suggestions, selections, or decisions for user A based on agent A103's opinions and evaluations regarding various tasks related to the vehicle and its interior, such as selecting the route ahead, choosing places to stop at such as tourist spots and restaurants, selecting content to play inside the vehicle, and controlling the air conditioning, lighting, seat angle, and seat temperature for user A. User A, understanding this, acknowledges to vehicle agent 101 by saying "Thanks" and requests that the process be carried out. 【0083】 This section illustrates an example of a procedure in which the in-vehicle system 102 displays connection information to the Cyber ​​Cabin (or vehicle agent 101) in the form of a QR code, and the user reads it via an application on their information terminal to connect their agent. 【0084】 Alternatively, instead of scanning a QR code, the in-vehicle system 102 may perform personal identification of user A105 using their face, voice, other biometric information, or gestures, and connect agent A103 to the cyber cabin when user A105 gets in the vehicle. Similarly, the in-vehicle system 102 may disconnect agent A103 from the cyber cabin when user A105 gets out of the vehicle. This is to prevent discrepancies in the information received by user A105 and agent A103 that would occur if agent A103 remained in the cyber cabin even though user A105 had gotten out of the real cabin. 【0085】 Alternatively, the information terminal A104 may be triggered to automatically connect agent A103 to the cyber cabin when it detects wireless communication information transmitted by the in-vehicle system 102 (such as the SSID of the Wi-Fi for inside the vehicle, the MAC address and link key of Bluetooth®). If the access information for the cyber cabin is fixed, this can be achieved by recording the access information from previous connections in memory 204. 【0086】 Alternatively, by touching the information terminal A104 to the NFC device installed in the vehicle, access information to the cyber bay can be read, and agent A103 can automatically connect to the cyber bay. 【0087】 The dialogue agent system 100 according to this embodiment includes a mechanism to connect to (board) and disconnect from (disconnect from) agent A103, which is being used by user A105, in synchronization with user A105 boarding / alighting from the real vehicle. As long as this is achieved, the specific technical method may be implemented in other embodiments, including the one described above. 【0088】 Figure 4 is a sequence diagram showing an example of the process of connecting a user's agent to the dialogue agent system 100 according to this embodiment. The sequence in Figure 4 includes both the scene of connecting agent A103, as described in Figure 3, to the cyber vehicle, and the scene of disconnecting the agent in synchronization with user A105's exit. In the following description, user A105 will be used as an example, as in Figure 3, but the same applies to other users (such as user B108). In this specification and drawings, processing units in sequence diagrams and flowcharts are denoted by the letter S. In this specification, for example, "step S401" will not be written, but simply as "S401". 【0089】 First, the in-vehicle system 102 (or the app of the vehicle agent 101 that operates in conjunction with the in-vehicle system 102; the same applies hereinafter) asks the newly boarded user A105, "Do you want to connect with the agent?" to confirm whether user A105 wants to connect their agent A103 (or information terminal A104; the same applies hereinafter) to the cyber cabin (step S401). When user A105 affirms with "Yes" (S402), the in-vehicle system 102 confirms that user A105 intends to connect agent A103 (S403). 【0090】 The in-vehicle system 102 displays connection information for other agents to connect to the cyber cabin via the UI unit 215 of the in-vehicle system 102 (S404), and instructs user A105 to scan the displayed QR code using the application on the information terminal A104 with the message, "Please scan the QR code with your smartphone" (S405). 【0091】 User A105 uses an application installed on information terminal A104 to access Agent A and reads its QR code (S406). This allows the application to obtain connection information to connect to the cyber cabin (S407). 【0092】 If user A105 has multiple agents, the user selects the agent to connect to the cyber room by touching the agent's avatar (S408). This allows the app to identify the agent to connect to (in this example, agent A103) (S409). The app then connects agent A103 to the vehicle's cyber room, which allows interaction with other agents online (S410). 【0093】 When the in-vehicle system 102 detects that agent A103 has connected to the cyber cabin (S411), it displays the avatar of the connected agent A103 on the UI unit 215 (S412), notifies the information terminal A104 that it has connected via the cyber cabin service, and notifies user A105 that the connection is complete by saying "Successfully connected to the agent" (S413). 【0094】 Having connected with agent A103, the in-vehicle system 102 notifies user A105 that "From now on, we will consult with agent A and guide you" (S414). User A105 acknowledges this with "Thank you" and requests further processing (S415). Subsequently, the in-vehicle system 102 (or vehicle agent 101) and agent A103 (or information terminal A104) communicate and cooperate in the cyber cabin to provide user A105 with a more valuable travel experience (S416). 【0095】 When the detection unit 213, including an in-vehicle camera, detects that user A105 has finished moving, that the vehicle has arrived at its destination, parked, the in-vehicle system has stopped, or that user A105 has left the real vehicle compartment (S417), the in-vehicle system 102 disconnects the connection with user A105's agent A103 by disconnecting the agent from the cyber vehicle compartment or by terminating the connection to the cyber vehicle compartment itself (S418). 【0096】 Furthermore, for security reasons, connection information to the cyber cabin may use variable access destinations and authentication codes, or a fixed access destination and authentication code may be used for each vehicle to allow devices that have been connected before to easily connect again. 【0097】 Figure 5 illustrates an example of connecting an agent used by a user to the dialogue agent system 100 according to this embodiment. The difference from Figures 3 and 4 is that multiple users connect (simultaneously). Just as user A105 connected agent A103 to the cyber room via the application on information terminal A104, user B108 can do the same thing using information terminal B107 (agent B106), thereby enabling agent A103 and agent B106 to connect to the cyber room. 【0098】 Furthermore, the process of disconnecting Agent A103 or Agent B106 from the cyber bay is the same as explained in Figures 3 and 4, so it will be omitted here. 【0099】 In some cases, multiple users may be riding in a vehicle simultaneously and traveling together. Therefore, if each user's agent in the vehicle can be connected to the cyber cabin, it would be possible to recommend, select, and decide on routes, stops, in-cabin air conditioning, and content to play in the cabin, so that each user can have a better travel experience. For this reason, the above mechanism, which displays a QR code on the UI unit 215 and allows each user to connect their agent to the cyber cabin by scanning it, is considered to be extremely practical and effective in such new use cases. 【0100】 Furthermore, the UI section 215 of the in-vehicle system 102 displays the avatar of the connected agent when an agent connection is established. The agent's avatar may be displayed continuously while connected to the in-vehicle system 102 (or vehicle agent 101, or cyber cabin). This allows the agent's user to easily confirm that the agent has connected to the cyber cabin. 【0101】 By connecting your agent in this way, you no longer need to communicate or specify your preferences and specific requests to the in-car system yourself; you can leave it to your agent. This reduces the psychological burden of traveling by vehicle and has the advantage of increasing opportunities to gain new insights and discoveries, especially in unfamiliar places. 【0102】 Figure 6 illustrates an example of adding the dialogue agent system 100 according to this embodiment to a conversation group in which an agent used by a user participates. 【0103】 This embodiment differs in part from the embodiment for connecting an agent to the cyber vehicle described above (hereinafter referred to as Embodiment - Connection Process A). This embodiment (hereinafter referred to as Embodiment - Connection Process B) enables dialogue between agents by having the vehicle agent 101 (temporarily) join the SNS (Social Networking Service) group in which user A105's agent A103 is participating. 【0104】 The above-described embodiment—connection process A—involves a personal AI agent that corresponds to the user joining a virtual space that the vehicle agent 101 is either handling or already residing in, whereas embodiment—connection process B—differs in that it invites the vehicle agent 101 to join an SNS group that the personal AI agent is already a member of. 【0105】 In this explanation, we have described inviting vehicle agent 101 to an SNS group in which agent A103 is a member. However, this disclosure is not limited to this, and it may also apply to an SNS group in which user A105 is a member, but whose agent is not a member. In that case, user A105 sets agent A103 as their substitute before, during, or after adding vehicle agent 101 to the group. If there is already an SNS group that includes users riding in the real vehicle but does not include their agents, vehicle agent 101 is added to this group (temporarily). This method has the advantage of requiring less effort, as only one user, rather than all users, needs to operate the information terminal. 【0106】 In this context, "temporarily participating" means that the vehicle agent 101 is participating in the group under the condition that it will be automatically removed (exited) when one or more users riding in the real vehicle disembark, or when the vehicle arrives at its destination. This allows the vehicle agent 101 to participate in the SNS group only when necessary, which has the advantage of mitigating concerns such as information leakage. Furthermore, the setting of these temporary participation conditions may be provided by the SNS app, or the in-vehicle system 102 (or vehicle agent 101) may remove itself from the SNS group when the participation of vehicle agent 101 is no longer needed or is deemed inappropriate. 【0107】 First, in (I), user A105 instructs vehicle agent 101 to "show the SNS account code (of vehicle agent 101)" in order to have vehicle agent 101 join an SNS group that existing user A105 or agent A103 is a part of. In response, vehicle agent 101's avatar replies "yes" to acknowledge acceptance. For the sake of explanation, it is simply referred to as SNS here, but information that identifies which SNS it is could also be provided to vehicle agent 101. 【0108】 Next, in (II), the vehicle agent 101 displays its SNS account code information on the UI unit 215 of the in-vehicle system 102 and notifies user A105 with the message, "Here is my code." In response, user A105 launches the application on the information terminal A104 and reads the account code information from the vehicle agent 101. 【0109】 Next, in (III), the SNS app confirms whether to add vehicle agent 101 as a friend. If user A105 has no problems, touch Yes to add vehicle agent 101 as a friend. 【0110】 Furthermore, instead of adding Vehicle Agent 101 as a friend, you may register them as a temporary agent with conditions that allow their account to be disabled or excluded under certain circumstances. For example, if you register them as a "Vehicle Agent" instead of a "Friend," they may be registered as an agent that will be automatically disabled or excluded if certain conditions are met, based on the location relationship between the currently riding vehicle and the user, or the user agreement. 【0111】 Next, in (IV), a screen is displayed on information terminal A104 for selecting the SNS group to which vehicle agent 101 will join. Here, confirmation is required to add vehicle agent 101 to the group in which agents A103 and B106 are members. When user A105 indicates their intention to confirm by pressing the "OK" button (not shown), vehicle agent 101 is added to the target group. 【0112】 Next, in (V), it is displayed on the screen of information terminal A104 that vehicle agent 101 has joined an SNS group consisting of agent A103 and agent B106, and this is notified to user A105. User A105 then knows that vehicle agent 101 can now autonomously interact with agent A103. 【0113】 Next, in (VI), the UI section 215 of the in-vehicle system 102 displays the avatars of vehicle agent 101, agent A103, and agent B106, who are connected in an SNS group. User A105 gives the instruction "Navigate as a group," requesting vehicle agent 101 to select or update the best route for user A105 and user B108 (not shown) who are in the real vehicle, with the other agents in the group. In response, vehicle agent 101 replies, "Yes. From here on, we will consult as a group and guide you," and, while conversing with agents A103 and B106, responds that they will guide you along the most appropriate route. 【0114】 In this way, vehicle agent 101 can be temporarily added to an existing SNS group that users (or the agents used by those users) who are riding in a real vehicle are participating in. From the user's perspective, the more users who are connected, the less effort it takes to connect vehicle agent 101 with agent A103, etc. Also, since the interactions between agents are saved as chat history within a familiar SNS group, it is easy to check what discussions and conversations took place between agents, which has the advantage of reducing and resolving concerns and anxieties regarding agent statements and the provision of users' private information. 【0115】 Furthermore, the vehicle agent 101 (or in-vehicle system 102) may also notify user A105 (or user A105) of the conditions under which the vehicle agent 101 will be removed from the group, withdrawn, or have its account deactivated when the vehicle agent 101 joins a group that includes the user A105. In this case, user A105 can use the conversational agent system 100 with a clearer understanding of how long the vehicle agent 101 will remain in the SNS group. 【0116】 Figure 7 is a sequence diagram showing an example of the process from adding the dialogue agent system 100 according to this embodiment to a conversation group in which agent A103, used by a user, is participating, to excluding it. The sequence in Figure 7 includes both the scene in which vehicle agent 101, as explained in Figure 6, is added to the SNS group, and the scene in which vehicle agent 101 is disabled / removed from the group based on the account disablement / removal conditions. In the following explanation, we will use user A105 as in Figure 6, but the process is similar even if another user (such as user B108) performs the procedure. 【0117】 User A105 requests the in-vehicle system 102 (or vehicle agent 101) to display the account code of vehicle agent 101 in order to add the vehicle agent 101 to an existing SNS group (S701). In response, the in-vehicle system 102 (or its vehicle agent 101, hereafter the same) requests the SNS account code representing vehicle agent 101 from the in-vehicle system 102 (or its SNS app, hereafter the same) (S702). The in-vehicle system 102 (or its SNS app) receives the request and sends the account code of vehicle agent 101 to the in-vehicle system 102 (or its vehicle agent 101) (S703). The in-vehicle system 102 (or its vehicle agent 101) then displays the account code representing itself on the SNS app in the UI unit 215 (S704). 【0118】 Although this process was described as being performed based on a user request, it is also possible to automatically notify or display a message via the UI unit 215 when it detects that the user has entered a real vehicle. 【0119】 Next, user A105 operates the information terminal A104 (and its SNS app) to read and obtain the account code displayed on the UI unit 215 (S705). User A105 then registers the vehicle agent 101 as a friend on the SNS app (S706). At this point, one or more conditions can be set to disable or remove the registration of the vehicle agent 101 as a friend (or conversely, to enable or continue the registration). Since the registration of the vehicle agent 101 is for temporary use, especially in the case of shared cars, and there is no reason to continue registering the vehicle agent 101 in an SNS group that includes at least one or all of the users who rode in the real car after the vehicle is no longer in use, conditions to disable or remove the vehicle agent 101 from SNS friends and groups can be set at the time of registration. 【0120】 When vehicle agent 101 is added as a friend on the SNS app, the in-vehicle system 102 (and its SNS app) is notified of this fact via the SNS app's operating service (S707). In addition, the in-vehicle system 102 (and its vehicle agent 101) monitors this notification from the SNS app or detects it via API integration, thereby recognizing that vehicle agent 101 has been added as a friend on user A105's SNS app (S708). 【0121】 Next, user A105 adds vehicle agent 101, which they have registered as a friend, to a social networking group that user A105 (or agent A103) is a part of (S709). Here, as with the friend registration, one or more conditions may be set to disable or remove vehicle agent 101's registration from the group (or to enable or continue registration from the group). Note that the conditions for friend registration and group registration do not have to be the same. For example, friend registration may remain active unless explicitly instructed otherwise by the user, while group registration may be limited to the same or equivalent validity period as the vehicle usage contract period, or temporarily disabled while the user is out of the vehicle, with more detailed conditions applied. 【0122】 Once the registration process to the group is completed on the information terminal A104 (and its SNS app), the in-vehicle system 102 (and its SNS app) is notified of this via the SNS app's operating service (S711). Subsequently, similar to the friend registration process (S708), the in-vehicle system 102 (and its vehicle agent 101) is notified or detected that the user has been registered to the group (S712). 【0123】 Following the connection between vehicle agent 101 and agent A103, user A105 requests the in-vehicle system 102 (vehicle agent 101) to guide the user inside the vehicle while collaborating with other members of the registered group (including at least 0 people and at least 1 agent) (S713). In response, the in-vehicle system 102 (vehicle agent 101) sets its own policy to recommend or decide on vehicle guidance taking into account the preferences and interests of the registered group members (S714). Then, it begins to guide the user or design the user's travel experience in collaboration with the group members (S715). This embodiment in which vehicle agent 101 collaborates with other agents to guide the vehicle and design the travel experience of a user moving in a real vehicle will be described in detail later. 【0124】 After registration, the information terminal A104 (or its SNS app or SNS operating system) determines whether the conditions for disabling or deactivating the friend registration (S705) or group registration (S709) on the SNS have been met, and detects that the conditions have been met (S720). 【0125】 Information terminal A104 (or its SNS app, or SNS operating system) disables or removes the registration of vehicle agent 101 as a friend and / or group (S721). The fact of the removal is notified to or detected by information terminal A (or its agent A103) (the same process as S722 and S708, so the explanation is omitted), and is also notified to user A105 via the UI section 205 of information terminal A104 (or its SNS app) (S723). 【0126】 The fact that the account has been removed is also notified to the in-vehicle system 102 (and its SNS app) via the SNS app's operating service (S724). The in-vehicle system 102 (and its vehicle agent 101) is then notified or detected of this fact (the process is the same as in S725 and S708, so the explanation is omitted). In response, the in-vehicle system 102 (and its vehicle agent 101) notifies user A105 via the UI unit 215 that the vehicle agent 101 has been removed from user A105's SNS app or from a group within it (S726). 【0127】 The above describes an embodiment in which the user's agent enters and exits the cyber cabin in sync with the user's entry and exit from the vehicle (real cabin). Here, an embodiment was described, based on an example, that connects to a cyber cabin that allows for easy interaction in order to enable coordinated operation between the vehicle agent representing the vehicle and each agent representing each user in the real cabin, and further, an embodiment that disables or disconnects when it is no longer needed. 【0128】 In this explanation, the method of connecting the user's agent in the vehicle to the in-vehicle system 102 (or vehicle agent 101) was described using online meeting room access information or SNS account information. However, the connection condition could also be that the devices on which the connecting agents operate (information terminal A104, information terminal B107, and in-vehicle system 102) are less than a predetermined distance apart, using a positioning method such as GPS. It is believed that the security of the connection can be enhanced by making such a positional relationship a connection condition. 【0129】 Alternatively, if information terminal A104 or information terminal B107 is aware of the access information to the in-vehicle system 102 (vehicle agent 101), it may detect that it is sufficiently close to the in-vehicle system 102 using short-range wireless communication (WiFi, Bluetooth, NFC, etc.) or positioning technology such as UWB or GPS, and use this as a trigger to display a connection confirmation on the screen of the UI unit 205 of the information terminal, or notify the user by voice or vibration. Similarly, it may detect that the distance between the vehicle and the information terminal has increased beyond a predetermined distance, and use this as a trigger to request confirmation to disconnect via the UI unit 205, or if conditions such as the distance exceeding a predetermined distance or the state continuing for a predetermined time are met, the connection may be automatically disconnected. 【0130】 [Collaborative review and selection of recommended candidates among agents in the cyber cabin] From here, we will specifically explain, using Figures 8 to 11, Embodiment 1 of the agreement-building process and operation process outlined in Figure 2B, which involves collaboration between the vehicle agent 101 and the connected agent regarding vehicle movement, various joint discussions, the formulation of recommended proposals for the user, and the operation process related to vehicle movement in response to the adopted recommendation (proposal). 【0131】 Figure 8 is a partially flowchart-style sequence diagram showing an example of the processing flow in the dialogue agent system 100 related to consensus formation process A and operation process A of this embodiment, in which the vehicle agent narrows down the candidates to recommend to the user using each user's agent, presents them to the user for a decision on acceptance or rejection, and reflects that decision in the in-vehicle system. 【0132】 First, a user (for example, user A105) requests the in-vehicle system 102 (or vehicle agent 101) to provide a recommendation regarding a certain matter (S801). 【0133】 These matters are diverse, but all involve the kind of research, analysis, and comparison that require considerable effort from the user to determine which single option is best or most suitable for their situation at that particular time, from among numerous candidates that include both those within and outside the user's knowledge. 【0134】 Typical examples include route selection from the current location to the destination, tourist destinations that users want to visit, food and beverages or restaurants that users want to eat / drink, shops that sell products that users want to buy, parking spaces that users want to use at intermediate stops / destinations, music content that users want to listen to, and video content that users want to watch. Because there are so many options, it is difficult to recognize all of them without missing anything, and it is also difficult to compare and evaluate them all without omission. These are all things that often happen when traveling by vehicle. 【0135】 Furthermore, in practical terms, it would be desirable to check for each candidate whether it is currently available (e.g., restaurant opening hours, traffic congestion, and parking availability), whether all users experiencing that candidate together are interested in it, and what the cost of that candidate is (e.g., driving time and cost for each route option). However, this would require extensive and painstaking research. Therefore, finding candidates that satisfy users while driving or traveling has not been a practically feasible solution until now. 【0136】 While advancements in generative AI technology have made it possible to interact with AI in a chat format using prompts, it has not yet been achieved to instantly narrow down and propose recommended options for a given matter, especially when multiple users are riding together in a real vehicle, taking into account not only the individual preferences and interests of the users but also real-time changing traffic conditions. One of the main reasons for this is the significant effort required to prepare the input data for the generative AI. Considering whether the options align with each user's preferences and interests, and whether the facilities will be available by the time of arrival at the destination, by creating prompts for each potential destination and discussing them with the AI ​​in a chat format remains time-consuming and impractical. The dialogue agent system according to this embodiment solves this problem using generative AI. It aims to instantly evaluate a large number of options for a given matter, taking into account real-time information inside and outside the vehicle, and select the optimal candidate for the user, presenting it to the user. This will make the user's travel experience richer and more full of new discoveries than ever before. 【0137】 Upon receiving a request from a user, the in-vehicle system 102 identifies the requester (one of the users inside the vehicle) and recognizes the content of the request based on their speech (S802). For example, the requester may be identified by analyzing lip-sync with the facial image at the time of speech, estimating the speech position using a microphone array, or by voiceprint identification. The content of the request may also be determined by speech recognition if it is a verbal request. The purpose of identifying the requester is to notify the agent connected to the cyber vehicle that who in the real vehicle made the statement. Specifying the requester and the content of the request as a pair increases the amount of information input to the agent, increasing the likelihood of obtaining an appropriate response from the agent. It should be noted that identifying the requester is not mandatory; for example, only the text of the speech-recognized result indicating the content of the request may be sent to the agent connected to the cyber vehicle. 【0138】 Upon recognizing the request, the in-vehicle system 102 notifies (sends) the vehicle agent 101 of the requester's identification information and the request details (e.g., speech-recognized text data) (S803). The notification may be sent via a virtual conference room on the internet (i.e., a cyber room) where participants can share their video, audio, and entered text, similar to an online meeting. Alternatively, the notification may be sent using a mechanism similar to that of an SNS group, where participants belonging to a specific SNS group share their video, audio, and entered text among the group members. 【0139】 Upon receiving the notification, the vehicle agent 101, together with the agents connected to the cyber cabin, generates recommended candidates for the matter requested by the user (S804). This step is performed simultaneously with the step in which the agent who received the request for consideration from the vehicle agent 101 collaborates with the vehicle agent 101 to create a proposed recommendation (S805). Details of these processing steps S804 and S805 will be described later. 【0140】 Here, the term "cyber cabin" is used to mean a digital virtual space where two or more agents are connected via network 109, and is not limited to specific connection forms such as online meeting rooms or SNS groups as described above. 【0141】 The vehicle agent 101, having generated recommended candidates through joint review with the agent, presents one or more of these recommended candidates to the real vehicle interior user via the UI unit 215, along with the reasons for recommending them (S806). Providing the reasons for the recommendation is not mandatory at this stage. This step will be explained in more detail later. 【0142】 The user confirms this recommended candidate via the UI unit 215 (S807). If the user does not accept the recommended candidate, the vehicle agent 101 is instructed to display the next recommended candidate (arrow returning from S807 to S806), and the vehicle agent 101 presents the next best recommended candidate to the user with reasons (S806), and the user confirms (S807), and this process is repeated. 【0143】 In this way, the user makes a final decision on one of the candidates recommended by the vehicle agent 101 and notifies the vehicle agent 101 of it (S808). The vehicle agent 101 then instructs the in-vehicle system 102 to configure or change the settings according to the decision (S809). 【0144】 The in-vehicle system 102, when instructed to set or change something, executes it (S810). Typical examples of settings or changes that the in-vehicle system 102 is instructed to set or change by the vehicle agent 101 include routes and parking lots managed by the navigation system, content played in the vehicle cabin, lighting and air conditioning in the vehicle cabin, and, in the case of autonomous vehicles, the autonomous driving mode (for example, sport driving mode, eco driving mode with low environmental impact, low-cost driving mode to reduce travel costs to the destination, and automatic tracking mode that continues to follow the vehicle in front). Furthermore, autonomous driving is not limited to LV5 fully autonomous driving. It also includes modes below LV5, such as lane keeping and cruising modes where autonomous driving is disengaged when the brakes are applied, and modes that allow autonomous driving only on expressways. It also includes auto valet parking, which automatically parks the vehicle in a parking lot from near the destination. 【0145】 For example, if a new stop (tourist attraction or restaurant) is added, the vehicle agent 101 sends a request to the navigation system to add a waypoint. Upon receiving this, the navigation system adds the new waypoint, recalculates the route, and provides navigation instructions according to this new route. Alternatively, if it is an autonomous vehicle, the calculation unit 212 drives the control unit 216 according to the new route, causing the vehicle to automatically drive along the new route. A specific example of this step will be described later. 【0146】 In this way, the vehicle agent 101 and each agent connected to the cyber cabin jointly consider recommended options for matters specified by the user, and present these recommendations to the user along with the reasons for them. This makes it easier for the user to make more appropriate or suitable choices and decisions from a vast number of unknown options. As a result, the conversational agent system 100 can realize a service that enhances the value of the user's travel experience. 【0147】 Figure 9 shows an example in the dialogue agent system 100 according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this is reflected in the in-vehicle system 102 (navigation system). This is a concrete example of the processing flow described in Figure 8. 【0148】 First, in (I), user A105 asks the in-vehicle system 102 (or vehicle agent 101) "What restaurant do you recommend?" In response, vehicle agent 101 replies "I will look into it" and works with an agent connected to the cyber cabin to select recommended candidates. 【0149】 Next, in (II), once the selection of recommended candidates is complete, the vehicle agent 101 notifies the user in the real vehicle interior that "there is a popular ramen restaurant 5 km ahead along the route," and presents the ramen restaurant's cooking video and route map to the restaurant via the UI unit 215. To select these recommended candidates, the vehicle agent 101 and agents A103 and B106, which are connected to the cyber vehicle interior, extract and evaluate restaurants that best match multiple conditions and organize the information as recommended candidates. 【0150】 A "restaurant that better matches multiple conditions" refers to a restaurant that better satisfies two or more conditions, such as being located in a place that can be visited from the currently set route, being open at the current time or the scheduled arrival time, being able to make a reservation at the scheduled arrival time, being determined by the agent connected to the Cyber ​​Cabin to be a good match for the preferences of the user being served, being determined by the agent connected to the Cyber ​​Cabin to be a place of interest that the user is registered in their map or social media information, having a high rating from users, being featured in popular media, or being a particularly famous restaurant or cuisine in the area including the current location. 【0151】 These various conditions can be broadly categorized as follows: the recommended candidates must be close to the user's current location and available; they must align with the user's preferences and interests within the vehicle; and they must have a high rating from third parties. If it is difficult to evaluate whether the candidates align with the user's preferences and interests within the vehicle, the degree of match between the user's preference and interest information, which is accessible to the agent used by the user, and the individual candidate candidates may be quantitatively evaluated. 【0152】 In response to a ramen restaurant recommended by vehicle agent 101, user A105 asks, "Is it crowded?" inquiring about the current crowding situation at the restaurant. This illustrates a scenario where the user is confirming information missing from the candidates recommended by the dialogue agent system 100 before deciding on a particular candidate. 【0153】 Next, in (III), in response to the question, vehicle agent 101 checks the restaurant's congestion status (via an online congestion status service) and replies, "It seems to be relatively empty right now." Following this, user B108 expresses their intention to stop by the recommended ramen restaurant, saying, "That's fine." 【0154】 Next, in (IV), the vehicle agent 101, having received the user's final decision, responds, "Okay, I'll take you to the restaurant," and notifies the user inside the vehicle that it will guide them to the ramen restaurant. User A105 hears this and responds affirmatively, "Thanks." 【0155】 Next, (V) shows the in-vehicle system 102 (its navigation system) guiding the vehicle to the chosen ramen restaurant. This means that within the in-vehicle system 102, processing (information display in the UI unit 215) has been handed over from the vehicle agent 101 application to the navigation system application. 【0156】 In this example, the display on the UI section 215 of the in-vehicle system 102 switches to the navigation system to guide the route. However, the vehicle agent 101 application and the navigation system application may run simultaneously on the in-vehicle system 102. The guidance could continue using the voice of the vehicle agent 101, and the navigation system could refer to and display a map of the area near the current location shown by the navigation system while guiding the route to the ramen restaurant. 【0157】 Deciding which restaurant to stop at while traveling by car requires considering various factors, as described above, which is practically very time-consuming and cumbersome. Therefore, it is common for users to make decisions based on limited information. However, with this service, an agent that understands the user's preferences and can autonomously consider options on their behalf, and a vehicle agent that understands the vehicle and recommended options, working together to provide users in the actual car with easy-to-understand recommendations, it is believed that users will be able to make optimal choices not only for restaurants to stop at during their travels, but also for tourist destinations and routes, which were previously impossible. 【0158】 Furthermore, the conversational agent system 100 is highly convenient not only because it selects and proposes recommended candidates based on the user's request, but also because it can conduct additional research to gather supplementary information for the user's final decision and can instruct and reflect the final decision in various functions of the vehicle. As in the example above, the fact that the user does not have to go through the troublesome process of manually operating the navigation system to add the ramen shop as a waypoint after deciding to stop by the ramen shop is effective not only in improving the travel experience, but also in leading to changes in user behavior based on the information provided. 【0159】 While this explanation describes how recommended candidates are suggested in response to user inquiries, this disclosure is not limited to this. For example, if the user's agent determines that there is a place in the vicinity of the user's current location that is very well matched to the user's preferences and interests, in an area where the user has not visited many places, the agent may notify the user (or vehicle agent 101) accordingly. 【0160】 When an agent notifies vehicle agent 101, the agent may ask vehicle agent 101 to notify the user, providing the name of the location, a brief description of the location, and an introductory image (or a link to it). Upon receiving this, vehicle agent 101 notifies the user via UI unit 215 that the location is nearby, based on the information received. 【0161】 Furthermore, the voluntary provision of information by the agent mentioned above may only occur when the user visits an area where they have no or little visit history, or the user may set in advance whether they want or do not to receive voluntary information from the agent, or the user may notify or request the agent in advance that they would like or are willing to receive such voluntary information. 【0162】 Figure 10 is a sequence diagram showing an example of the processing flow in the dialogue agent system 100 according to this embodiment, in which the vehicle agent 101 uses each user's agent to narrow down the candidates to recommend to the user, allows the user to decide whether to accept or reject them, and reflects that decision in the in-vehicle system (navigation system). 【0163】 Figure 11 is a diagram illustrating an example of a dialogue history between agents in the cyber vehicle interior of the dialogue agent system 100 according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this decision is reflected in the in-vehicle system (navigation system). 【0164】 The scenes shown in Figures 10 and 11 are the same as those explained in Figure 8, so we will use both figures to continue the explanation. Also, parts that overlap with previous explanations may be omitted. 【0165】 User A105 asks the in-vehicle system 102 (or vehicle agent 101, hereafter the same) "What restaurant do you recommend?" (S1001). The in-vehicle system 102 recognizes this as speech and understands the request, then generates questions to elicit candidates and requirements from the agent connected to the cyber cabin (S1002). The content of the conversation between the user and vehicle agent 101 may be shared with the agent connected to the cyber cabin each time (not shown in Figure 10; see the first row of Figure 11). 【0166】 This question uses natural language and could be something like, "@Agent A, @Agent B. If you have any restaurant recommendations for a place to stop by now, please provide the specific restaurant name, location, and reason for your recommendation. Or, please list your requirements." The "@Agent A, @Agent B" part at the beginning specifies (mentions) the agents who are expected to respond to this message. This allows Agent A103 and Agent B106 to detect that the following message is addressed to them and that some kind of response is required. 【0167】 The vehicle agent 101 that generated the above message sends this message to the cyber galley, or to agents A103 and B106 connected to the cyber galley, requesting responses regarding restaurant candidates and requirements (S1003). 【0168】 Upon receiving this message, information terminal A104 (agent A103) and information terminal B107 (agent B106) respectively extract restaurant candidates and requirements on behalf of the user and generate a response based on the message received from vehicle agent 101 (S1004, S1005). 【0169】 Agent A103 then sends the restaurant candidates and responses to the requirements to the cyber room, or to Agent B106 and Vehicle Agent 101 connected to the cyber room (S1006). 【0170】 This response uses natural language and might look something like this: "Restaurant A: XX Town, offers a wide selection of local dishes. Restaurant B: □□ Town, a popular ramen shop I recently checked out." In response to a message from Vehicle Agent 101, it recommends two restaurants, including the name, location, and reason for each recommendation. 【0171】 Furthermore, the recommendation reason for Restaurant A, "Offers a wide selection of local dishes," is an example of a recommendation reason found by Agent A103 through an online search for restaurants near their current location. Additionally, the recommendation reason for Restaurant B, "A popular ramen shop recently checked out," is an example of a recommendation reason based on User A's past activity history, as it is a ramen shop that User A105 recently learned about and registered on social media. 【0172】 Similarly, agent B106 sends responses to restaurant candidates and requirements to the cyber galley, or to agent A103 and vehicle agent 101 connected to the cyber galley (S1007). 【0173】 This response uses natural language, such as "I'd like a restaurant with healthy menu options," and addresses the general requirements for a restaurant to visit, rather than providing specific restaurant suggestions. 【0174】 Having received these responses, vehicle agent 101 searches the internet for other restaurant candidates to add (S1008). 【0175】 Furthermore, the databases and websites searched here may be provided by third parties. For example, candidates may be extracted from a restaurant search database / website provided by a third party. In this case, a special incentive arrangement may be made between the third party and the operating company of Vehicle Agent 101. In exchange for Vehicle Agent 101 suggesting restaurants near the user's current location from the third-party database, the operating company of Vehicle Agent 101 may receive advertising fees from the third party. In addition, to increase user interest, these restaurant recommendations may be accompanied by special discount coupons, allowing users to use them at a discounted price or with some kind of benefit. These coupons may be in a form that can be read by information terminals, for example, a QR code containing a specific URL. 【0176】 The vehicle agent 101, having collected restaurant candidates, organizes the information on these candidates so that agents connected to the cyber bay can evaluate them (S1009). When organizing the information, or before or after this process, the vehicle agent 101 may check online whether each restaurant candidate is available at the estimated time of the vehicle's arrival, and may keep only the available restaurant candidates for evaluation. 【0177】 Once the restaurant candidates are organized in this way, vehicle agent 101 requests the cyber compartment, or agents A103 and B106 connected to the cyber compartment, to evaluate the restaurant candidates (S1010). 【0178】 This request is, for example, an evaluation request using natural language, such as the following: 【0179】 "@Agent A, @Agent B. User A asked for restaurant recommendations, and I've narrowed it down to the following restaurants. Please rate these restaurants on a scale of 1 to 10 and provide a brief review." Restaurant A {Features}{Access}{URL} Restaurant B {Features}{Access}{URL} Restaurant C {Features}{Access}{URL} 【0180】 Restaurant A, B, and C are the names of the restaurants. {Features} is a brief description of the characteristics of the corresponding restaurant. {Access} is information indicating the travel time or distance from the current location or currently set route to the corresponding restaurant. {URL} is the URL of the website introducing the corresponding restaurant. 【0181】 Upon receiving this message, information terminal A104 (and its agent A103) and information terminal B107 (and its agent B106) generate responses that rate each restaurant on a scale of 1 to 10 and include brief comments, based on the received message (S1011, S1012). 【0182】 Agent A103 then sends the generated response to the cyber cabin, or to Agent B106 and Vehicle Agent 101 connected to the cyber cabin (S1013). 【0183】 This answer is, for example, a response using natural language, such as the following: 【0184】 "@Vehicle Agent" Restaurant A: 8 points. Good because it offers many local dishes. Restaurant B: 9 points. They have the popular ramen. Restaurant C: 4 stars. Expensive. 【0185】 Similarly, Agent B106 sends a response, including an evaluation of each restaurant candidate, to the Cyber ​​Room, or to Agent A103 and Vehicle Agent 101 connected to the Cyber ​​Room (S1014). 【0186】 For example, the answer is as follows: 【0187】 "@Vehicle Agent" Restaurant A: 6 points. Low ratings on social media. Restaurant B: 8 points. A popular restaurant located along the route. Restaurant C: 6 points. It's far from the route and takes a long time to get to. 【0188】 Both Agent A103's and Agent B106's responses begin with "@Vehicle Agent," which specifies Vehicle Agent 101, the agent expected to respond to this message. This allows Vehicle Agent 101 to detect that a response is clearly required for the following message. 【0189】 Upon receiving these responses, the vehicle agent 101 selects / organizes recommended candidates based on the agent's evaluation results (S1015). 【0190】 This compilation of recommended candidates may be shared with other agents in the cyber bay. For example, it may be organized using natural language as follows: 【0191】 "We will propose the following options in order of highest evaluation score." Restaurant B (★★★★☆) There's a popular ramen shop 5km along the route. {Additional Information} Restaurant A (★★★☆☆) They offer a wide variety of local dishes. {Additional Information} Restaurant C (★★☆☆☆) A high-end sushi restaurant. It's a bit far from the main route. {Additional Information} 【0192】 The ★ marks indicate a visual representation of the quantitative evaluation results described later, showing the quality of the evaluation between agents on a 5-point scale as an example. {Supplementary information} includes things like video information of the dishes used to introduce the corresponding restaurant, route information to the restaurant, or the restaurant's URL. 【0193】 Details regarding the selection of recommended candidates will be described later, but if evaluation scores are obtained, candidates with higher total scores may be recommended first. In addition, the priority of recommended candidates may be changed not only based on evaluation scores but also on the reasons for the recommendation. Furthermore, when the vehicle agent 101 presents recommended candidates to the user, it may include not only the name of the recommended candidate but also a concise explanation of the reason for the recommendation, and supplementary information such as photos and route information. 【0194】 Next, the vehicle agent 101 briefly presents a recommended option, such as "There is a popular ramen restaurant 5 km along the route," along with the reason for the recommendation. It then notifies the user via the UI unit 215 (S1016), along with a photo of the dish to supplement the recommendation and route information. This is the proposed form shown in Figure 9(II). This notification may also be shared with agents connected to the cyber vehicle (in this example, agents A103 and B106) (not shown in Figure 11). 【0195】 In response, user A105 asks vehicle agent 101, "Is it crowded?", expressing concern about the restaurant's crowding level (S1017). Following this, vehicle agent 101 obtains information from an external service regarding the restaurant's current crowding level, or its expected crowding level at the estimated arrival time (S1018). It then replies to user A105, "It appears to be relatively uncrowded at the moment" (S1019). 【0196】 In response, user B108 tells vehicle agent 101 that the restaurant is fine (S1020). Based on this, vehicle agent 101 decides to stop at the ramen restaurant mentioned above (S1021). Then, it sets the ramen restaurant as a waypoint in vehicle system 102 (its navigation system), or requests that it be set (S1022). This setting request may be implemented through API linkage from the vehicle agent 101's app to the navigation system's app. Furthermore, it informs the user that it will now guide them to the ramen restaurant (S1023). As a result, the user can find a suitable restaurant without having to search for one themselves. Also, once the restaurant is finalized, route guidance to that restaurant begins, providing a very smooth travel experience. 【0197】 For example, if the vehicle is an autonomous vehicle, such as a robotaxi or other driverless taxi, when the navigation system adds a new waypoint and sets a new route, the autonomous driving control system (one of the software / functions operating in the calculation unit 212; specifically, the driving assistance / autonomous driving control software 1908), which drives and controls the control unit 216 for autonomous driving, detects this, or receives a notification from the navigation system that the route has been updated, and drives and controls the control unit 216 according to the new route to drive the vehicle automatically. In this case, since no one in the real vehicle needs to drive, the user can travel to that location simply by deciding on the destination and waypoints while interacting with the conversational agent system 100. 【0198】 The table in Figure 11 shows the speaker's identification information, the time of their utterance, and the content of the communication (chat text) that took place within the cyber cabin. If the cyber cabin is in the form of an online meeting room, this can be automatically recorded like meeting minutes and shared among agents. If the cyber cabin is in the form of a social networking group, this chat content can be recorded as a conversation log for that group and shared among group members. Regardless of the implementation of the cyber cabin, users can later review what discussions and evaluations took place among the agents. This is useful because it allows users to later confirm whether the behavior of agents who provided information, opinions, suggestions, or evaluations on their behalf met their expectations, and whether they carelessly disseminated information that might cause concern to the user. 【0199】 Figure 12 illustrates a different embodiment (Embodiment - Consensus Formation Process B and Operation Process B) of the consensus formation process A and operation process A in the embodiments described in Figures 8 to 11 above. 【0200】 This flowchart shows an example of the process flow in the dialogue agent system 100 according to this embodiment, where the vehicle agent 101 uses each user's agent to narrow down the candidates to recommend to the user and allows the user to make a decision. 【0201】 The main differences between this process and previous systems are that the vehicle agent 101 performs the following five steps. P1) Proceed with discussions with agents who understand user preferences. P2) When creating recommended candidates, collect recommended candidates from each agent. P3) Have each agent quantitatively evaluate each recommended candidate. P4) Aggregate the evaluation results from each agent and propose the top candidates to the user. P5) Recommended candidates should be presented to the user with the reasons for the recommendation and supplementary information. 【0202】 Unlike previous systems where a single computer system or program would run and complete the process, this system differs significantly in several ways: multiple computer systems and programs (i.e., agents) collaborate in their deliberations; vehicle agent 101 manages the deliberations and compiles the results; agents who know the preferences of real car users participate in the deliberations; and agents who know the user preferences propose options to the user, including the reasons for their recommendations. 【0203】 The flowchart in Figure 12 shows the processing of the vehicle agent 101 on the left and the processing of each user's agent connected to the cyber cabin on the right. The processing in this figure is initiated when the vehicle agent 101 is requested by the user to propose recommended candidates, or when it receives a spontaneous proposal from the vehicle agent 101 or another agent. Messages are also sent and received during the processing flow, but these are the same as the example messages described in Figures 10 and 11, so the explanation is omitted. 【0204】 First, the vehicle agent 101 generates a request message for each user's agent to submit proposed candidates or requirements for the current topic of consideration, and sends it to the cyber bay or each agent connected to the cyber bay (S1201; see S1002, S1003). 【0205】 Upon receiving this, the agent generates a response message regarding recommended candidates and requirements, referring to data on the user's preferences and interests (see S1202, S1004, S1005). Furthermore, the generated response message is sent back to the vehicle agent 101 or the cyber cabin (see S1203, S1006, S1007). 【0206】 When this is received, or when a predetermined time has elapsed since step S1201, the vehicle agent 101 aggregates the candidates and requirements recommended by each agent that it has obtained (S1204). Candidate aggregation is the process of listing the candidates obtained from the agents in a way that there are no duplicates. Requirements aggregation is the process of listing the requirements obtained from the agents while refining the expression to avoid semantic duplication or omissions. 【0207】 The processes from step S1201 to step S1204 correspond to P1) and P2) above. If information from social media accounts that the user follows or browsing / registration history on online services is available, this information can be used to provide more personalized suggestions to the user. This can be achieved if the user's agent is authorized to access the above information of the user. Alternatively, if there is quantitative information in the agent's settings that shows the user's preferences and level of interest in specific types / genres, that information can be used. 【0208】 Furthermore, the vehicle agent 101 does not generate the candidate population on its own, but rather collects data from agents connected to the cyber cabin as a substitute for the user to form the population. This increases the likelihood of recommending candidates that are highly satisfying because they align with the user's preferences and interests. In this way, suggestions based on the user's preferences and interests using agents can have higher quality and accuracy compared to suggestions based on evaluations by third parties who are completely unaware of the user's preferences and interests. 【0209】 Next, the vehicle agent 101 searches the network 109 for any other suitable candidates besides those acquired from the agent (S1205; see S1008). In this additional candidate search, the vehicle agent 101 may also search for and add new candidates from a database provided by a third party, as explained in step S1008. Detailed local event information and campaign information provided by third parties such as travel agencies and local organizations is undoubtedly beneficial to users and has the potential to create unique encounters. Furthermore, it provides the operating company of the vehicle agent 101 with an opportunity to earn advertising revenue, increasing the likelihood of sustained service provision. 【0210】 Vehicle agent 101 checks via network 109 whether all collected recommended candidates are available around the estimated time of vehicle arrival (S1206; see S1009). If there are no available candidates, or if the number is less than a predetermined number, the process proceeds to "No" and additional recommended candidates are searched for and added via network 109 (S1205). Vehicle agent 101 performs the search based on the list of requirements collected from the agent. On the other hand, if the number of available candidates is greater than or equal to a predetermined number, the process proceeds to "Yes". 【0211】 Next, vehicle agent 101 generates a request message for each recommended candidate collected so far, asking them to provide a quantitative evaluation and a brief comment explaining the reasons for the evaluation, and sends it to the cyber bay or each agent connected to the cyber bay (S1207; see S1010). 【0212】 The reason for requesting quantitative evaluations here is for the sake of simplicity, objectivity, and fairness in the aggregation process. Technically, it seems possible for vehicle agent 101 to rank priorities based on qualitative evaluations, and such a format is also acceptable. However, in that case, concerns cannot be eliminated that vehicle agent 101's intentions and biases may influence the evaluation. Therefore, we will mainly explain a method for determining the recommendation priority in a way that can be simply calculated from the evaluations of agents acting as substitutes for the user, without vehicle agent 101 participating in the evaluation. 【0213】 While a brief comment is not strictly required, if one is obtained, it can be presented as supplementary information in step S1211, as described later, to assist the user in making a decision. 【0214】 Upon receiving an evaluation request message, the agent provides a quantitative evaluation and brief commentary for each candidate, referencing the user's preferences and interests, and creates an evaluation result that includes these (S1208; see S1011, S1012). Furthermore, each agent sends a message containing the evaluation result they have created back to the cyber cabin or vehicle agent 101 (S1209; see S1013, S1014). 【0215】 Vehicle agent 101, having received evaluation result messages from each agent, aggregates the evaluation results from each agent (see S1210 and S1015). The method for aggregating the evaluation results can be as simple as described above. In the example above, if each recommended candidate is quantitatively evaluated on a scale of 1 to 10 and a brief comment is given for that candidate, the total score obtained by each agent who responded for each recommended candidate on a scale of 1 to 10 can be used as the quantitative evaluation result for that candidate from each agent. Candidates with higher scores in this quantitative evaluation result can be given higher priority and recommended to the user. 【0216】 Furthermore, if the evaluation scores for recommended candidates are invalid, such as not being provided, you may use the average valid evaluation score, which is calculated by summing the valid evaluation scores and dividing them by the number of agents who provided valid evaluation scores. For example, you may recommend candidates with higher average valid evaluation scores to users as higher-priority candidates, in descending order of average valid evaluation scores (highest priority). 【0217】 Furthermore, if a recommended candidate receives significantly low ratings, such as 3 out of 10 or less, from one or more agents, it may indicate that the user of that agent strongly dislikes that candidate. In such cases, you may consider lowering the priority of that recommended candidate or even removing it from the list of candidates recommended to users. 【0218】 Furthermore, if the brief comments include strong affirmations or negativity, the vehicle agent 101 may adjust the priority accordingly. For example, if a restaurant serves many foods that the user is allergic to, the agent might write in the brief comments that "many dishes contain allergens." In this way, in addition to preferences and interests, if there are constraints related to the user's health (allergies, calories, etc.), beliefs (religion, etc.), or creed (vegan, etc.), the vehicle agent 101 may change or set the priority not only based on the quantitative evaluation results but also based on the content of the brief comments, or even prioritize the brief comments over the quantitative evaluation results. 【0219】 The process from step S1207 to step S1210 corresponds to P3) above. Except in cases where there are strong restrictions regarding health, faith, or beliefs as described above, a quantitative evaluation can be requested and the scores can be simply tallied to ensure fairness. 【0220】 Following the determination of the priority of recommended candidates, the vehicle agent 101 proposes the candidate with the highest evaluation from the remaining candidates to the user via the UI unit 215 (S1211; see S1016). In this proposal, the reasons for recommending the candidate may be generated based on the agent's comments and added as supplementary information. Similarly, video information, audio information, current location and distance from the route, and evaluation comments from third parties may also be added as supplementary information to introduce the recommended candidate. 【0221】 For example, if the topic of consideration is a tourist destination, the proposal may include one or more of the following: the name of the tourist destination, its characteristics, and a video of the tourist destination. If the topic of consideration is a restaurant or retail store, the proposal may include one or more of the following: the name of the store, the characteristics of the products / services the store handles, and a video of the products / services the store handles. If the topic of consideration is video / audio content, the proposal may include one or more of the following: the name of the content, the characteristics of the content, and a video illustrating the content. If the topic of consideration is a route, the proposal may include one or more of the following: the estimated travel time for the route, the estimated toll fees, the characteristics of the route, and the route displayed on a map. Furthermore, for each characteristic included in the proposal, a brief comment that matches the user's interests based on a brief review, or a comment including a third-party evaluation, may be included. 【0222】 The vehicle agent 101 (or in-vehicle system 102) detects the user's response via the detection unit 213, which may be one or more of the following: verbal statements, touch operations, facial expressions, gestures, etc., and determines whether the user has decided to adopt at least one of the currently proposed recommended candidates (S1212). If a negative response from the user or a response requesting other recommended candidates is detected, the system proceeds to "No" and returns to step S1211, proposing the next best recommended candidate. On the other hand, if the system detects that the user has decided to adopt (or select) one of the currently proposed recommended candidates, the system proceeds to "Yes". 【0223】 When the vehicle agent 101 detects the above decision made by the user, it determines that the user has made a final decision on the recommended candidate selected by the user (S1213) and terminates the process. 【0224】 The processes from step S1210 to step S1213 correspond to P4) and P5) above. Based on the agent's evaluation results, the priority of recommended candidates is determined, and the candidates with the highest priority are presented to the user along with the reasons for the recommendation. This allows the user to easily understand the appeal of the recommended candidates and easily make a final selection of one candidate. 【0225】 From here, we will explain the details of the information leakage countermeasures for the consensus-building process in the embodiment shown in Figure 2A. We will also explain the details of user data and how it is handled. 【0226】 [Measures to prevent information leakage to agents] Up to this point, we have described examples of connection configurations or systems that allow vehicle agent 101 and each agent connected to the cyber room to view common information. With this connection configuration, information symmetry is ensured, and discussions and evaluations among agents can be conducted with transparency. However, since each agent can hold or refer to user preference information and behavioral history information, there is a concern that an agent may unilaterally transmit private information that the user does not want to disclose to the cyber room or other agents connected to the cyber room. Therefore, from here on, we will specifically describe an embodiment of measures to prevent information leakage among agents connected to the cyber room. 【0227】 Figure 13 is a sequence diagram showing an example of the processing flow in the dialogue agent system 100 according to this embodiment, in which the vehicle agent uses each user's agent to narrow down the candidates to recommend to the user, the user decides whether to accept or reject them, and this decision is reflected in the in-vehicle system (navigation system). 【0228】 Figure 13 shows the exact same scene as Figure 10 and is similar to an example of the processing flow of the dialogue agent system 100 shown in Figure 10; therefore, only the differences will be explained. The processing steps that differ are those indicated in the S1300 series. 【0229】 The process from step S1302 to step S1307 involves collecting candidates / requirements for a discussion topic provided by the user's agent. In Figure 10, the cyber room is described as an online meeting room or a group chat on social media, where all participants can access the same information. On the other hand, in Figure 13, the vehicle agent 101 and each agent exchange information through individual chat-style interactions or via a predetermined API, ensuring that private information contained in each agent's response is not shared with other third-party agents. It should be noted that the cyber room in this embodiment, where the vehicle agent 101 and each agent exchange information through individual chat-style interactions or via a predetermined API, i.e., a cyber room where two parties can converse, is an example of a cyber private room. 【0230】 Vehicle agent 101 added "@Agent A" to the candidate / requirement question message to indicate that it was a direct message (mention) to agent A. If vehicle agent 101 determines that there is a possibility of information being exchanged between agents connected to the cyber cabin that may be of concern to be shared, it will send a request message to the agent individually. For example, if the request message is for agent A103, it will generate a message that does not mention other agents, such as "@Agent A. If you have any restaurant recommendations for a place to stop by now, please give the specific restaurant name, location, and reason for recommendation. Or, please give your requirements." (S1302). 【0231】 Vehicle agent 101 sends a request message generated individually to each agent connected to the cyber room (S1303). Alternatively, it sends a request message to each agent's endpoint via API. Note that the request message itself does not contain any information that would infringe on privacy, so this process may also be carried out in a way that shares the same request message to each agent connected to the cyber room, as in steps S1002 and S1003. 【0232】 Next, agents A103 and B106, who received the request message, extract candidates / requirements for the topic under consideration based on the request message (S1004, S1005). This process is the same as in Figure 10. 【0233】 Next, Agent A103 and Agent B106 each send their generated response messages back to Vehicle Agent 101. These response messages may contain private information such as health, beliefs, creed, preferences, and behavioral history. Therefore, each agent sends a reply only to Vehicle Agent 101, who made the request (S1306, S1307). 【0234】 In this way, when collecting candidates / requirements for a topic under consideration, it is possible to prevent any user's private information from being leaked to third-party agents other than vehicle agent 101 and the agent in question. Users can also confirm that this mechanism is in place through chat history in the Cyber ​​Room, allowing them to use the service with peace of mind. 【0235】 The same process as described above will be used when evaluating potential restaurants. 【0236】 Step S1309, which involves organizing the restaurant candidates, differs from step S1009 in that, as explained in step S1302, only the agent sending the request message is mentioned within the request message. 【0237】 Step S1310 of the candidate evaluation request differs from step S1010 in that, as explained in step S1303, a request message is sent to each individual agent. 【0238】 Steps S1313 and S1314, in which the agent replies with an evaluation, differ from steps S1013 and S1014 in that, as explained in steps S1306 and S1307, the reply is only sent to the vehicle agent 101 who made the request. 【0239】 Figures 10 and 13 depict the same scene and almost the same processing flow, but by switching to individual communications when there are privacy concerns regarding information communication between agents, the risk of information leakage to unrelated third-party agents is eliminated. 【0240】 This scenario assumes the use of partner-type agents in daily life that can store or access user preference information, interest information, and behavioral history information for each user. Therefore, in a situation where multiple agents are connected, for example, in a cyber cabin and share all information, there is a risk that private and sensitive information, such as user preferences and beliefs, could be shared without the user's consent. This is a problem that has not existed until now. 【0241】 A problem arises when private or sensitive information, such as one's preferences or beliefs, is transmitted to another person's agent. Neither the user who owns the information nor the agent of the user who transmitted it can control how that other agent handles that information. Therefore, in this embodiment, when multiple (three or more) agents exchange information and handle a user's private or sensitive information, communication is limited to the minimum necessary number of agents (e.g., two), excluding third-party agents. 【0242】 Furthermore, this embodiment describes a case in which an agent is used while a user is traveling in a vehicle. Therefore, the vehicle agent 101, which is an agent that substitutes for the vehicle, is in a position to facilitate and coordinate discussions with the user's agent connected to the cyber cabin, and this is a different position from the user's agent. For this reason, in this embodiment, when it is assumed that such private or sensitive information will be handled, the vehicle agent 101 will not share the information with all agents but will exchange information individually with each agent. 【0243】 Figure 14 is a flowchart illustrating an example of the process flow in the dialogue agent system according to this embodiment, where the agent switches the recipient of information based on the nature of the information when sending information. 【0244】 This processing flow applies when an agent transmits information to other agents connected to the cyber cabin (i.e., an online space where multiple agents can exchange information). Therefore, it applies to vehicle agent 101, agent A103, and agent B106. 【0245】 An information-transmitting agent determines whether or not the information it intends to transmit contains private or sensitive information of an entity it represents or substitutes for, or quantifies the degree to which it contains such information (S1401). Here, the entities it represents or substitutes for are the in-vehicle system 102 representing the vehicle, agent A103 representing user A105, and agent B106 representing user B108. 【0246】 An in-vehicle system 102 representing the vehicle may determine that information regarding the vehicle owner's personal information, accident history, and driving history (such as where the vehicle was driven) (described in the vehicle data file) is sensitive information (or private information) for the vehicle. On the other hand, it may determine that information such as the vehicle's speed, destination, route to the destination, and fuel / battery level is not sensitive information. This determination may not be a simple yes / no, but rather a quantitative determination on an individual basis. 【0247】 Furthermore, an agent acting on behalf of a user may determine that the user's personal information (such as name, address, telephone number, date of birth (age), etc.), biometric information (such as gender, health status, medical history, etc.), religious information (such as religion, etc.), and belief information (such as strong restrictions on diet, politics, culture, etc.) are private or sensitive information (described in the user data file). On the other hand, the agent may determine that the user's preferences for food genres, areas / matters of high interest, and matters already shared with the information recipient are not private or sensitive information. This determination may not be a simple yes / no, but rather a quantitative determination on an individual basis. Moreover, this quantitative determination may take into account the relationship between the information recipient and the user. 【0248】 Next, the information-transmitting agent determines whether the information to be transmitted contains private information or sensitive information, or whether the result of quantifying the degree to which such information is included is above a predetermined value (S1402). 【0249】 If the result of this determination is "Yes," the information-transmitting agent will send the information only to the receiving agent in order to handle the information individually (S1403). If there are multiple receiving agents, the information should be sent to each of them individually. Individual communication can be achieved by the information-transmitting agent and the receiving agent exchanging information via a communication method shared only between the two parties. 【0250】 For example, the agent's endpoint (such as an API URL) could be disclosed, and information could be exchanged by sending and receiving text messages using communication protocols such as HTTP requests to that endpoint. Alternatively, just like humans, the agents could communicate in a chat room accessible only to both parties using a social networking service. This individual communication method is not limited to any specific communication method or technique, as long as it allows both parties to exchange information directly without sharing information with third parties. 【0251】 If the determination in step S1402 is "No", the agent transmitting the information does not need to handle the information individually, and therefore transmits the information in a format that can be viewed by all agents connected to the cyber room, including the agent receiving the information (S1404). 【0252】 This form, for example, means that if the cyber room is in the form of an online meeting room, all participants can acquire and view the information, such as comments and chat texts in the online meeting room. Similarly, if the cyber room is in the form of a social networking group, all members of the group can acquire and view the information as a conversation log. This embodiment does not limit the form of information sharing, as long as it is shared and shared among agents that satisfy a predetermined relationship, such as those connected to the cyber room. 【0253】 In this way, the agent providing the information determines the nature of the information. If it determines that the information is private or sensitive, it sends it individually to the recipient agent. If it determines that the information is not private or sensitive, it sends it in a format that can be viewed by all relevant agents. When sending information publicly, it is desirable to send it in a format that can be shared with everyone unless there is a reason to send it individually, as this is useful in terms of transparency, symmetry, and simultaneity of information among all relevant agents. 【0254】 Figure 15 shows an example in the dialogue agent system 100 according to this embodiment, in which a vehicle agent responds to a user's privacy inquiry by communicating individually only with the user's agent. 【0255】 There may be situations where a user wants to communicate about privacy-related matters with their agent while traveling in a vehicle. This diagram illustrates a scenario where vehicle agent 101 dynamically switches its connection destination to communicate individually with the user's agent. After that interaction is complete, the vehicle agent resumes its conversation with the user, sharing information with other agents as before. 【0256】 First, in (I), there are users A105 and B108 in the real vehicle cabin, and user A105 is asking the in-vehicle system 102, "When was the last time I was here?", indicating when they last arrived at their current location. The UI section 215 of the in-vehicle system 102 displays the avatars of vehicle agent 101, agent A103, and agent B106, all connected to the cyber cabin. For example, in this state, the conversation between vehicle agent 101 and user A105 or user B108 in the real vehicle cabin, and the conversation between agents connected to the cyber cabin, are shared among the agents as text data. The important point is that both agent A103 and agent B106 have access to exactly the same information, and there is no information asymmetry between them. 【0257】 Next, in (II), the vehicle agent 101, having received a question from user A105, determines that the question concerns user A105's activity history and therefore involves private information. Then, in order to proceed to step S1403, it sends the question message to agent A103 individually. In the UI section 215 of the in-vehicle system 102, pictures and marks are displayed to the user in the real vehicle (user A105) to show that the vehicle agent 101 is exchanging information individually with agent A103. As an example, here the avatar of agent B106 is removed from the UI section 215, and a bidirectional arrow is displayed between the vehicle agent 101 and agent A103. 【0258】 If Agent B106's avatar is not displayed in the UI section 215, it becomes difficult to determine whether the system is connected to the cyber vehicle. Therefore, it may be possible to gray out only Agent B106 and continue to display it, or to display additional information indicating that Vehicle Agent 101 and Agent A103 are individually connected. For example, additional information such as Vehicle Agent 101 and Agent A103 talking on a string telephone may be displayed. 【0259】 Next, in (III), vehicle agent 101 receives a response from agent A103 and tells user A105, "I came here three years ago." Since vehicle agent 101 and agent A103 continue to exchange information individually, the same avatar connection status as in (II) is displayed in the UI section 215. 【0260】 Furthermore, the questions that user A105 asked about their own activity history were audible to user B108, who was in the same real-world vehicle, and user B108 also heard vehicle agent 101's response. In other words, it could be said that the private information that user A105 had been near their current location three years ago was leaked to user B108. However, since user A105 asked the question while user B108 was in the same real-world vehicle, it can be assumed that user A105 had no concern, or a very low concern, about the answer being conveyed to user B108. 【0261】 Rather, the serious data breach for the service providers of user A105, agent A103, or vehicle agent 101 is not against user B108, but against agent B106, which user B108 uses. Because agent B106 is autonomous software, if it acts maliciously, or even if it malfunctions without malicious intent, it could potentially damage the reputation and credibility of the service providers of user A105, agent A103, or vehicle agent 101 in cyberspace. Furthermore, there is concern that information once known to agent B106 may be learned by agent B106 or recorded in its learning database. 【0262】 For example, agent B106 might use user B108's social media account to widely share the information it obtained with an unspecified number of people. If this happens, there is a risk that the information will remain on the internet indefinitely. 【0263】 To address the new challenges described above, this embodiment proposes a mechanism for switching the flow of information between agents according to the nature and content of that information. Furthermore, it proposes to visualize the status of this information flow in an easy-to-understand manner and present it to the user. 【0264】 Next, in (IV), user A105 asks a question about traffic conditions, "How much further to my destination?", which is unrelated to anyone's privacy information. The in-vehicle system 102 recognizes this question from user A105 and forwards it to the vehicle agent 101. 【0265】 Next, in (V), the vehicle agent 101, which was asked the question, obtains real-time traffic information from a third-party service via the network 109 and responds with an estimated arrival time at the destination: "It looks like we will arrive in about 30 minutes." 【0266】 Vehicle agent 101 determines that this exchange regarding the estimated time of arrival at the destination does not contain anyone's private or sensitive information, and the process in step S1404 is carried out. Therefore, this exchange is shared among all agents connected to the cyber cabin, and to indicate this, agent B106's avatar is displayed in the UI unit 215. 【0267】 In this embodiment, the vehicle agent 101 mediates the flow of information between the user in the real vehicle and the agent in the cyber vehicle. Therefore, the agent in the cyber vehicle (i.e., each user's agent) cannot detect the situation in the real vehicle or the status of the vehicle. In other words, the user's agent cannot know about the conversations between users taking place in the real vehicle or the current status of the vehicle (e.g., vehicle speed or fuel level) unless they are shared within the cyber vehicle. 【0268】 In this dialogue agent system 100, the vehicle agent 101 mediates and controls the flow of information between the real vehicle and the cyber vehicle, thereby eliminating the risk of the aforementioned private information being leaked to the agent. If the in-vehicle system 102 or the vehicle agent 101 were to share conversations within the real vehicle with an agent connected to the cyber vehicle without restriction, everything said in the real vehicle, including the user's private information, would be learned or remembered by an unrelated third-party agent and could be used in a way unintended by the user. To prevent such unforeseen circumstances, it is highly significant to have the vehicle agent 101 control the flow and sharing of information between the real vehicle and the cyber vehicle. 【0269】 There are two types of agents connected to the cyber cabin: the vehicle agent 101 and agents that act as a substitute for the user. However, only the vehicle agent 101 has contact with the real world of the real cabin and vehicle. The user's agent has contact with the digital data shared within the cyber cabin, but this information about the real cabin and vehicle is only conveyed through the vehicle agent 101. In other words, by making the information asymmetry between the vehicle agent 101 and the user's agent the basic design of the dialogue agent system 100, the user's private and sensitive information can be protected from other users' agents while using this dialogue agent system 100. 【0270】 In the above description, when dealing with privacy information or sensitive information, individual communication is carried out between the relevant agents, and the state is visualized. However, even if other agents can receive the communication, if there is no learning or recording of information related to the communication as digital data (including writing on the network or transmitting to other agents), this concern is considered to be reduced. Therefore, a mode indicating that the agent does not learn or record any information received may be provided, and it may be visually or auditorily expressed to the user that the agent is operating in that mode. For example, it may be meant that the avatar of the agent is wearing sunglasses or does not have an information terminal such as a personal computer or a smartphone, and is operating in a mode where it does not learn or record anything at all. 【0271】 FIG. 16 is a sequence diagram showing an example of a flow in which the vehicle agent 101 responds to a user's inquiry regarding the user's privacy by communicating individually only with the user's agent in the dialogue agent system 100 according to the present embodiment. 【0272】 Here, when the information transmitting agent (such as the vehicle agent 101) determines that the information exchange includes someone's privacy information or sensitive information (or the degree of inclusion is above a predetermined level), the state where the information exchange is carried out by individual communication only among the relevant parties is called the "secret mode". On the other hand, when the information transmitting agent (such as the vehicle agent 101) determines that the information exchange does not include anyone's privacy information or sensitive information (or the degree of inclusion is less than a predetermined level), the state where the information exchange is carried out by communication shared with all relevant parties is called the "normal mode". 【0273】 That is, when the determination in step S1402 of FIG. 14 is Yes, the operation mode of the dialogue agent system 100 (or individual agents) is switched to communicate in the secret mode, and when it is No, it is switched to communicate in the normal mode. 【0274】 Since the scene described here is the same as the scene in Figure 15, some redundant explanations may be omitted. 【0275】 First, user A105 asks the in-vehicle system 102 what time it was when they last came to this location (S1601). In response, the in-vehicle system 102 notifies the vehicle agent 101 of this question (as a result of voice recognition). 【0276】 Vehicle agent 101 determines that this question (or the series of exchanges including the answer) contains, or has a predetermined or greater likelihood of containing, private information (S1602). Then, it switches to secret mode, a mode in which information is exchanged separately with user A105 and its agent A103 regarding this exchange. 【0277】 The fact that this secret mode is in operation is also notified to user A105 via the UI unit 215 (S1603). For example, the avatar of agent B106, who is not subject to information exchange in secret mode, may be grayed out, or shown wearing earplugs, or only the subjects of secret mode (in this case, vehicle agent 101 and agent A103) may be highlighted, connected by a string telephone, or shown as being in a state where they are having a confidential conversation away from other agents. Alternatively, when vehicle agent 101 responds to the user, it may explain that it is operating in secret mode, switch to a voice or whispering tone for secret mode, display the answer on a monitor visible only to those involved in the conversation on the UI unit 215, or control the display method of the screen (such as a bias filter) so that it is only visible from the direction of user A105 who asked the question. 【0278】 Vehicle agent 101, operating in secret mode, forwards messages containing or highly containing user A105's private information (user A105's questions) individually only to the relevant user A105's agent A103 (S1604). 【0279】 Upon receiving this, agent A103 refers to user A105's behavioral history information and generates a response message indicating that the user had visited three years ago (S1605). Here, it is assumed that agent A103 has been granted prior permission (configured) to access user A105's privacy information, including that behavioral history information. 【0280】 Agent A103 then sends the response message "You were here 3 years ago" only to vehicle agent 101, who sent the question individually (S1606). Based on this response, the avatar of vehicle agent 101 or agent A103 responds (individually) to user A105 via the UI unit 215 (S1607). As a result, the response message "You were here 3 years ago" is notified to user A105 in the real vehicle room as video and audio information (S1608), and agent A103, which is connected to the cyber vehicle room, is also individually notified that user A105 has been responded to in this way. 【0281】 Steps S1602 through S1608 are an example of the conversational agent system 100 operating in secret mode because it determined that it was handling private information. 【0282】 Next, user A105 asks the in-vehicle system 102 a new question: "How much further is it to the destination?" Upon receiving this, the vehicle agent 101 makes the determination in step S1402 and determines that it does not contain anyone's private or sensitive information, or that the degree to which it does is below a predetermined level (S1611). 【0283】 Information such as the travel time and distance to the destination is a common topic for users in the actual vehicle cabin, and although it depends on traffic conditions to the destination, it is not related to the privacy or sensitive information of the vehicle or the individual user. Therefore, the vehicle agent 101 decides to change from secret mode to normal mode, switches the operation mode of the dialogue agent system 100 to normal mode, and notifies the user of this fact via the UI unit 215 (S1612). 【0284】 The notification that the dialogue agent system 100 is operating in normal mode is the opposite of the notification that it is operating in secret mode. Information exchange between agents connected to the real vehicle and the cyber vehicle is shared with all participants. Therefore, the UI unit 215 displays all the avatars of agents connected to the cyber vehicle, as shown in Figure 15(V), indicating that everyone is sharing information. If this display indicating normal mode takes up too much space in the UI unit 215, the normal mode can be indicated with an icon, or identification information (an icon indicating secret mode, or between agents interacting in secret mode) can be displayed only when in secret mode. 【0285】 When the vehicle agent 101 switches to normal mode, it shares the content of user A105's question with all agents connected to the cyber cabin (S1613). Furthermore, it obtains estimated time information to the destination (through API integration from the in-vehicle system 102's navigation system, or by coordinating with external services via the network 109, etc.) (S1614). Then, based on the information obtained, the vehicle agent 101 generates the answer message "We should arrive in about 30 minutes" (S1615). Subsequently, because it is in normal mode, it sends this generated answer message to all agents connected to the cyber cabin and to user A105 who asked the question (S1616). 【0286】 In this manner, if the conversational agent system 100 detects that an exchange involving user or vehicle privacy or sensitive information is occurring, it switches the information flow to secret mode and exchanges information only among the minimum necessary agents. If it detects that the information is not of that nature, it switches the information flow back to normal mode and shares the information of the exchange with all agents connected to the cyber cabin. By switching the method of information flow according to the information being handled, it is possible to prevent information from being shared or leaked to other (third-party) agents connected to the cyber cabin. As mentioned above, a mechanism to prevent unnecessary information sharing and information leakage to agents that can act spontaneously in cyberspace is expected to become very much in demand in the future world where agents are becoming more widespread. 【0287】 Furthermore, this disclosure is not limited to embodiments that switch the method of information distribution based on the information being handled. Below, we will describe a method that does not involve switching such operating modes. 【0288】 Based on the explanation so far, whether the conversational agent system 100 operates in normal mode, The explanation described how the system operates in secret mode, with agents communicating individually with each other depending on the information being handled. In another embodiment, all user agents connected to the vehicle agent 101 may communicate individually with the vehicle agent 101, regardless of the information being handled. 【0289】 In FIGS. 3 to 7, it was explained that the user agent obtains access information for an online meeting where multiple users can enter and leave, and account information / group information of SNS, and connects to other agents including the vehicle agent 101 via the network 109. These are an example of a method of sharing information with multiple agents in the cyber space. However, instead of these access information for online meetings and account information of SNS, it is also possible to obtain the endpoints of the agents to be connected (such as URLs for directly exchanging information with the agents via APIs). 【0290】 For example, user A105 may use an app (on the information terminal A104 and including a camera and agent A103) to read a QR code storing the endpoint of the vehicle agent 101 displayed on the UI unit 215 of the in-vehicle system 102. 【0291】 Agent A103 may send its own endpoint (access information) to the obtained endpoint of the vehicle agent 101 so that the vehicle agent 101 can send a message to agent A103. 【0292】 Also, agent A103 may textify and send and receive messages to and from the endpoint of the vehicle agent 101 via communication protocols such as HTTP requests (POST, GET, etc.), WebSocket, and gRPC. 【0293】 Also, for secure communication, connection authentication may be performed using an API key or token at the start of connection. Thereby, it can be confirmed that the agents recognize each other appropriately and message exchange is permitted. 【0294】 Figure 17 shows an example of how the conversational agent system 100 according to this embodiment uses an API (HTTP request / response) when the vehicle agent 101 communicates individually with the user A105's agent A103. In particular, specific examples of the exchange in steps S1310 and S1313 of Figure 13 will be explained here. 【0295】 In step S1310, vehicle agent 101 sends a request message to agent A103 to evaluate restaurant candidates. The process of establishing a connection between vehicle agent 101 and agent A103 will not be explained here, but if these two are connected using the HTTP protocol, then vehicle agent 101 may send an HTTPS request to agent A103 as shown in the example HTTPS request (left side) of Figure 17. 【0296】 The first line indicates that this request is being sent to Agent A103's endpoint (the URL being accessed) https: / / AgentA / {session_ID} / messages using the HTTP version 1.1 POST method. {session_ID} should be replaced with the unique session ID for this communication session. 【0297】 The second line is the Authorization header, where the {access_token} part specifies the access token used for authentication with the server on which agent A103 is running. The access token is a secret authentication code issued to secure this connection and is used by the server to decide whether to approve the request. 【0298】 The third line is the Content-Type header, which specifies that the data to be sent is in JSON format. The request body that follows describes the content of the message to be sent in JSON format. Here, the message is text data requesting Agent A103 to evaluate restaurant candidates. Specifically, the parts that represent mentions, "@Agent A, @Agent B", have been removed from the content explained in Figure 11. This is because it is a separate HTTPS request sent from Vehicle Agent 101 to Agent A103, so mentions to specify the recipient are not necessary. 【0299】 In step S1313, agent A103 sends an evaluated response message to vehicle agent 101 regarding the restaurant candidates. If these two agents were connected via the HTTP protocol, agent A103 could send an HTTPS response to vehicle agent 101 as shown in the example HTTPS response (right side) of Figure 17. 【0300】 The first line indicates that the request was processed successfully by returning a status code "200 OK" as a response. 【0301】 The second line is the Content-Type header, which specifies that the data format to be sent is JSON. 【0302】 The subsequent response body describes the message that Agent A103's server sends in JSON format. Here, the message contains Agent A103's evaluation and brief comments on the restaurant candidates as text data. 【0303】 In this way, the two agents can use an endpoint to connect directly via a predetermined communication protocol and send and receive messages. In this case, even if another agent is connected to the cyber cabin, the communication between the vehicle agent and each agent is separate and not shared with other agents, thus eliminating the risk of information leakage and unauthorized use to the aforementioned agents. 【0304】 Furthermore, the cyber cabin of the conversational agent system 100 may be connected solely through individual communication between the two agents. In this case, since the agents of the same user do not directly share information with each other, it is believed that a conversational agent system 100 with high psychological safety can be realized, such as when an unspecified number of users are riding in the same vehicle (real cabin). 【0305】 Figure 18 shows an example of user data referenced by the user's agent in the dialogue agent system according to this embodiment. In this figure, the agent used by the user connects to the cyber cabin and transmits information while referencing the user's health, beliefs, creed, preferences, interests, behavioral history, cabin environment, etc., and this is an example of the data referenced in that process. 【0306】 As shown here, user data is described separately into main categories, subcategories, and data. This allows the agent to generate personalized responses to the user by referring to this data, such as the user's level of interest. It is also assumed that the agent has been configured to access this user data in advance. 【0307】 Furthermore, this user data is securely managed in the memory of the computer system on which the agent operates. For example, when agent A103 (or its program) runs on information terminal A104, it refers to the user data file securely managed in memory 204 of information terminal A104. 【0308】 On the other hand, when the vehicle agent 101 (or its program) is running in the in-vehicle system 102, it refers to a vehicle data file securely managed in the memory 214 of the in-vehicle system 102. The vehicle data file may contain information such as the personal information of the vehicle owner, accident history, and driving history (where the vehicle has been driven). 【0309】 The main health category includes subcategories such as health checkups. For health checkups, the user's health checkup results are recorded in the data. For example, it might record that the systolic blood pressure is 120 mmHg. This information is used, for example, to provide recommendations regarding diet and exercise. 【0310】 The main category of "faith" includes subcategories such as religion. The user's religious beliefs are recorded in the data. For example, it might state that the user does not practice Buddhism. This information is used, for example, to provide recommendations for dining, sightseeing, and shopping. 【0311】 The main category of beliefs includes subcategories such as food. The user's beliefs regarding food are described in the data. For example, it might state that they are not vegan. This information is used, for example, for recommendations regarding food and shopping. 【0312】 The main category of preferences includes subcategories such as food. The user's preferences for food are described in the data. For example, the degree to which they like each dish is quantified and described, such as ramen being rated 8 and sushi 7. This information is used, for example, for making recommendations related to food. Similarly, in the music subcategory, the user's preferences for music are described in the data. For example, the degree to which they like each music genre is quantified and described, such as Japanese music being rated 8 and Western music 7. This information is used, for example, for making recommendations regarding the music played by the in-car system 102. 【0313】 The main category of interests is further subdivided into subcategories such as genres. For each genre, the user's level of interest is quantified and described. For example, the level of interest for each genre might be 9 for food and drink, and 8 for scenic spots. This information is used, for example, to recommend nearby places to visit. Additionally, the subcategory of registered map locations includes data on locations registered by the user in the map service. For example, location information such as the latitude and longitude of a ramen shop might be included. This information is used, for example, to recommend nearby places to visit. 【0314】 The main category of activity history includes subcategories such as social media browsing. The user's browsing history for social media is recorded in the data. For example, it might record that the user viewed content posted by user □□ at https: / / ... This information is used, for example, to recommend places to visit. Similarly, the user's browsing history for web browsing is recorded in the data. For example, it might record that the user viewed content on the webpage called △△ Guide (https: / / ...). This information is used, for example, to recommend tourist destinations, restaurants, shopping, content to play, routes, and places to visit. 【0315】 The main vehicle category includes subcategories such as temperature. The user's recommended temperature settings are described in the data. For example, the air conditioning temperature might be described as 25°C. This information is used, for example, to make recommendations regarding the control of the cabin environment. Additionally, the user's recommended settings for autonomous driving modes are described in the data. For example, the recommendation for an eco-driving mode with low environmental impact might be described. This information is used, for example, to make recommendations regarding vehicle drive control. 【0316】 As described above, the agent can use the information in these user data files to generate, evaluate, or request recommendations from other agents (e.g., vehicle agent 101) regarding the degree of recommendation for the user, such as recommended tourist spots, restaurants, routes to intermediate / destination points, video / audio content to play, cabin and seat lighting, temperature, and autonomous driving mode. 【0317】 In the above explanation, the vehicle agent 101 and one or more user agents (agent A103, agent B106) were connected via network 109 to search, evaluate, recommend, or select a wide variety of options. However, such considerations may also be carried out collaboratively by summoning other agents, especially specialized agents with detailed knowledge in specific fields. 【0318】 Specifically, a regional tourism-focused agent that introduces local attractions including the vehicle's current location can be connected to the Cyber ​​Room (by the vehicle agent 101) via Network 109, and the agent can recommend suitable places to visit or introduce the characteristics of recommended candidates. The vehicle agent 101 may also select a regional tourism-focused agent based on the vehicle's current location and have them participate in the Cyber ​​Room. By adding such regional tourism-focused agents as evaluators, it is likely that the system will be able to extract, evaluate, or recommend local sightseeing, dining, shopping, and other experiences based on more locally relevant information, thereby improving the user's travel experience. 【0319】 In another use case, a specialized regional traffic monitoring agent, familiar with traffic conditions near the vehicle's current location, could be similarly connected to the cyber cabin to collaboratively consider route selection. 【0320】 The vehicle agent 101 can broaden and deepen the scope of discussions among agents by connecting to one or more such specialized external agents depending on the topic of discussion, or by disconnecting them as appropriate. As a result, the conversational agent system 100 can not only improve the user's travel experience, but also reduce environmental impact caused by inefficient route selection and alleviate traffic congestion. 【0321】 Figure 19 shows an example of the coordination between the software and hardware configurations of the dialogue agent system according to this embodiment. Here, the symbols used in Figure 2 are reused as much as possible. Excluding newly assigned symbols, the detection unit 213 of the in-vehicle system 102 is described separately as a detection unit (in-vehicle) 213a for detecting the status of the user, seat belts, seats, etc. inside the vehicle, and a control unit (control unit / outside vehicle) 213b for detecting the external conditions around the vehicle and the vehicle's powertrain (here, this refers to the entire drive system of the vehicle, including generating, transmitting, and rotating the tires). 【0322】 Furthermore, the control unit 216 is described in two parts: a control unit (in-cabin) 216a that controls and drives the in-cabin equipment (seats, lighting, air conditioning, etc.), and a control unit (vehicle drive) 216b that controls the physical mechanisms for driving (driving and stopping) the vehicle, including the vehicle's powertrain and driving devices. The control unit (vehicle drive) 216b includes the mechanism and control functions that physically realize the automatic driving function, which automatically operates the vehicle's driving devices (steering wheel, accelerator, brakes, etc.) based on instructions from the automatic driving control software that automatically drives the vehicle to a destination registered in the navigation system, as well as the vehicle's driving assistance functions (assistant driving functions such as stopping at obstacles or driving while maintaining the lane). 【0323】 The newly coded blocks are the software programs that operate within the arithmetic unit. The software that runs on the arithmetic unit 202 of the information terminal, or runs in conjunction with other computer systems via the network 109, includes the information terminal management software 1901 that oversees the overall functions of the information terminal, and the user agent software 1902 that performs tasks such as learning and inference processing for the user's agent, and controlling the movement and speech of the avatar that represents the agent's body and appearance. 【0324】 Furthermore, the software executed by the arithmetic unit 212 of the in-vehicle system 102, or executed in conjunction with other computer systems via the network 109, includes vehicle management software 1903 that oversees the overall functions of the in-vehicle system 102, vehicle agent software 1904 that performs learning and inference processing for the vehicle's agents, and controls the movement and speech of the vehicle agent's avatar, content playback software 1905 that controls the playback of video content, music content, etc. via the UI unit 215, navigation system software 1906 that searches, sets, and updates routes to the vehicle's destination and waypoints, cabin control software 1907 that controls in-cabin equipment (seats, lighting, air conditioning, etc.), and driving assistance / autonomous driving control software 1908 that supports the safe driving of the vehicle or instructs the control unit 216b to perform driving operations for autonomous driving according to the sensor data of the vehicle's detection unit 213b and the destination / route registered in the navigation system software 1906. 【0325】 [Action Processing] From here, we will explain the details of the operation process of the embodiment described in Figure 2B. 【0326】 The following section, using Figure 19, will explain in detail an example of how the processes from steps S808 to S810 in Figure 8 (or from steps S1020 to S1022 in Figure 12) are carried out by hardware and software. 【0327】 User B108's response (utterance) "That's fine" is acquired through the voice microphone of the detection unit 213a and the in-vehicle camera (for example, speech recognition by lip reading or intent estimation by gesture). Based on this sensing data, the vehicle management software 1903 (or vehicle agent software 1904) performs speech recognition or image recognition processing and recognizes that user B108 uttered "That's fine." 【0328】 The user's speech inside the vehicle (corresponding to one line of chat content in Figure 11) may be notified from the vehicle management software 1903 to the vehicle agent software 1904 via a predetermined API, or the vehicle agent software 1904 may recognize it by performing speech recognition processing or the like. 【0329】 Based on the results of speech recognition, the vehicle agent software 1904, using natural language processing, recognizes the user's decision regarding recommended candidates and sends a request for setting or modification processing related to the selected candidate to the software that controls and manages the corresponding function. This software includes software that controls vehicle equipment and components installed in the vehicle. While there are multiple patterns, two representative processing request patterns will be explained. 【0330】 (1. Pattern of requesting changes to the navigation system) If the discussion topic in the dialogue agent system 100 is a tourist spot, restaurant, or shop to visit, and it is decided to make a stop there, the route needs to be changed to include that specific location. Therefore, the vehicle agent software 1904 requests the navigation system software 1906 to change the route to include the new waypoint / destination. 【0331】 When the navigation system software 1906 receives this request via a predetermined API, it searches for a new route based on the current location and the new waypoint / destination. For the sake of simplicity, this explanation assumes that there is only one route option. If there are multiple route options, they may be displayed in the UI section 215 for the user to select. In addition, to enable the user to make a decision based on traffic conditions for each route option, the system may also display information such as travel time, distance, and toll fees. 【0332】 Once the route is determined, the navigation system software 1906 notifies the driver assistance / autonomous driving control software 1908 of the route information via a predetermined API. Alternatively, the driver assistance / autonomous driving control software 1908 may check and obtain the currently set route information from the navigation system software 1906 (or vehicle management software 1903, etc.) based on the occurrence of specific events (such as route change events or changes in traffic conditions) or the fulfillment of conditions (such as after a predetermined time has elapsed or when the vehicle is started). 【0333】 In response, the driver assistance / autonomous driving control software 1908, in cooperation with the detection unit 213b, instructs the control unit 216b to assist in the safe driving of the vehicle and to perform driving operations for autonomous driving according to the newly registered destination / route. Accordingly, the control unit 216b physically executes and controls the powertrain and driving devices to drive along the newly registered route, making it possible to safely and / or automatically move the user inside the vehicle to the tourist destination or restaurant that the conversational agent system 100 has decided to visit. 【0334】 Although the vehicle agent software 1904 is described as sending a route change request to the navigation system software 1906, it may also send it to other software (for example, the vehicle management software 1903, the driver assistance / autonomous driving control software 1908, etc.). 【0335】 For example, if you use a robotaxi as a means of transportation while traveling with friends, if it's equipped with this conversational agent system 100, you can ask the agent to tell you about local specialties, suggest tourist spots, restaurants, and shopping areas that suit everyone's interests, and since the ride to your desired destination is autonomous, once you've decided on a place, it will automatically take you there, making for a completely different travel experience. In other words, there's no need for a driver, a guide, or someone to find nearby places to stop, so everyone can enjoy themselves equally, and it's expected to be a trip where it's easier to discover and experience the unique charm of the place. 【0336】 (2. Pattern for requesting a change to content playback) Another typical processing request pattern involves changing the content being played using the UI unit 215 (its display and speakers). 【0337】 If the topic of discussion in the dialogue agent system 100 is video content such as movies or dramas that you want to watch, or audio content such as music or radio that you want to listen to, and it is decided to play a certain content, the vehicle agent software 1904 will, in accordance with that decision, send a playback request for the decided content to the content playback software 1905. 【0338】 This playback request may include one or more of the following: the name of the content used in suggesting candidates by the dialogue agent system 100, a summary, playback time, a thumbnail image representing the content, the name of the distribution service, the URL from which the content was obtained, the names of the performers / artists, and the content identification code. 【0339】 When the content playback software 1905 receives this request via a predetermined API, it acquires the content data via the communication unit 211 based on the content information to be played, decodes it, and plays it using the display and speakers of the UI unit 215. This allows the user to quickly and smoothly view the content determined by the conversational agent system 100 in the vehicle without having to search for the content themselves by operating a menu screen or the like. 【0340】 As described above, by coordinating the software on the in-vehicle system 102 and controlling the vehicle's hardware in accordance with the decisions made by the dialogue agent system 100, it becomes possible to, for example, allow everyone to watch video content they want to see in the real car cabin, play their favorite background music, automatically set a route to a newly decided destination, or, in the case of an autonomous vehicle, automatically drive and take them to that destination. 【0341】 [Other uses besides vehicles] In the various embodiments described above, a dialogue agent system that interacts with a user traveling in a vehicle was explained as an example. However, information processing that interacts with a user inside a room and sets and controls hardware related to the room is not limited to this disclosure. 【0342】 For example, the car interior in one example of this embodiment can also be interpreted as a dwelling. 【0343】 The conversational agent system 100 shown in Figure 2 can be implemented by replacing the in-vehicle system 102 with a home control system that manages the equipment and appliances of a house, while keeping the same configuration. This allows the home control system to perform tasks such as investigation, analysis, evaluation, consideration, and proposal using the user's agent in response to requests from one or more users inside the house. For example, the following use cases can be considered for the conversational agent system 100 for users inside a house. 【0344】 Fitness suggestions: Based on users' health data and activity history within their homes, exercises that can be done at home are suggested if there is a lack of exercise. These include stretching, yoga, and simple strength training programs, tailored to each user's physical condition and health goals. 【0345】 Meal suggestions: Suggest recipes tailored to the health, preferences, and dietary restrictions (religious or nutritional) of users within the home, as well as nearby healthy restaurants and delivery options. For example, suggest low-calorie meals or vegetarian menus. 【0346】 Air conditioning management proposals: Based on the health data of users in the home and the current living environment (humidity, temperature, etc.), we propose appropriate air conditioning settings and humidifier use. For users with allergies, we recommend turning the air purifier on or off. 【0347】 Religious Event Suggestions: Based on the user's beliefs within the home, the system will notify them of prayer times and religious events, and assist with preparations. It will also automate the creation of quiet spaces within the home and adjust lighting and volume levels. 【0348】 Content Suggestions: We propose video / audio content that can be enjoyed by all users in the living room of a house, supporting shared family time. We recommend content that is balanced to be enjoyable for everyone, taking into account the preferences of each user. 【0349】 Monitoring suggestion: If abnormalities are detected in the health status of a user inside the home (e.g., heart rate or body temperature), an alert will be automatically sent to a nearby hospital or family member, and necessary support will be suggested. 【0350】 In the aforementioned dialogue agent system 100, simply by replacing the in-vehicle system 102 with a home control system, the cyber cabin with a cyber home, and the vehicle agent 101 with a home agent, it becomes possible to support a richer, more comfortable, and more convenient lifestyle within the home for themes such as those considered in the use cases described above. Therefore, the dialogue agent system 100 is considered to have great value not only for users who travel by vehicle, but also for users who spend time at home. 【0351】 Furthermore, the dialogue agent system 100 disclosed herein is not limited to vehicles or residences. It is a system that can be used in any space where one or more users are present, and where an agent evaluates and proposes to users an event with a large number of choices that all users in that space experience in common. For example, it could be used to propose or automatically change spatial controls (lighting, air conditioning, sound, display content, etc.) in places such as shops, restaurants, offices, public transport vehicles, train stations and airports, large buildings, shopping malls, schools, cram schools, and libraries, based on the evaluations of the agents present. 【0352】 Furthermore, the programs executed by each device of the interactive agent system 100 according to each of the above embodiments may be provided as files in an installable or executable format, recorded on a computer-readable recording medium (Computer Program Product) such as a CD-ROM, FD, CD-R, or DVD. 【0353】 Furthermore, the programs executed by each device of the dialogue agent system 100 according to each of the above embodiments may be stored on a computer connected to a network such as the Internet and provided by downloading them via the network. Alternatively, the programs executed by each device of the dialogue agent system 100 according to each of the above embodiments may be provided or distributed via a network such as the Internet. 【0354】 Furthermore, the programs executed by each device of the dialogue agent system 100 according to the above embodiments may be pre-installed and provided in ROM or the like. 【0355】 According to at least one embodiment described above, further improvements can be achieved using a user-associated, conversational agent. 【0356】 While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols] 【0357】 100 Dialogue Agent Systems 101 Vehicle Agent 102 In-vehicle systems 103 Agent A 104 Information Terminal A 105 User A 106 Agent B 107 Information Terminal B 108 User B 109 Network 201, 211 Communications Department 202, 212 Arithmetic section 203, 213, 213a, 213b Detection Unit 204, 214 memory 205, 215 UI section 216, 216a, 216b control unit 1901 Information Terminal Management Software 1902 User Agent Software 1903 Vehicle Management Software 1904 Vehicle Agent Software 1905 Content Playback Software 1906 Navigation System Software 1907 In-cabin control software 1908 Driver assistance / autonomous driving control software

Claims

[Claim 1] This is an information processing method for AI agents executed by computers included in the system associated with the room. The aforementioned AI agent includes: A room AI agent that receives and responds to instructions from the user in the room via a voice-activated user interface (VUI), It includes one or more personal AI agents that can serve one or more users in a room. The aforementioned information processing method is The one or more personal AI agents and the room AI agent are connected to a virtual space where they can interact with each other. When the room AI agent connected to the virtual space receives a suggestion request regarding a topic from one or more users in the room, it recognizes the user who made the suggestion request and the accepted topic via the voice interaction interface (VUI), and notifies the result of the recognition to the one or more personal AI agents connected to the virtual space. One or more personal AI agents connected to the virtual space generate one or more suggestions based on the recognized topic and the recognized user profile. If a proposal is selected by the user from among the one or more proposals generated, The computer applies settings or processing to the system associated with the room in accordance with the adopted proposal, The one or more personal AI agents include a first agent corresponding to a first user in the room and a second agent corresponding to a second user in the room. The aforementioned virtual space is The first cyber cabin is a cyber cabin in which each of the aforementioned AI agents can interact with each other, The first cyber private room is a cyber cabin in which the aforementioned room AI agent and the first agent can communicate with each other. The room includes a second cyber private room which is a cyber room where the aforementioned room AI agent and the second agent can communicate with each other, The aforementioned AI agent in the room, If the first user or the second user requests a proposal regarding a first topic that does not relate to the user's privacy information, the first cyber room will interact with at least the personal AI agent corresponding to that user. If the first user or the second user requests a proposal regarding a second topic related to the user's privacy information, the first cyber private room or the second cyber private room, where information is exchanged between the two parties, will interact with the personal AI agent corresponding to the user. Information processing methods. [Claim 2] The above one or more proposals include a destination, The system associated with the aforementioned room includes hardware that displays information within the room. The aforementioned settings or processes cause the hardware of the system associated with the room to display guide information and / or access information to the proposed destination. The information processing method according to claim 1. [Claim 3] The above one or more proposals include a destination, The aforementioned room is the passenger compartment of the vehicle. The setting or process involves setting the destination in the vehicle's navigation system and / or setting the destination in the vehicle's automatic driving control system. The information processing method according to claim 1. [Claim 4] The above one or more proposals include a destination, The one or more personal AI agents connected to the virtual space generate an evaluation of the destination based on the recognized user's profile and accompany the one or more suggestions. The information processing method according to claim 1. [Claim 5] The above one or more proposals include a destination, The one or more personal AI agents connected to the virtual space access local information and map information of the region to which the destination belongs, determine the validity of the destination based on the local information and map information, and generate the one or more suggestions based on the result of the validity determination. The information processing method according to claim 1. [Claim 6] The aforementioned room is the passenger compartment of the vehicle. The aforementioned virtual space is a cyber cabin, The connection of the one or more personal AI agents to the cyber cabin is performed after the user corresponding to the personal AI agent enters the cabin. The information processing method according to claim 1. [Claim 7] The aforementioned room is the passenger compartment of the vehicle. The aforementioned virtual space is a cyber cabin, The connection of the one or more personal AI agents to the cyber vehicle is terminated when the user corresponding to the personal AI agent leaves the vehicle or when the vehicle arrives at its destination. The information processing method according to claim 1. [Claim 8] The system associated with the aforementioned room includes hardware that displays information within the room. The computer causes the hardware of the system attached to the first cyber room, the first cyber room, and the second cyber room, with respect to the cyber room in which the room's AI agent is having a conversation, to display an image indicating that an avatar representing the character of an AI agent capable of conversation in that cyber room is present, and that other characters are not present. The information processing method according to claim 1. [Claim 9] Each of the one or more personal AI agents is used on an information terminal brought into the vehicle interior. The computer causes the display mounted in the vehicle compartment to display a code for connecting to the virtual space associated with the vehicle, Each of the one or more personal AI agents is connected to the virtual space when the information terminal reads the code. The information processing method according to claim 6. [Claim 10] Each of the one or more personal AI agents is selected from among multiple available AI agents by operation on the information terminal. The information processing method according to claim 9. [Claim 11] The aforementioned virtual space is an SNS in which a group of one or more personal AI agents participates. The aforementioned computer, The AI ​​agent in the vehicle's cabin displays a code on a display in the cabin requesting to join the SNS group. The information terminal reads the code requesting participation, The personal AI agent of the information terminal, based on the read code, permits the room AI agent to join the SNS group in which the room AI agent is a member. The information processing method according to claim 9. [Claim 12] In the settings of the aforementioned SNS group, a condition for removing the aforementioned room AI agent from the group is set in advance. The aforementioned vehicle is a rental car with a specified rental period. The aforementioned termination condition is the end of the rental period for the aforementioned rental car. The information processing method according to claim 11. [Claim 13] Processor and A memory containing a program for causing the processor to execute the information processing method described in any one of claims 1 to 12, The aforementioned chamber is equipped with or connected via a network. computer. [Claim 14] To cause the computer to execute the information processing method described in any one of claims 1 to 12. program.

Images