Ai emotion recognition-based thermal control plush doll and interactive charater system using it

KR103013846B1Active Publication Date: 2026-09-04GREENGATE CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
KR1020250208697
Authority / Receiving Office
KR · KR
Patent Type
Patents
Current Assignee / Owner
Filing Date
2025-12-24
Publication Date
2026-09-04
Estimated Expiration
2045-12-24

Smart Images

  • Figure 112025146330637-PAT00001_ABST
    Figure 112025146330637-PAT00001_ABST
Patent Text Reader

Abstract

The present invention relates to an AI emotion recognition-based heat-controlled plush doll and an interactive character system using the same. More specifically, it relates to an AI emotion recognition-based heat-controlled plush doll and an interactive character system using the same that provides accurate emotion care functions, differentiates itself as a heat-controlled doll specialized for winter, enables natural character-based interaction, expands brand experience, and allows safe use by all ages through a heat structure that ensures safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technology Field

[0001] The present invention relates to an AI emotion recognition-based heat-controlled plush doll and an interactive character system using the same. More specifically, it relates to an AI emotion recognition-based heat-controlled plush doll and an interactive character system using the same that provides accurate emotion care functions, differentiates itself as a heat-controlled doll specialized for winter, enables natural character-based interaction, expands brand experience, and allows safe use by all ages through a heat structure that ensures safety. Background Technology

[0002] Prior art related to the present invention includes character dolls and attachment dolls. Patent Document 1, an offline character doll control device and method using user's emotional information, automatically expresses the user's emotional information—expressed through messages or profiles created by the user via an application service—through an offline character doll possessed by a designated counterparty, thereby easily and quickly transmitting the user's emotional information to the counterparty. Additionally, Patent Document 2, a child emotion-based interactive attachment doll for children, identifies a specific child's emotional state from the child's voice (speech) using an emotion dictionary, and induces smooth psychological counseling with the child by transmitting a response message according to a response scenario or a response message from a guardian / manager via voice. Furthermore, Patent Document 3, a method and device for notifying dangerous situations based on daily pattern emotion analysis using artificial intelligence doll conversation, enables conversation between an artificial intelligence doll conversation device and the user of the device through an artificial intelligence doll conversation service. It constructs a daily pattern emotion analysis database record through the analysis of the artificial intelligence doll conversation history, and detects dangerous situations through learning-based predictive analysis of the daily pattern emotion analysis database record, thereby enabling rapid and accurate notification to a guardian terminal, medical institution, or control agency. In addition, the method and device for providing cognitive card analysis information based on daily pattern emotion analysis using AI doll conversations described in Patent Document 4 enables conversation between an AI doll conversation device and a user of the device through an AI doll conversation service, and constructs a daily pattern emotion analysis database record through the analysis of the AI ​​doll conversation history, and constructs cognitive card analysis information through learning-based predictive analysis of the daily pattern emotion analysis database record, and can provide it to a guardian terminal, a medical institution, or a control agency. Prior art literature

[0003] Registered Patent Publication No. 10-1906500 Registered Patent Publication No. 10-2584773 Registered Patent Publication No. 10-2701869 Registered Patent Publication No. 10-2701870 The problem to be solved

[0004] The present invention aims to provide an AI emotion recognition-based heat-controlled plush doll that analyzes a user's emotional state and the temporal characteristics of the emotion using artificial intelligence, and performs a heating function and various reaction actions according to the results of the emotion analysis, as well as an interactive character system using the same. means of solving the problem

[0005] An AI emotion recognition-based thermal control plush doll and an interactive character system using the same according to a preferred embodiment of the present invention are characterized by comprising: a doll that performs emotion, heat generation, character, store linkage, and multi-mode (peace, excitement, sleep, vitality); and a server that performs an emotion analysis engine, a character engine, a thermal control algorithm, and a mode switching algorithm.

[0006] In addition, the doll is characterized by providing thermal output, visual facial expression output, and voice output considering the direction of emotional change, duration, or whether it occurs repeatedly; providing thermal output with different temperature rising curves, maintenance times, or decreasing patterns depending on the emotional state; visually outputting intermediate facial expression states corresponding to transitions between emotional states; and providing voice output with different tones, speeds, or dialogue depending on the emotional state and character.

[0007] In addition, the doll comprises: a heating unit including a PTC heating element; a sensor unit including a voice emotion sensor, a hug pressure sensor, a touch sensor, and an environment sensor; an audio unit that outputs voice; a communication unit; and a control unit; wherein the control unit is characterized by performing sensing that detects voice, hug (torso, chest), touch (head, arm, chest), and environment (temperature, illuminance, noise) using a sensor; and performing operations that perform voice output, LED output, heating output, and motor output.

[0008] In addition, the server comprises a communication unit; a storage unit; and a control unit; and the control unit is characterized by performing sensing that detects voice, hug, touch, and environment using a sensor; and performing operations that perform voice output, LED output, heat output, and motor output.

[0009] Additionally, the system further includes a store that responds to customers, wherein the store includes a kiosk; a customer section; and a control section; and wherein the control section performs customer response by adding customer-tailored actions to the actions of the doll and the server.

[0010] In addition, the character part of the doll is characterized by including: a feature DB that stores character features of Fruity, Lolly, Mashi, and Puppy; a learning part that learns information from the feature DB; an inference part that infers character features from input data; and an output part that outputs the inferred features. Effects of the invention

[0011] The present invention utilizes artificial intelligence to analyze a user's emotional state and models emotions as a continuous state. By performing heating functions and various reaction actions based on the analysis results, it enables natural emotional response and provides accurate emotional care. Furthermore, it is differentiated as a heating doll specialized for winter, facilitates natural character-based interaction, expands the brand experience, and offers the effect of being safe for all ages to use due to its safety-assured heating structure. Brief explanation of the drawing

[0012] Figure 1 is an example diagram showing the configuration of the AI ​​emotion recognition-based heat control plush doll of the present invention and an interactive character system using the same. FIG. 2 is a block diagram showing the configuration of the doll of the present invention. Figure 3 is a block diagram showing the configuration of the server of the present invention. FIG. 4 is a block diagram showing the configuration of the store of the present invention. FIG. 5 is a block diagram showing the configuration of the character part of the present invention. Specific details for implementing the invention

[0013] Hereinafter, an AI emotion recognition-based thermal control plush doll and an interactive character system using the same according to an embodiment of the present invention will be described in detail with reference to the drawings. Descriptions of conventionally known matters below are omitted or simplified to clarify the gist of the present invention. The components included in the description of the present invention operate individually or in a composite combination.

[0014] Figure 1 is an example diagram showing the configuration of the AI ​​emotion recognition-based heat control plush doll of the present invention and an interactive character system using the same.

[0015] Referring to FIG. 1, the AI ​​emotion recognition-based heat control plush doll and the interactive character system using the same include: a doll (1) that performs emotion, heat generation, character, store linkage, and multi-mode (peace, excitement, sleep, vitality); a server (2) that performs an emotion analysis engine, a character engine, a heat control algorithm, and a mode switching algorithm; and a store (3) that responds to customers.

[0016] The doll (1) detects the user's emotional input (voice, facial expression, etc.) and the temporal characteristics of the emotion, and performs heating and character mode (expression) according to the instructions of the server (2).

[0017] The server (2) analyzes the data received from the doll (1) to determine the emotion, the direction of the change in emotion, the duration, or whether it occurs repeatedly, and executes the doll's heat control algorithm and mode switching algorithm accordingly.

[0018] The store (3) is operated in conjunction with the system to provide customer service and recommends customized products based on the condition of the doll (1).

[0019] The server (2) acts as the brain of the system. The emotion analysis engine analyzes data transmitted from the doll (1) (e.g., voice tone, words, facial expression images, etc.) to determine the user's current emotional state (e.g., peaceful, excited, sad, energetic, etc.). The character engine determines and instructs the doll to switch to a character mode (peaceful, excited, sleep, energetic) based on the determined emotion. The thermal control algorithm calculates the temperature (heat level) most suitable for the user's emotional state and instructs the doll to apply it. Sadness / Anxiety -> Warm heat (providing comfort / sense of stability), Excitement / Energy -> Reduced heat or maintaining a constant temperature. The mode switching algorithm determines the doll's multi-mode based on the results of the emotion analysis.

[0020] When a user is stressed and speaks to the doll (1), the user says to the doll in a low, tired voice, "Today was so hard..." The doll (1) captures this voice data and transmits it to the server (2). The server (2) uses an emotion analysis engine to analyze the transmitted voice data. It determines the emotion to be "sadness" or "anxiety / fatigue." The server (2) uses a character engine and mode switching algorithm to decide to switch to the "peace" mode, which is most suitable for the "sadness" emotion. The thermal control algorithm instructs the doll's temperature to rise to 38 degrees to induce the "peace" mode. The doll (1) activates the internal heating system of the doll (1) in accordance with the instructions of the server (2) to become warm. Additionally, it performs visual / auditory expressions suitable for the "peace" mode (e.g., emitting soft light or playing a comforting voice message in a low tone). The server (2) transmits this emotional state information to the store (3). The store (3) can send messages to customers recommending "aromatherapy products good for mental and physical relaxation" or "sleep-inducing music for relaxation." When the user hugs the doll (1), the doll (1) becomes warm and provides a "comforting" feeling, and performs the role of an interactive character with a user-customized response.

[0021] FIG. 2 is a block diagram showing the configuration of the doll of the present invention.

[0022] Referring to FIG. 2, the doll (1) includes a heating unit (11) including a PTC heating element; a sensor unit (12) including a voice emotion sensor, a hug pressure sensor, a touch sensor, and an environment sensor; an audio unit (13) that outputs voice; a communication unit (14); and a control unit (15). The control unit (15) performs sensing (16) that detects voice, hug (body, chest), touch (head, arm, chest), and environment (temperature, light intensity, noise) using a sensor; and performs operations (17) that perform voice output, LED output, heating output, and motor output.

[0023] The doll (1) is responsible for input (sensing) and output (operation) at the point of contact with the user. The control unit (15) acts as the center to collect data from the sensor unit (12) and performs operations (17) according to the analysis results of the server (2). The heating unit (11) uses a PTC heating element to prevent overheating and maintain a constant temperature. It provides emotional stability by providing a temperature similar to or warmer than the user's body temperature. The sensor unit (12) identifies the user's intention and state, detects voice tone and size, detects pressure on the torso / chest (degree of hugging), detects touch by distinguishing between head (stroking) and arm / chest (patting), and measures environment by measuring ambient temperature, illuminance, and noise. The audio unit (13) outputs the character's voice or sound effects through a speaker. The communication unit (14) transmits and receives data with the server (2) (Wi-Fi, Bluetooth, LTE, etc.). The control unit (15) oversees the collection and preprocessing of sensing (16) data and the execution of operation (17) commands. The control unit (15) outputs heat, visual expressions, and voice by considering the direction of change in emotion, duration, or whether it occurs repeatedly. The operation (17) outputs heat differently depending on the emotional state, the temperature rise curve, maintenance time, or decrease pattern, visually outputs an intermediate expression state corresponding to the transition between emotional states, and outputs voice with different tones, speeds, or dialogue depending on the emotional state and character.

[0024] "Situation of soothing an anxious child," let's assume a situation where the user (child) is startled by the sound of thunder and tightly hugs the doll (1). Sensing (16) performed: Environment sensor: detects a sudden increase in noise (thunder), hug pressure sensor: detects the child strongly pressing the doll's torso and chest, voice emotion sensor: detects the child's trembling voice saying "I'm scared," touch sensor: detects the child tightly gripping the doll's arms. The control unit (15) sends this data to the server (2) via the communication unit (14). The server (2) determines this as a 'fear / anxiety' state and issues a 'Comfort Mode' command.) Operation (17) performed: Heat output: The PTC heater of the heating unit (11) is activated to gradually raise the temperature to 38 degrees, which is slightly higher than body temperature. (Induces psychological stability through warmth) LED output: The LEDs on the eyes or chest slowly blink in a warm orange color (breathing rhythm). Voice output: Outputs a calm, low-toned voice saying "It's okay, I'll protect you" through the audio unit (13). Motor output: Creates a regular vibration (Heartbeat) similar to a heartbeat inside the doll to help the child calm down.

[0025] The camera (vision) sensor recognizes the user's facial expressions (smile, frown) or performs eye contact, and when the child laughs, the doll also makes a laughing sound or turns its head in the direction the child is looking.

[0026] The biosignal sensor (heart rate) measures the user's heart rate when holding the doll (1) to precisely measure the level of tension, and if the user's heart rate is high (tension), it maintains calming music and a warming mode until the heart rate decreases.

[0027] The gyro / accelerometer sensor detects the movement of the doll (1) (shaking, throwing, dropping), and if the doll is handled roughly, it responds with "It hurts," or if it is shaken gently (cradle), it switches to sleep mode.

[0028] The olfactory sensor (gas / odor) detects air quality or specific odors in the house, sounds an emergency alert when smoke or gas odors are detected, or sends out a message encouraging ventilation when there is a musty smell.

[0029] The joint drive actuator provides physical movement to the arms, head, and ears of the doll (1), and when the user says "Hug me," the arms open, or when the hug sensor detects, the doll also performs a Hugging Back action.

[0030] The fragrance dispensing module provides aromatherapy effects according to the emotional mode; it finely diffuses a lavender scent in 'Sleep Mode' and a citrus scent in 'Vitality Mode'. (Promotes fragrance diffusion in conjunction with the heating unit)

[0031] A beam projector (miniature) projects images / light from the belly or head of the doll (1) to the ceiling, displays constellation images on the ceiling when inducing sleep, or projects related images onto the wall when telling a story.

[0032] The smart fabric (color-changing material) causes the color of the doll's (1) fur or clothes to change according to temperature or emotion, and the fur color gradually changes to pink when the doll is in a good mood and to a calm blue when comforting the user.

[0033] The wireless charging module charges simply by placing it on the stand without a cable connection, and when the doll says "I'm hungry (low battery)," the child lays it on the doll bed (charging pad) to encourage charging play.

[0034] The UV sterilization module is used to maintain hygiene by connecting to the inside of the doll (1) or the stand, and since the doll (1) is frequently touched and hugged, the built-in UV lamp operates during charging or sleep mode to suppress the growth of surface bacteria.

[0035] The GPS / location tracking module locates the doll (1) and prevents it from getting lost (when going out), and if the child goes out with the doll and loses it, it finds the location via an app or sends a notification to the guardian if the child moves out of a certain area.

[0036] Figure 3 is a block diagram showing the configuration of the server of the present invention.

[0037] Referring to FIG. 3, the server (2) includes a communication unit (21); a storage unit (22); and a control unit (23). The control unit (23) performs sensing (24) that detects voice, hug, touch, and environment using a sensor; and performs an operation (25) that performs voice output, LED output, heat output, and motor output.

[0038] The server (2) acts as the 'brain' of the system, interpreting data sent by the doll (1), determining what action to take, and issuing commands. The communication unit (21) is an interface for transmitting and receiving data to and from the doll (1), the store (3), the manager terminal, etc. (Uses MQTT, HTTP, WebSocket protocols, etc.). The storage unit (22) stores data necessary for system operation. DB: User profile, emotion records, character personality data; Content: Audio files, motion pattern library. The control unit (23) performs emotion analysis and thermal control, and processes the overall logic. The control unit (23) performs the main functions of sensing (24) and operation (25), where 'sensing' in the server refers to data interpretation rather than physical detection, and 'operation' refers to command generation rather than physical movement. Sensing (24) performs data interpretation and emotion judgment: receives raw data (voltage value, waveform, etc.) transmitted from the doll (1) and converts it into meaningful information, and when microphone sensor values ​​are received, analyzes frequency / decibels and judges it as "high-pitched laughter (joy)," and when pressure sensor values ​​are received, analyzes duration and intensity and judges it as "long-term tight hug (attachment / anxiety)." Action (25) performs control command generation and transmission: determines the optimal action the doll should perform according to the judged emotion and generates a command packet, and when joy is judged, generates LED: Rainbow, heat: maintain, motor: rhythm vibration commands.

[0039] "When the user speaks to the doll while feeling depressed," data reception (communication unit (21)): Voice data ("I had a really hard time today...") and weak touch data from the doll (1) are received by the server (2). The control unit (23) senses (24): The voice analysis engine analyzes the text content ("It was hard") and voice tone (low and slow) to derive the emotion of 'sadness'. The character engine loads the personality of 'Mashi (comforting character)'. Command generation (control unit (23) - operation (25)): Heat control: Command to set to 39 degrees, which is warmer than body temperature, to relieve depression; Voice control: Select the comforting script "What happened? I'll listen."; Command transmission: These commands are transmitted to the doll through the communication unit (21).

[0040] The Natural Language Processing (NLP) engine understands the context of the user's speech to enable free-flowing conversation, and if the user says "I'm hungry," it prompts the conversation by responding not with a simple reply, but with "Oh dear, have you had dinner? Shall I recommend something delicious?"

[0041] The personalized learning module learns user-specific habits and preferences using machine learning, and automatically reflects the fact that User A likes hot temperatures (40 degrees) when sad, while User B likes lukewarm temperatures (36 degrees).

[0042] The emotion history analyzer (Big Data) analyzes long-term trends in emotional changes to generate a report and provides analysis results, such as "Anxiety levels have been high every night at 10 PM for the past week," to the guardian app or store kiosk.

[0043] The predictive control algorithm predicts the user's pattern in advance and operates preemptively. If the user always hugs the doll at 7:00 PM after work, the heating is turned on in advance from 6:50 PM to prepare it to be warm.

[0044] The OTA (Over-The-Air) manager automatically updates the firmware or character voice pack of the doll (1) remotely, and when the Christmas season comes, the server (2) distributes the 'carol singing function' to the doll in bulk.

[0045] User account / permission management (IAM) distinguishes users and separates data when multiple family members use one doll (1), and distinguishes voice tones when a father holds it and when a child holds it, responding to different character modes.

[0046] The data encryption / security module protects sensitive voice data and privacy information, and performs encryption and anonymization processing before transmission to prevent children's voice data from being leaked externally.

[0047] The smart home (IoT) linkage interface links with other home appliances (lights, speakers, thermostats) in the house, and when the doll enters 'sleep mode', the server (2) instructs the smart light bulb in the room to turn off and the curtains to close.

[0048] The guardian monitoring dashboard allows guardians to check the status of their children or elderly relatives in real time via the app, and sends notifications to the guardian if an elderly person living alone has not interacted with the doll for a long time or if feelings of 'depression' persist.

[0049] The Content Management System (CMS) continuously updates and manages the content provided by the dolls, such as fairy tales, music, and quizzes, and uploads new fairy tale content to the server every week to keep the stories the dolls tell rich.

[0050] The API gateway provides a channel for external developers or services to connect with the doll system, and integrates with a healthcare app to extend functionality so that when the user hugs the doll after a workout, it says, "You worked out hard today!"

[0051] FIG. 4 is a block diagram showing the configuration of the store of the present invention.

[0052] Referring to FIG. 4, the store (3) includes a kiosk (31); a customer section (32); and a control section (33). The control section (33) performs customer response (34) that adds customer-tailored actions to the actions of the doll (1) and the server (2).

[0053] The store (3) utilizes information from the doll (1) and the server (2) to provide customized services and product recommendations to customers and achieve actual commercial goals.

[0054] The kiosk (31) is an unmanned information terminal that allows customers to directly interact with the system, register dolls, check personal emotion records, and view customized product recommendations.

[0055] The customer department (32) includes a physical space and personnel where customer service and product sales take place, and includes display stands, consultation staff, and a payment system.

[0056] The control unit (33) controls the operation of the store and customer response based on data received from the server (2).

[0057] The control unit (33) receives the results of the sentiment analysis from the server (2) and performs customer-customized actions (34) in the actual store environment.

[0058] Customer response (34) provides personalized services through the store environment, personnel, and system based on the results of the emotion / mode analysis of the server (2), and provides relevant product information via kiosk or mobile according to the customer's current emotion (e.g., excitement / energy mode), Emotion: Energy -> "You are having an active day today! Recommend a thermal / cooling tumbler suitable for outdoor activities," adjust the store environment, finely adjust the lighting, music (BGM), temperature, etc. inside the store according to the customer's emotional state, Emotion: Peace / Sleep mode -> Change the store BGM to calm classical music and lower the brightness, provide a customized response script to the staff based on the customer's emotional state, Emotion: Sadness detection -> Give instructions to the staff such as "Please respond in a calm and comforting tone with a warm drink," and immediately issue a relevant product discount coupon via kiosk or app only to customers in a specific emotional state, Emotion: Anxiety detection -> Provide a 10% discount coupon for anxiety-relieving aroma oil.

[0059] The experience zone linkage system reflects the mode (temperature, color) of the doll (1) directly into the experience space (e.g., sleeping space, rest chair) in the store, allowing customers to directly experience the doll's condition, and activates the heated seat of the experience chair when in the calming mode.

[0060] The emotional history visualization display shows a graph of a customer's emotional changes over a week through a kiosk (31) or personal app and provides lifestyle improvement tips based on this.

[0061] The social media integration module provides features that allow customers to easily share records of interactions with the doll or recommended products on social media.

[0062] The VIP / regular customer customized notification system immediately displays a welcome message tailored to the doll's mode on a kiosk or employee's tablet based on server information when a specific customer visits the store.

[0063] The in-store voice assistant function responds based on server information when a customer asks, "How is my doll doing right now?" via microphones and speakers installed throughout the store, even without the doll present.

[0064] The inventory / sales linked analysis system analyzes customer sentiment patterns and products that led to actual purchases in real time to optimize future inventory quantity forecasting and shelf placement.

[0065] The employee training and evaluation system evaluates whether the server (2) complies with customer response guidelines and measures whether the employee's response style had a positive effect on the customer's emotional state (change in doll mode).

[0066] CCTV-based customer movement analysis is linked with CCTVs in the store (3) to determine which product corner a customer stayed in the longest when in a specific emotional state, thereby increasing product placement efficiency.

[0067] The online shopping mall integration gateway allows you to immediately add recommended products from the kiosk or app in the store (3) to the online shopping mall shopping cart and conveniently purchase them from home.

[0068] The digital signage (advertising board) automatic switching system automatically changes advertising content based on the average emotional state of customers detected in real-time by the digital advertising board in the store (3), and broadcasts advertisements for healing products that provide a sense of calm when many are in an excited mode.

[0069] The mobile app-based pickup notification system informs customers of the preparation time and pickup location via the mobile app when they reserve or purchase doll-related services.

[0070] FIG. 5 is a block diagram showing the configuration of the character part of the present invention.

[0071] Referring to FIG. 5, the character unit (171) includes a feature DB (1711) that stores character features of Fruity, Lolly, Mashi, and Puppy; a learning unit (1712) that learns information from the feature DB (1711); an inference unit (1713) that infers character features from input data; and an output unit (1714) that outputs the inferred features.

[0072] A character unit (171) that may be located inside a control unit (15) or a server (2) selects one of the doll characters (Frutti, Lolly, Mashi, Puppy) according to the user's emotional state and interaction data, or determines a customized response that matches the characteristics of the character.

[0073] The feature DB (1711) is a database that stores information such as the unique personality, preferred reactions (temperature, voice), and behavior patterns of all characters in the system.

[0074] The learning unit (1712) learns user interaction data received from the feature DB (1711) and the server (2), and learns patterns to determine what reaction each character should have in a given situation and optimizes it.

[0075] The inference unit (1713) determines (infers) the most suitable character response for the current situation using learned information based on input data (emotion, touch, environment).

[0076] The output unit (1714) transmits the final character action command (temperature, voice, motor) determined by the inference unit (1713) to the doll's operation unit (17) or the server's operation unit (25).

[0077] In the example of Feature DB (1711), Fruity is cheerful and lively, energetic, specialized in excitement / vitality mode, has little temperature change, and has a bright and fast voice; Rolly is calm and rational, seeks stability, specialized in anxiety / sleep mode, maintains constant warmth, and has logical and calm conversation; Masa is emotional and comforting, has high empathy, specialized in sadness / comfort mode, has a rapid temperature rise, and has an empathetic low-tone voice; and Puppy is loyal and responsive, sensitive to touch, responds to all modes, and provides immediate motor / vibration output upon detecting hugs / touches.

[0078] Input data is received when the 'Mashi' character responds to a sad user, and the inference unit (1713) receives emotion data ("sadness") and interaction data ("hug pressure: medium") from the server / doll control unit. The feature DB (1711) queries the feature DB and confirms that Mashi has the features of "sadness / comfort mode specialized" and "rapid heat increase." The learning unit (1712) learns based on past data that user satisfaction was highest when Mashi applied rapid heat to 39 degrees upon detecting "sadness." Using this learned information, the inference unit (1713) infers "Mashi mode, heat 39 degrees, comfort voice pattern" as the final response in the current situation. The output unit (1714) transmits the inferred command to the doll's operation unit (17). The heating unit (11) heats up to 39 degrees, and the audio unit (13) outputs Mashi's comforting voice.

[0079] The emotion transfer / strategy module sets a specific target emotion (e.g., inducing a sense of stability within 30 minutes) beyond a simple reaction, and establishes and executes a multi-step response strategy to achieve it.

[0080] The relationship / intimacy database records the intimacy score with each character (Frutti, Mashi, etc.) for each user, reacts cautiously when the intimacy is low, and attempts deeper interaction when it is high.

[0081] The Persona Creation / Combination Module dynamically generates a new custom character persona by combining Fruity's liveliness and Lolly's calmness based on user preference, in addition to the existing four fixed characters.

[0082] Ethics and safety control filters serve as safeguards that halt responses or trigger emergency alerts in situations where sentiment analysis results could harm the user (e.g., extreme anger / implyations of self-harm) or inappropriate conversation requests.

[0083] The situation / time-based behavior scheduler utilizes environmental sensor information (illumination, time zone) to schedule actions to occur voluntarily at specific times without user interaction, and at 7 AM, 'Frutti' outputs an energetic notification voice.

[0084] The interactive content database stores thousands of customized dialogue scripts, humor, and knowledge for each character, ensuring that responses to user questions remain rich rather than simply repetitive.

[0085] The multimodal output mapping module provides a unified emotional experience by simultaneously and harmoniously mapping an inferred emotional state (e.g., joy) to all of the doll's output means, such as heat, voice, LED, and motor, and provides Joy -> (warm heat) + (fast voice) + (bright LED) + (fast motor vibration).

[0086] The user feedback collection module collects data to improve the accuracy of the learning unit (1712) by asking the user for simple feedback, such as "Was this reaction good?", after the doll reacts.

[0087] The character version management system manages each character's personality and reaction algorithms by version, ensuring stability between versions during system updates and easily identifying the impact of new feature introductions.

[0088] The cloud synchronization module provides a consistent experience by synchronizing the user's intimacy and learned character features with cloud storage, even if the doll is changed or used in a different environment.

[0089] The AI ​​conversational engine generates new sentences appropriate to the situation in real-time, in addition to fixed scripts, making character dialogue more human and natural.

[0090] The present invention is not limited to the specific preferred embodiments described above, and any person skilled in the art to which the invention pertains can make various modifications without departing from the essence of the invention as claimed in the claims, and such modifications will be within the scope of the claims. Explanation of the symbols

[0091] 1: Doll 2: Server 3: Store 11: Heating part 12: Sensor section 13: Audio Section 14: Communications Department 15: Control unit 16: Sensing 17: Action 21: Communications Department 22: Storage section 23: Control unit 24: Sensing 25: Action 31: Kiosk 32: Customer Department 33: Control unit 34: Customer Response 171: Character Department 1711: Feature DB 1712: Learning Department 1713: Inference section 1714: Output section

Claims

Claim 1 An AI emotion recognition-based heat control plush doll comprising: a doll that performs emotion, heat generation, character, store linkage, and multi-mode (peace, excitement, sleep, vitality); a server that performs an emotion analysis engine, a character engine, a heat control algorithm, and a mode switching algorithm; and a store that responds to customers; wherein the store comprises a kiosk; a customer section; and a control section; wherein the control section performs customer response that adds customer-tailored actions to the operations of the doll and the server; and wherein the character section of the doll comprises a feature DB that stores character features of Fruity, Lolly, Mashi, and Puppy; a learning section that learns information from the feature DB; an inference section that infers character features from input data; and an output section that outputs the inferred features. Claim 2 The AI ​​emotion recognition-based heat control plush doll according to claim 1, characterized in that the doll outputs heat, visual expressions, and voices by considering the direction of change, duration, or recurrence of emotion, and outputs heat differently according to the emotional state, the temperature rise curve, maintenance time, or decrease pattern, visually outputs an intermediate expression state corresponding to the transition between emotional states, and outputs voices with different tones, speeds, or dialogues according to the emotional state and character. Claim 3 In claim 1, the doll comprises: a heating unit including a PTC heating element; a sensor unit including a voice emotion sensor, a hug pressure sensor, a touch sensor, and an environment sensor; an audio unit that outputs voice; a communication unit; and a control unit; wherein the control unit performs sensing that detects voice, hug (torso, chest), touch (head, arm, chest), and environment (temperature, illuminance, noise) using a sensor; and operations that perform voice output, LED output, heating output, and motor output. Claim 4 In claim 1, the server comprises a communication unit; a storage unit; and a control unit; and the control unit is characterized by performing sensing that detects voice, hug, touch, and environment using a sensor; and performing operations that perform voice output, LED output, heat output, and motor output. Claim 5 delete

Citation Information

Patent Citations

  • Action control system

    JP2024159561A

  • Character toy capable of communicating with portable terminal and childcare training system using the same

    KR101855178B1

  • Smart character toy embedded artificiall intelligence function

    KR1020200050441A