System

Abstract

A system is provided.SOLUTION: A system comprising means for analyzing facial expression data of a user using an emotion recognition algorithm to identify an emotional state, means for analyzing gesture data of the user to recognize meaningful actions, means for analyzing voice data to identify speech content and emotion tones, means for generating an AI model that generates appropriate responses based on analysis results, means for providing the generated responses to the user, and means for collecting feedback from the user and using the feedback as training data for the model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] The current challenge is the difficulty of achieving smooth communication with people with disabilities and their families, people with dementia, and animals with highly developed brains. In particular, conventional communication methods are unable to accurately understand and convey emotions and intentions, resulting in insufficient communication. By solving this problem, it is necessary to improve the quality of social life and reduce stress. [Means for solving the problem]

[0005] The present invention solves a specific problem by providing a system including: means for analyzing a user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to distinguish between spoken content and emotional tone; a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses to the user; and means for collecting user feedback and using it as training data for the model. Furthermore, the system further includes means for analyzing the user's gaze data using eye-tracking technology to recognize intent, and means for providing the generated responses as voice messages, thereby achieving even more accurate communication.

[0006] An "emotion recognition algorithm" is an algorithm for analyzing a user's facial expression data and identifying their emotional state.

[0007] "Gesture data" refers to the movement of a user's hands or body captured as digital data.

[0008] "Voice data" refers to a digital representation of a user's speech signal, including the speech content and emotional tone of the speech signal.

[0009] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on the user's emotional state and intentions.

[0010] "Eye tracking technology" is a technology that analyzes a user's gaze data to recognize their intentions and points of gaze.

[0011] "Feedback" refers to the evaluation or opinion a user gives of a system's response.

[0012] A "response" is a message or action that the system provides to the user based on the analysis results. [Brief explanation of the drawings]

[0013] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0014] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0015] First, the terms used in the following description will be explained.

[0016] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0017] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0018] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0019] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0020] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0021] [First embodiment]

[0022] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0023] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0024] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0025] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0026] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0027] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0028] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0029] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0030] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0031] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0032] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0033] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0034] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user, and its details are as follows.

[0035] Collecting input

[0036] User

[0037] The user provides input to the device, for example, by using a camera and microphone to record their facial expressions and voice.

[0038] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time.

[0039] Terminal

[0040] The device includes hardware for collecting facial expression data, gesture data, and voice data from the user, including a camera, a microphone, and an eye-tracking sensor.

[0041] server

[0042] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0043] Data Preprocessing

[0044] server

[0045] The server pre-processes the received data, which includes noise removal, necessary data format conversion, and extraction of the user's facial features.

[0046] Examples: Removing background noise from audio data to clarify what the user is saying and the tone. Identifying the position of the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.).

[0047] Analysis and Recognition

[0048] server

[0049] The server uses emotion recognition algorithms to analyze the user's facial expression data and identify their emotional state, gesture recognition algorithms to recognize meaningful actions from the user's hand and body movements, a voice analysis engine to identify the content and emotional tone of the user's speech from the voice data, and eye-tracking technology to analyze the user's gaze data and recognize their intentions.

[0050] For example: If a user smiles, an emotion recognition algorithm will determine this as "happiness." If a user waves, a gesture recognition algorithm will recognize this as "greeting." From the user's voice, it will be analyzed that the user said "hello" and that the emotional tone is friendly. Eye-tracking technology will recognize that the user is looking at a specific icon on the screen.

[0051] Communication generation

[0052] server

[0053] The server generates an appropriate response based on the analysis results, leveraging generative AI models to create the optimal response based on the user's emotional state and intent. This response can be in the form of audio, text, or video.

[0054] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[0055] Terminal

[0056] The terminal provides the generated response sent from the server to the user, either played through the speaker in the case of a voice message or displayed on the screen in the case of a text message.

[0057] Gathering feedback and self-learning

[0058] User

[0059] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[0060] server

[0061] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[0062] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0063] The above is an embodiment of the present invention. This system solves the problems that existed in conventional communication methods, and enables smooth communication, particularly with people with disabilities and animals.

[0064] The processing flow will be explained below.

[0065] Step 1:

[0066] User

[0067] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[0068] Step 2:

[0069] Terminal

[0070] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[0071] Step 3:

[0072] Terminal

[0073] The device sends the collected data to a server, where video data, audio data, and gaze data are sent together.

[0074] Step 4:

[0075] server

[0076] The server preprocesses the received data: for video data, noise reduction is performed and a face detection algorithm is used to identify the location and features of the user's face; for audio data, a noise reduction filter is applied to extract clear audio information.

[0077] Step 5:

[0078] server

[0079] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. It then uses a gesture analysis algorithm to analyze the user's movements and recognize specific actions. It then uses a voice analysis engine to analyze the voice data and identify the content and emotional tone of the speech. Based on the gaze data, it uses eye-tracking technology to analyze the user's intentions.

[0080] Step 6:

[0081] server

[0082] The server uses a generative AI model to generate an appropriate response based on the analysis results, which is encoded in the optimal format (audio, text, video, etc.) depending on the user's emotional state and intent.

[0083] Step 7:

[0084] Terminal

[0085] The terminal receives the generated response sent by the server, which is played through the speaker if it is a voice message, or displayed on the screen if it is a text message.

[0086] Step 8:

[0087] User

[0088] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[0089] Step 9:

[0090] server

[0091] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve its accuracy.

[0092] Example 1

[0093] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0094] Conventional communication systems lack the ability to comprehensively analyze multiple input data, including not only a user's facial expressions and voice, but also their gaze and gestures. This makes it difficult to accurately understand the user's emotional state and intentions and generate appropriate responses based on them. Furthermore, they lack a mechanism for effectively utilizing user feedback and continuously improving the system's response accuracy. In particular, there have been technical challenges in achieving smooth communication with people with disabilities and animals.

[0095] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0096] In this invention, the server includes: means for collecting a user's facial expression data using a camera and microphone; means for transmitting the collected data to the server; means for performing noise reduction and format conversion on the data received by the server to extract the user's facial features; means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing audio data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means for including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses to the user; and means for collecting user feedback and using it as training data for the model. This enables the system to comprehensively analyze a variety of user input information and generate more accurate responses. Furthermore, the system's accuracy can be improved using user feedback, enabling smoother communication, particularly with people with disabilities and animals.

[0097] "Facial expression data" is video information captured using a camera to show the facial expressions of a user.

[0098] "Voice data" refers to acoustic information captured using a microphone, such as the content of a user's speech and the tone of their voice.

[0099] "Noise reduction" is the process of removing background noise and unwanted sounds from audio data.

[0100] "Format conversion" is the operation of converting collected data into a format that is easy to analyze.

[0101] "Feature points" are information that indicates the main points of the user's face (eyes, mouth, eyebrows, etc.).

[0102] An "emotion recognition algorithm" is a method or program for analyzing facial expression data to identify a user's emotional state.

[0103] "Gesture data" is video information that captures the movements of the user's hands and body.

[0104] A "gesture recognition algorithm" is a method or program for analyzing gesture data and recognizing meaningful actions from user movements.

[0105] "Gaze data" is information captured using eye-tracking technology to determine where a user is looking.

[0106] "Eye tracking technology" is a technology for capturing the direction of a user's gaze and point of gaze.

[0107] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[0108] "Feedback" is the process of collecting user feedback and opinions on responses.

[0109] "Training data" is collected feedback information used to improve the performance of a model.

[0110] A "response" is a message or instruction that the generative AI model provides to the user based on the analysis results.

[0111] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user.

[0112] Collecting input

[0113] The user uses a camera and microphone to provide their facial expressions and voice to the device. For example, when the user speaks in front of the camera, their facial expressions and voice are captured in real time.

[0114] The device includes hardware for collecting facial, gesture, and voice data from the user, including a camera, microphone, and eye-tracking sensors, such as a Logitech C920 camera, a Blue Yeti microphone, and a Tobii Eye Tracker 5.

[0115] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0116] Data Preprocessing

[0117] The server then performs noise reduction and format conversion on the received data, which includes removing background noise from the audio data and making the content and tone of the user's speech clearer. It also locates the user's face from the video data and extracts feature points (eyes, mouth, eyebrows, etc.).

[0118] Analysis and Recognition

[0119] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. For example, the Microsoft Azure Face API is used. It also uses a gesture recognition algorithm to recognize meaningful actions (e.g., greeting, pointing) from the user's hand and body movements. It uses a voice analysis engine (e.g., Google Speech-to-Text) to identify the content and emotional tone of the user's speech from the voice data. Furthermore, it uses eye-tracking technology to analyze the user's gaze data and recognize the user's intentions.

[0120] Communication generation

[0121] The server generates an appropriate response based on the analysis results. It uses a generative AI model (e.g., OpenAI GPT-4) to create the optimal response (voice, text, or video) based on the user's emotional state and intent. For example, in response to the utterance "Hello," it generates the text message "Hello! How is your day?"

[0122] The terminal provides the generated response sent from the server to the user, either playing it through the speaker in the case of a voice message or displaying it on the screen in the case of a text message.

[0123] Gathering feedback and self-learning

[0124] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[0125] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[0126] For example, if a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0127] An example of a prompt sentence is a user's utterance, "What is the weather today?" The server analyzes this voice data and generates a response, "The weather today is sunny."

[0128] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0129] Step 1:

[0130] Providing user input

[0131] The user uses a camera and microphone to provide their facial expressions and voice to the device. The user speaks in front of the camera, and the video and audio are captured in real time. Specifically, a camera (Logitech C920) and a microphone (Blue Yeti) are used. The input of this step is the user's facial expressions and voice, and the output is the captured raw video and audio data.

[0132] Step 2:

[0133] Data collection

[0134] The device uses a camera, microphone, and eye-tracking sensor to collect facial expression data, voice data, and gaze data. Tobii Eye Tracker 5 is used for eye-tracking. For example, when the user looks to the right, the gaze data is captured. The input of this step is the raw data output from step 1, and the output is organized facial expression data, voice data, and gaze data.

[0135] Step 3:

[0136] Sending data

[0137] The device sends the collected data to the server in real time. A secure protocol (e.g., HTTPS) is used for communication. The input of this step is the data collected in step 2, and the output is the data transfer to the server.

[0138] Step 4:

[0139] Temporary data storage

[0140] The server temporarily stores the data received from the terminal in storage. This process allows for efficient management of large amounts of data. The input to this step is the transferred data, which is the output of step 3, and the output is the data stored in storage.

[0141] Step 5:

[0142] Noise reduction and format conversion

[0143] The server removes background noise from the audio data and unnecessary parts from the video data to extract facial features. A noise-canceling algorithm is used to remove noise from the audio data, and a video analysis algorithm is used to extract facial features. The input to this step is the data saved in step 4, and the output is clear pre-processed audio data and facial expression data with extracted features.

[0144] Step 6:

[0145] Emotion and movement analysis

[0146] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. Specifically, the Microsoft Azure Face API classifies emotions into categories such as "happiness," "sadness," and "surprise." It also uses a gesture recognition algorithm to analyze the user's movements. The input of this step is the clear data output from Step 5, and the output is the user's emotional state and gesture identification results.

[0147] Step 7:

[0148] Audio analysis

[0149] The server uses a speech recognition engine to analyze the speech content and emotional tone from the audio data. Google Speech-to-Text is used for this process. The input is the audio data processed in step 5. For example, "Hello" and a friendly tone are identified. The output is text data of the speech content and emotional tone.

[0150] Step 8:

[0151] Line of sight analysis

[0152] The server uses eye-tracking technology to analyze the user's gaze data and recognize their intention. Specifically, it analyzes data showing the user's gaze on a specific icon on the screen. The input is the gaze data from step 2, and the output is text data about the user's gaze area and intention.

[0153] Step 9:

[0154] Generating a response

[0155] The server generates an appropriate response based on the analysis results. Using a generative AI model (e.g., OpenAI GPT-4), it generates the optimal response based on the user's emotional state and intention. For example, in response to the utterance "Hello," a response such as "Hello! How is your day?" is generated. The input is the analysis results from steps 6 to 8, and the output is the generated voice or text message.

[0156] Step 10:

[0157] Sending a Response

[0158] The server sends the generated response to the terminal. The data is transferred to the terminal in the appropriate format (audio, text, video). The input is the data generated in step 9, and the output is the response data to the terminal.

[0159] Step 11:

[0160] Providing a response

[0161] The terminal provides the generated response received from the server to the user, either playing it through the speaker if it is a voice message or displaying it on the screen if it is a text message. The input is the response data from step 10, and the output is the response the user receives.

[0162] Step 12:

[0163] Providing feedback

[0164] The user provides feedback on the system's response. Using a smartphone or PC, the user evaluates whether the response was appropriate. The input for this step is the user's impression, and the output is feedback data.

[0165] Step 13:

[0166] Collecting feedback

[0167] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, which improves the accuracy of the response. The input of this step is the feedback data from step 12, and the output is the training data.

[0168] Step 14:

[0169] Training the model

[0170] The server retrains the generative AI model with the newly collected feedback, which allows the system to self-improve. The input for this step is the training data from step 13, and the output is an updated generative AI model.

[0171] (Application example 1)

[0172] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0173] Conventional customer service systems in brick-and-mortar stores have difficulty accurately grasping customers' emotions and intentions, making smooth communication difficult. Furthermore, they rely heavily on the skills of employees, making it difficult to maintain a consistent level of service quality. When employees are dealing with a large number of customers, they need to respond efficiently and appropriately, but there has been no effective way to achieve this.

[0174] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0175] In this invention, the server includes: means for analyzing a user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to identify the content of speech and emotional tone; a generative AI model for generating appropriate responses based on the analysis results; means for providing the generated responses to the user; means for collecting feedback from the user and using it as training data for the model; means for capturing customer facial expressions and gestures using a camera; and means for collecting customer voices using a microphone and displaying the generated responses on the smart glasses based on the analysis results. This makes it possible to grasp the customer's emotions and intentions in real time and generate and provide appropriate responses. As a result, high-quality service can be provided and smooth communication can be achieved without relying on the skills of employees.

[0176] An "emotion recognition algorithm" is a technology for analyzing a user's facial expression data and identifying their emotional state.

[0177] "Means for analyzing user gesture data" refers to technology that detects the movements of the user's hands and body and recognizes the meaning of those movements.

[0178] "Means for analyzing voice data" refers to technology that identifies the content and emotional tone of a user's speech and extracts appropriate information.

[0179] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[0180] The "means for providing a response to a user" refers to a technique for transmitting the generated response to a user in the form of voice, text, or the like.

[0181] "Means for collecting feedback and using it as model training data" refers to techniques for collecting user evaluations and using that data to improve the system's performance.

[0182] "Camera-based means" refers to the device or system used to acquire video data.

[0183] "Smart glasses" are devices equipped with a camera, display, microphone, etc., that display various information in real time when worn by a user.

[0184] "Means using a microphone" refers to a device or system for capturing audio data.

[0185] This invention is a customer service support system for brick-and-mortar stores that facilitates smooth communication between store employees wearing smart glasses and customers. The system combines emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, and gaze tracking technology.

[0186] Hardware and Software Configuration

[0187] Smart glasses: Smart glasses are equipped with a camera, microphone, display, and eye-tracking sensor to collect customers' facial expressions, gestures, and voice in real time.

[0188] Server: The server receives the data sent by the smart glasses, analyzes and processes it. Specifically, the following processes are performed:

[0189] Emotion recognition: Based on the video data captured by the camera, an emotion recognition algorithm is run to identify emotions from the user's facial expressions.

[0190] Gesture recognition: Uses technology to recognize user gestures (e.g., hand movements) from video data captured by a camera.

[0191] Voice analysis: Analyzes the content and emotional tone of the user's speech based on the voice data captured by the microphone.

[0192] Eye tracking: Analyzes data acquired from an eye tracking sensor to identify the object and direction the user is looking at.

[0193] Generative AI Model: Uses an artificial intelligence model that generates the optimal response based on the analysis results. This response can be in the form of text or voice.

[0194] Operation procedure and example

[0195] As an example of use in a physical store, the system would be implemented in the following steps:

[0196] 1. User data collection:

[0197] A customer approaches an employee wearing smart glasses and begins a conversation. The camera in the glasses captures the customer's facial expressions in real time, and the microphone collects their voice.

[0198] 2. Data Analysis:

[0199] The collected data is sent to a server, where an emotion recognition algorithm analyzes the facial expression data to identify the customer's emotional state (e.g., joy, surprise), while gesture recognition technology detects hand movements and other gestures, and voice analysis technology analyzes the content and emotional tone of spoken words.

[0200] 3. Generate a response:

[0201] Based on the analysis, the generative AI model generates an appropriate response. For example, if it determines that a customer is tired, it generates a voice message saying, "Would you like to take a break?" This response can be displayed on the smart glasses' display or delivered through the audio speaker.

[0202] 4. Eye-tracking information provision:

[0203] Using eye-tracking technology, if a customer is looking at a particular product, information about that product will be displayed on the smart glasses' display and provided to the customer.

[0204] 5. Gathering feedback and self-learning:

[0205] Feedback from customers and employees is collected and sent to a server, where it is used as training data for a generative AI model, helping to improve the accuracy of the system.

[0206] Examples and prompts

[0207] Examples:

[0208] A customer begins a conversation with an employee wearing smart glasses. The camera captures the customer's face and determines their emotion as "joy." The microphone collects the voice, saying "Hello, how are you today?", which is then analyzed by the smart glasses. Based on the analysis results, the generative AI model suggests, "Hello! Would you like to check out our new promotions?"

[0209] Example prompt sentence:

[0210] A user said: Hello, how are you today?

[0211] Generate an appropriate response to it.

[0212] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0213] Step 1:

[0214] Device data collection

[0215] When a customer approaches an employee wearing smart glasses, the device (smart glasses) uses a camera to capture the user's facial expression and gesture data, and also uses a microphone to collect the user's voice data.

[0216] Input: User facial expression video, gesture video, voice

[0217] Output: Facial expression data, gesture data, voice data

[0218] Step 2:

[0219] Sending data

[0220] The terminal transmits the collected facial expression data, gesture data, and voice data to a server.

[0221] Input: facial expression data, gesture data, voice data

[0222] Output: Facial expression data, gesture data, and voice data sent to the server

[0223] Step 3:

[0224] Server data preprocessing

[0225] The server receives the transmitted data and performs noise removal and necessary data format conversion. Specifically, it removes noise from audio data and extracts facial features from video data.

[0226] Input: Facial expression data, gesture data, and voice data sent to the server

[0227] Output: Noise-removed audio data, facial expression data with extracted features

[0228] Step 4:

[0229] Performing emotion and gesture recognition

[0230] The server uses an emotion recognition algorithm to analyze the facial expression data to identify the user's emotional state, and a gesture recognition algorithm to analyze the gesture data to recognize the user's actions.

[0231] Input: Facial expression data and gesture data with extracted feature points

[0232] Output: Emotional state, recognized gestures

[0233] Step 5:

[0234] Performing audio analysis

[0235] The server uses a voice analysis engine to identify the user's vocal content and emotional tone from the transmitted voice data, filtering out background noise to ensure clear voice data.

[0236] Input: Denoised audio data

[0237] Output: Vocal content, emotional tone

[0238] Step 6:

[0239] Performing eye tracking

[0240] Data obtained from the device's eye-tracking sensor is sent to a server, which then analyzes the user's eye-gaze data to identify the object and direction the user is looking at.

[0241] Input: Gaze data

[0242] Output: Object and direction being gazed at

[0243] Step 7:

[0244] Response generation using generative AI models

[0245] The server uses a generative AI model to generate an appropriate response based on the user's emotional state, recognized gestures, vocal content, emotional tone, and gaze data, which can be in the form of text or voice.

[0246] Input: Emotional state, recognized gestures, vocal content, emotional tone, gaze data

[0247] Output: The generated response (text or audio)

[0248] Step 8:

[0249] Providing a generated response

[0250] The server sends the generated response to the terminal, which then provides the generated response to the user: if it is a voice message, it is played through the speaker, and if it is a text message, it is displayed on the display of the smart glasses.

[0251] Input: The generated response

[0252] Output: The audio or text message provided to the user.

[0253] Step 9:

[0254] Collecting feedback

[0255] Users provide feedback on the responses provided, which is collected and sent to a server that uses it as training data for a generative AI model.

[0256] Input: User feedback

[0257] Output: Feedback data, improving the generative AI model

[0258] Through the above processing steps, a customer service support system is realized that can accurately grasp the emotions and intentions of customers and provide appropriate responses in real time.

[0259] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0260] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's emotional state and intentions with high accuracy and generates appropriate responses based on that. The system functions through interactions between the terminal, the server, and the user, and its details are as follows:

[0261] Collecting input

[0262] User

[0263] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[0264] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time, and it also recognizes whether their gaze is directed in a specific direction.

[0265] Terminal

[0266] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[0267] server

[0268] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0269] Data Preprocessing

[0270] server

[0271] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the user's face and its features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is similarly preprocessed to identify the user's gaze point.

[0272] Examples include: filtering out background noise from audio data to clarify what the user is saying and the tone of their speech; locating the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.); and determining which part of the screen the user is looking at from eye-tracking data.

[0273] Analysis and Recognition

[0274] Emotion Engine

[0275] The emotion engine in the server analyzes the collected facial expression, voice, and gesture data to comprehensively identify the user's emotional state. It also uses gaze data to complement the emotional state.

[0276] For example: If the user is smiling, the emotion engine will determine "happiness." If the user waves, the gesture recognition algorithm will recognize this as a "greeting." From the audio, it will be analyzed that the user said "hello" and that the emotional tone is friendly. From the gaze data, it will be recognized that the user is looking at a specific icon on the screen.

[0277] Analysis results and response generation

[0278] server

[0279] The server uses a generative AI model to generate an appropriate response based on the analysis results of the emotion engine. This response is encoded in the optimal format (audio, text, video, etc.) according to the user's emotional state and intention.

[0280] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[0281] Terminal

[0282] The terminal provides the generated response sent from the server to the user: if it is a voice message, it is played through the speaker, if it is a text message, it is displayed on the screen.

[0283] Gathering feedback and self-learning

[0284] User

[0285] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[0286] server

[0287] The server accumulates the feedback collected from users and stores it in a database. This feedback data is used as training data for the generative AI model and is used to improve the model's accuracy.

[0288] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0289] The above is an embodiment of the present invention. By combining this system with an emotion engine, it is possible to solve the problems that existed in conventional communication methods, and to realize smooth communication, particularly with people with disabilities and animals.

[0290] The processing flow will be explained below.

[0291] Step 1:

[0292] User

[0293] The user speaks to the terminal, for example, saying "hello" and smiling.

[0294] Step 2:

[0295] Terminal

[0296] The device uses a camera and microphone to capture the user's facial expression and voice data in real time, and an eye-tracking sensor to capture the user's gaze data, which is temporarily stored in a buffer.

[0297] Step 3:

[0298] Terminal

[0299] The device sends the collected data, including facial expression data, voice data, and gaze data, to a server.

[0300] Step 4:

[0301] server

[0302] The server pre-processes the received data, removing background noise from the video and applying a face detection algorithm to identify facial features. The audio data is also cleared through a noise reduction filter.

[0303] Step 5:

[0304] server

[0305] The server sends the preprocessed data to the emotion engine, which analyzes the facial expression data to identify the user's emotional state. It also analyzes the gesture data to recognize the user's actions. It also analyzes the voice data using a natural language processing (NLP) engine to analyze the speech content and emotional tone. It also uses gaze data to recognize the user's gaze point and intention.

[0306] Step 6:

[0307] server

[0308] The server uses a generative AI model based on the analysis results of the emotion engine to generate an appropriate response, which can be in the form of voice, text, or video depending on the user's emotional state and intentions.

[0309] Step 7:

[0310] Terminal

[0311] The device receives the generated response sent by the server, plays it over the speaker if it is a voice message, or displays it on the screen if it is a text message.

[0312] Step 8:

[0313] User

[0314] The user provides feedback on the system's response, such as rating it "good" or "needs improvement."

[0315] Step 9:

[0316] server

[0317] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve the accuracy of the system.

[0318] Example 2

[0319] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0320] Conventional communication systems have struggled to accurately recognize a user's emotions and intentions and generate appropriate responses based on them. In particular, the technology for comprehensively analyzing various inputs, such as voice, facial expressions, and gaze, to comprehensively assess the user's state was immature. Furthermore, the functionality for collecting feedback on the generated responses and using it to improve the system was insufficient. This made it difficult to communicate smoothly, particularly with people with disabilities and animals.

[0321] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0322] In this invention, the server includes means for capturing and temporarily storing a user's facial expression data with a camera, means for collecting and temporarily storing the user's voice data with a microphone, means for acquiring and temporarily storing the user's gaze data with an eye tracking sensor, means for preprocessing the received data and identifying the position and feature points of the face using a face detection algorithm, means for applying a noise reduction filter to extract clear voice information, means for an emotion engine to analyze the facial expression data, voice data, and gesture data to identify the user's emotional state, means for generating an appropriate response based on the analysis results using a generative AI model, means for providing the generated response to the user in the form of a voice message, a text message, or a video, and means for collecting user feedback, storing it in a database, and using it as training data for the generative AI model. This makes it possible to recognize user emotions and intentions with high accuracy, generate appropriate responses, and improve the system based on the feedback.

[0323] "User" refers to an individual who operates the system and provides input data.

[0324] "Terminal" refers to a device that collects and temporarily stores a user's facial expression data, voice data, and gaze data.

[0325] "Server" refers to a central processing unit that receives data sent from terminals and performs pre-processing, analysis, and response generation.

[0326] "Facial Expression Data" refers to information about a user's facial expressions captured through a camera.

[0327] "Voice data" refers to acoustic information about a user's speech collected through a microphone.

[0328] "Gaze data" refers to information about the direction of a user's gaze and point of gaze obtained through an eye tracking sensor.

[0329] "Preprocessing" refers to the initial processing of received data, such as noise reduction and feature point extraction.

[0330] A "face detection algorithm" refers to a computational method for identifying the position and features of a face from video data.

[0331] "Noise reduction filter" refers to a processing technology for removing unwanted background noise from audio data and extracting clear audio information.

[0332] "Emotion engine" refers to a program that analyzes facial expression data, voice data, and gesture data to identify the user's emotional state.

[0333] "Gesture data" refers to information about hand and body movements made by a user.

[0334] "Generative AI model" refers to an artificial intelligence model used to generate appropriate responses based on analytical results.

[0335] "Response" refers to the output information, such as a voice message, text message, or video, generated by a generative AI model.

[0336] "Feedback" refers to information regarding evaluations and opinions provided by users to the system.

[0337] "Database" means a data storage system for storing and managing collected feedback and other data.

[0338] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's facial expression data, voice data, and gaze data with high accuracy, and generates appropriate responses based on that data. The system functions through the interaction of three parties: the terminal, the server, and the user.

[0339] Collecting input

[0340] User

[0341] The user speaks or gestures into the device, for example, by waving while saying "hello," and the camera and microphone capture this input.

[0342] Examples:

[0343] When a user turns to the device and says "hello" and waves, the camera captures the user's facial expressions and gestures, and the microphone collects audio.

[0344] Terminal

[0345] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data using an eye-tracking sensor. These input data are temporarily stored in a buffer.

[0346] Examples:

[0347] The device uses a camera and an eye-tracking sensor to capture when the user is looking at a specific part of the screen, and stores this as gaze data.

[0348] server

[0349] The server receives the data sent from the terminal and temporarily stores it in storage for preprocessing.

[0350] Data Preprocessing

[0351] server

[0352] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[0353] Examples:

[0354] The server detects the user's face from the video data and extracts the positions of the eyes and mouth as feature points, and removes background noise from the audio data to extract clear audio information.

[0355] Analysis and Recognition

[0356] Emotion engine (on the server)

[0357] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also uses gaze data to complement the emotional state.

[0358] Examples:

[0359] The emotion engine recognizes when a user is smiling as "happy," when a user is waving as "greeting," and when a user says "hello" in a friendly tone.

[0360] Generating a response

[0361] server

[0362] Based on the analysis results of the emotion engine, the generative AI model generates an appropriate response, which can be encoded in the form of audio, text, video, etc.

[0363] Examples:

[0364] If the server senses that the user is tired, it generates a voice message saying "Do you want to take a break?" If the user says "Hello", it generates a text message saying "Hello! How's your day?"

[0365] Providing a response to the user

[0366] Terminal

[0367] The device receives a response from the server and provides it to the user: a voice message is played over the speaker, a text message is displayed on the screen.

[0368] Examples:

[0369] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays text on the screen saying, "Hello! How are you today?"

[0370] Gathering feedback and self-learning

[0371] User

[0372] The user provides feedback to the system, including the appropriateness of the response and areas for improvement.

[0373] Examples:

[0374] The user gives feedback saying, "This response was appropriate."

[0375] server

[0376] The server stores the feedback in a database and uses it as training data for the generative AI model, improving the accuracy of the system.

[0377] Examples:

[0378] Based on user feedback, the server retrains the generative AI model to improve the accuracy of the next response.

[0379] The above is an embodiment of the present invention. This system enables communication that combines an emotion engine, and is expected to facilitate smooth communication, especially with people with disabilities and animals.

[0380] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0381] Step 1:

[0382] Collecting User Input

[0383] User

[0384] The user speaks or gestures into the device, and the camera and microphone capture the user's input. For example, if the user waves while saying "hello," their facial expression and voice are collected.

[0385] Input: User voice and gestures

[0386] Output: Video and audio data captured by the camera and microphone

[0387] Specific behavior:

[0388] When a user speaks or waves at the device, the camera and microphone capture the input, which is then temporarily stored in a buffer.

[0389] Step 2:

[0390] Data buffering

[0391] Terminal

[0392] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor, and temporarily stores this input data in a buffer.

[0393] Input: Captured video data, audio data, and gaze data

[0394] Output: Buffered input data

[0395] Specific behavior:

[0396] The device uses a camera, microphone, and eye-tracking sensor to collect data and temporarily store it in a buffer.

[0397] Step 3:

[0398] Data Transfer and Storage

[0399] server

[0400] The server receives the input data sent from the terminal and temporarily stores it in storage for preprocessing.

[0401] Input: Input data stored in a buffer

[0402] Output: Data temporarily saved in storage

[0403] Specific behavior:

[0404] The server receives input data from the terminal and temporarily stores it in storage, which is used for subsequent preprocessing.

[0405] Step 4:

[0406] Data Preprocessing

[0407] server

[0408] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[0409] Input: Data stored in storage

[0410] Output: Denoised and feature-pointed data

[0411] Specific behavior:

[0412] The server detects the user's face from the video data and extracts the positions of the eyes and mouth. It also removes background noise from the audio data to extract clear audio. It also identifies the user's gaze point from the gaze data.

[0413] Step 5:

[0414] Emotion analysis and recognition

[0415] Server (emotion engine)

[0416] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also analyzes gaze data to complement the emotional state.

[0417] Input: Preprocessed facial expression data, speech data, and gaze data

[0418] Output: User's emotional state

[0419] Specific behavior:

[0420] The emotion engine identifies a user's emotional state (e.g., joy, surprise, sadness, etc.) from subtle facial movements, vocal tone, and gesture patterns, and complements emotion analysis by using gaze data to understand where the user is looking.

[0421] Step 6:

[0422] Generating a response

[0423] server

[0424] Based on the emotion engine's analysis, a generative AI model generates an appropriate response, which can be encoded in the form of audio, text, or video.

[0425] Input: User's emotional state

[0426] Output: The generated response message

[0427] Specific behavior:

[0428] The generative AI model uses prompts based on the analysis results to generate responses such as "Hello! How are you today?"

[0429] Step 7:

[0430] Providing a response

[0431] Terminal

[0432] The terminal provides the user with the response sent by the server, either a voice message played over the speaker or a text message displayed on the screen.

[0433] Input: The generated response message

[0434] Output: Providing a response to the user (voice, text, etc.)

[0435] Specific behavior:

[0436] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays a text message on the screen saying, "Hello! How's your day going?"

[0437] Step 8:

[0438] Collecting feedback

[0439] User

[0440] The user provides feedback on the system's response, including whether the response was appropriate and what could be improved.

[0441] Input: User feedback (ratings and opinions)

[0442] Output: Feedback information sent to the server

[0443] Specific behavior:

[0444] The user provides feedback such as "this response was appropriate," which is sent to the server via the terminal.

[0445] Step 9:

[0446] Self-study

[0447] server

[0448] The server stores the collected feedback in a database and uses this data as training data for the generative AI model, thereby improving the accuracy of the system.

[0449] Input: User feedback data

[0450] Output: Feedback reflected as training data

[0451] Specific behavior:

[0452] The server analyzes the collected feedback and uses it to retrain the generative AI model so that it generates a better response next time.

[0453] (Application example 2)

[0454] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0455] There is a need to improve work efficiency in factories while ensuring the safety of employees. However, it is not easy for robots to quickly and accurately recognize situations in which employees are in trouble or need assistance, and generate appropriate responses. Furthermore, understanding emotional states and intentions requires integrating and analyzing multiple data sets, including facial expressions, gestures, voice, and gaze, making the process complex. There is a need for a way to solve these problems and enable smooth communication between employees and robots.

[0456] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for analyzing the user's (employee's) facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses as voice messages through a speaker; means for collecting user feedback and using it as training data for the model; and means installed in a robot that accurately identifies the emotional state and intention of the employee through interaction with the employee and generates appropriate responses. This makes it possible to quickly and accurately recognize situations in which the employee is in difficulty and provide appropriate support. This improves work efficiency in the factory and ensures employee safety.

[0457] An "emotion recognition algorithm" is a computational method for analyzing facial expression data to identify a user's emotional state.

[0458] "Gesture data" is data that includes the user's movements and hand movements, and is analyzed to recognize meaningful actions.

[0459] "Voice data" refers to data containing sounds and words uttered by a user, and is used to identify the content and emotional tone of the speech.

[0460] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analyzed data.

[0461] "Eye tracking technology" is a technology that acquires a user's gaze data and analyzes the direction of their gaze and the point of focus.

[0462] A "robot" is an automated mechanical device used in a factory to interact with workers.

[0463] "Feedback" refers to the ratings and opinions collected from users that are used as training data for generative AI models.

[0464] "Interaction" is the exchange of information between a user and a system (or robot).

[0465] This invention is a system that uses emotion recognition algorithms and generative AI models to achieve smooth communication between robots and workers in factories. This system is implemented by combining the following hardware and software.

[0466] Hardware

[0467] Camera: A device used to capture facial expressions and gestures of employees. Used to obtain video data.

[0468] Microphone: A device for collecting employee voice data. Used to analyze speech content and emotional tone.

[0469] Eye tracking sensor: Used to capture employee gaze data and analyze user intent.

[0470] Speaker: A device for providing generated audio messages to employees.

[0471] Robot: An automated mechanical device that interacts with workers in a factory.

[0472] software

[0473] Python: A programming language for implementing the entire program.

[0474] TensorFlow: A machine learning library for building emotion recognition algorithms and generative AI models.

[0475] OpenCV: A library for preprocessing video data and identifying facial positions and feature points.

[0476] PyAudio: A library for preprocessing and denoising audio data.

[0477] Eye tracking library: A library for acquiring gaze data and analyzing user intent.

[0478] System Operation

[0479] 1. Collecting input

[0480] The camera captures the employee's facial expression data and uses OpenCV to detect the face position and feature points.

[0481] The microphone collects audio data and PyAudio is used to filter out background noise.

[0482] An eye-tracking sensor captures the employee's gaze data.

[0483] 2. Data Preprocessing

[0484] Noise is removed from the video data using OpenCV, and facial feature points are identified.

[0485] The audio data is denoised using PyAudio to extract clear audio information.

[0486] The gaze data is preprocessed by an eye-tracking library to identify the gaze point.

[0487] 3. Analysis and Recognition

[0488] Emotion recognition algorithms analyze facial, voice, and gesture data to identify the employee's emotional state, and also analyze gaze data to complement the emotional state.

[0489] 4. Analysis results and response generation

[0490] The server uses a generative AI model to generate an appropriate response based on the analysis of the emotion recognition algorithm, which can be in the form of a voice message, text, or other format.

[0491] 5. Gather feedback and self-study

[0492] User feedback is sent to the server, and the generative AI model uses this feedback to self-train.

[0493] Specific examples

[0494] If an employee says with a troubled expression, "I need another person to do this task," the robot will respond by generating a voice message saying, "Would you like additional support?" This action allows the employee to quickly receive the necessary support, improving work efficiency within the factory.

[0495] Prompt Sentence Examples

[0496] "The user (employee) looks troubled and says, 'I need another person to do this task.' What would a generative AI model generate as an appropriate response to this situation?"

[0497] This system combines emotion recognition algorithms and generative AI models to enable smooth communication between workers and robots, which is expected to improve work efficiency and worker safety in factories.

[0498] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0499] Step 1:

[0500] The user provides input by speaking into the camera and microphone or waving their hand, while simultaneously capturing gaze data using an eye-tracking sensor. This allows the user to generate video, audio, and gaze data.

[0501] Step 2:

[0502] The device captures the user's facial expression data with a camera and detects the position and feature points of the face using OpenCV. The input is video data, and the output is the position and feature points of the face.

[0503] Step 3:

[0504] The device collects the user's voice data with a microphone and uses PyAudio to filter out background noise. The input is the voice data, and the output is clear, noise-removed voice information.

[0505] Step 4:

[0506] The device acquires gaze data using an eye-tracking sensor and uses an eye-tracking library to identify the gaze direction and gaze point. The input is the gaze data, and the output is the gaze direction and gaze point.

[0507] Step 5:

[0508] The data collected by the device (facial expression data, voice data, gaze data) is temporarily stored in a buffer and then sent to the server. The input is a set of preprocessed data, and the output is the data sent to the server.

[0509] Step 6:

[0510] The server preprocesses the received data and uses an emotion recognition algorithm to identify the user's emotional state. The input is the received data, and the output is the identified emotional state (e.g., "confused").

[0511] Step 7:

[0512] The server uses a generative AI model to generate an appropriate response based on the analysis results. The input is the emotional state identification result and the prompt sentence, and the output is the generated response (e.g., "Would you like to arrange additional support?").

[0513] Step 8:

[0514] The server sends the generated response to the terminal, which then provides it to the user: a voice message played through the speaker, a text message displayed on the screen. The input is the generated response, and the output is the response provided to the user.

[0515] Step 9:

[0516] The user provides feedback such as evaluation and opinion on the system's response. The input is the user's feedback, which is sent by the terminal to the server.

[0517] Step 10:

[0518] The server collects feedback from users and uses it as training data for the model. This allows the generative AI model to self-learn and improve its accuracy. The input is the feedback data, and the output is an updated generative AI model.

[0519] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0520] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0521] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0522] [Second embodiment]

[0523] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0524] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0525] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0526] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0527] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0528] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0529] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0530] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0531] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0532] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0533] In the smart glasses 214, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0534] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0535] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user, and its details are as follows.

[0536] Collecting input

[0537] User

[0538] The user provides input to the device, for example, by using a camera and microphone to record their facial expressions and voice.

[0539] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time.

[0540] Terminal

[0541] The device includes hardware for collecting facial expression data, gesture data, and voice data from the user, including a camera, a microphone, and an eye-tracking sensor.

[0542] server

[0543] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0544] Data Preprocessing

[0545] server

[0546] The server pre-processes the received data, which includes noise removal, necessary data format conversion, and extraction of the user's facial features.

[0547] Examples: Removing background noise from audio data to clarify what the user is saying and the tone. Identifying the position of the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.).

[0548] Analysis and Recognition

[0549] server

[0550] The server uses emotion recognition algorithms to analyze the user's facial expression data and identify their emotional state, gesture recognition algorithms to recognize meaningful actions from the user's hand and body movements, a voice analysis engine to identify the content and emotional tone of the user's speech from the voice data, and eye-tracking technology to analyze the user's gaze data and recognize their intentions.

[0551] For example: If a user smiles, an emotion recognition algorithm will determine this as "happiness." If a user waves, a gesture recognition algorithm will recognize this as "greeting." From the user's voice, it will be analyzed that the user said "hello" and that the emotional tone is friendly. Eye-tracking technology will recognize that the user is looking at a specific icon on the screen.

[0552] Communication generation

[0553] server

[0554] The server generates an appropriate response based on the analysis results, leveraging generative AI models to create the optimal response based on the user's emotional state and intent. This response can be in the form of audio, text, or video.

[0555] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[0556] Terminal

[0557] The terminal provides the generated response sent from the server to the user, either played through the speaker in the case of a voice message or displayed on the screen in the case of a text message.

[0558] Gathering feedback and self-learning

[0559] User

[0560] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[0561] server

[0562] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[0563] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0564] The above is an embodiment of the present invention. This system solves the problems that existed in conventional communication methods, and enables smooth communication, particularly with people with disabilities and animals.

[0565] The processing flow will be explained below.

[0566] Step 1:

[0567] User

[0568] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[0569] Step 2:

[0570] Terminal

[0571] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[0572] Step 3:

[0573] Terminal

[0574] The device sends the collected data to a server, where video data, audio data, and gaze data are sent together.

[0575] Step 4:

[0576] server

[0577] The server preprocesses the received data: for video data, noise reduction is performed and a face detection algorithm is used to identify the location and features of the user's face; for audio data, a noise reduction filter is applied to extract clear audio information.

[0578] Step 5:

[0579] server

[0580] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. It then uses a gesture analysis algorithm to analyze the user's movements and recognize specific actions. It then uses a voice analysis engine to analyze the voice data and identify the content and emotional tone of the speech. Based on the gaze data, it uses eye-tracking technology to analyze the user's intentions.

[0581] Step 6:

[0582] server

[0583] The server uses a generative AI model to generate an appropriate response based on the analysis results, which is encoded in the optimal format (audio, text, video, etc.) depending on the user's emotional state and intent.

[0584] Step 7:

[0585] Terminal

[0586] The terminal receives the generated response sent by the server, which is played through the speaker if it is a voice message, or displayed on the screen if it is a text message.

[0587] Step 8:

[0588] User

[0589] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[0590] Step 9:

[0591] server

[0592] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve its accuracy.

[0593] Example 1

[0594] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0595] Conventional communication systems lack the ability to comprehensively analyze multiple input data, including not only a user's facial expressions and voice, but also their gaze and gestures. This makes it difficult to accurately understand the user's emotional state and intentions and generate appropriate responses based on them. Furthermore, they lack a mechanism for effectively utilizing user feedback and continuously improving the system's response accuracy. In particular, there have been technical challenges in achieving smooth communication with people with disabilities and animals.

[0596] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0597] In this invention, the server includes: means for collecting a user's facial expression data using a camera and microphone; means for transmitting the collected data to the server; means for performing noise reduction and format conversion on the data received by the server to extract the user's facial features; means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing audio data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means for including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses to the user; and means for collecting user feedback and using it as training data for the model. This enables the system to comprehensively analyze a variety of user input information and generate more accurate responses. Furthermore, the system's accuracy can be improved using user feedback, enabling smoother communication, particularly with people with disabilities and animals.

[0598] "Facial expression data" is video information captured using a camera to show the facial expressions of a user.

[0599] "Voice data" refers to acoustic information captured using a microphone, such as the content of a user's speech and the tone of their voice.

[0600] "Noise reduction" is the process of removing background noise and unwanted sounds from audio data.

[0601] "Format conversion" is the operation of converting collected data into a format that is easy to analyze.

[0602] "Feature points" are information that indicates the main points of the user's face (eyes, mouth, eyebrows, etc.).

[0603] An "emotion recognition algorithm" is a method or program for analyzing facial expression data to identify a user's emotional state.

[0604] "Gesture data" is video information that captures the movements of the user's hands and body.

[0605] A "gesture recognition algorithm" is a method or program for analyzing gesture data and recognizing meaningful actions from user movements.

[0606] "Gaze data" is information captured using eye-tracking technology to determine where a user is looking.

[0607] "Eye tracking technology" is a technology for capturing the direction of a user's gaze and point of gaze.

[0608] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[0609] "Feedback" is the process of collecting user feedback and opinions on responses.

[0610] "Training data" is collected feedback information used to improve the performance of a model.

[0611] A "response" is a message or instruction that the generative AI model provides to the user based on the analysis results.

[0612] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user.

[0613] Collecting input

[0614] The user uses a camera and microphone to provide their facial expressions and voice to the device. For example, when the user speaks in front of the camera, their facial expressions and voice are captured in real time.

[0615] The device includes hardware for collecting facial, gesture, and voice data from the user, including a camera, microphone, and eye-tracking sensors, such as a Logitech C920 camera, a Blue Yeti microphone, and a Tobii Eye Tracker 5.

[0616] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0617] Data Preprocessing

[0618] The server then performs noise reduction and format conversion on the received data, which includes removing background noise from the audio data and making the content and tone of the user's speech clearer. It also locates the user's face from the video data and extracts feature points (eyes, mouth, eyebrows, etc.).

[0619] Analysis and Recognition

[0620] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. For example, the Microsoft Azure Face API is used. It also uses a gesture recognition algorithm to recognize meaningful actions (e.g., greeting, pointing) from the user's hand and body movements. It uses a voice analysis engine (e.g., Google Speech-to-Text) to identify the content and emotional tone of the user's speech from the voice data. Furthermore, it uses eye-tracking technology to analyze the user's gaze data and recognize the user's intentions.

[0621] Communication generation

[0622] The server generates an appropriate response based on the analysis results. It uses a generative AI model (e.g., OpenAI GPT-4) to create the optimal response (voice, text, or video) based on the user's emotional state and intent. For example, in response to the utterance "Hello," it generates the text message "Hello! How is your day?"

[0623] The terminal provides the generated response sent from the server to the user, either playing it through the speaker in the case of a voice message or displaying it on the screen in the case of a text message.

[0624] Gathering feedback and self-learning

[0625] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[0626] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[0627] For example, if a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0628] An example of a prompt sentence is a user's utterance, "What is the weather today?" The server analyzes this voice data and generates a response, "The weather today is sunny."

[0629] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0630] Step 1:

[0631] Providing user input

[0632] The user uses a camera and microphone to provide their facial expressions and voice to the device. The user speaks in front of the camera, and the video and audio are captured in real time. Specifically, a camera (Logitech C920) and a microphone (Blue Yeti) are used. The input of this step is the user's facial expressions and voice, and the output is the captured raw video and audio data.

[0633] Step 2:

[0634] Data collection

[0635] The device uses a camera, microphone, and eye-tracking sensor to collect facial expression data, voice data, and gaze data. Tobii Eye Tracker 5 is used for eye-tracking. For example, when the user looks to the right, the gaze data is captured. The input of this step is the raw data output from step 1, and the output is organized facial expression data, voice data, and gaze data.

[0636] Step 3:

[0637] Sending data

[0638] The device sends the collected data to the server in real time. A secure protocol (e.g., HTTPS) is used for communication. The input of this step is the data collected in step 2, and the output is the data transfer to the server.

[0639] Step 4:

[0640] Temporary data storage

[0641] The server temporarily stores the data received from the terminal in storage. This process allows for efficient management of large amounts of data. The input to this step is the transferred data, which is the output of step 3, and the output is the data stored in storage.

[0642] Step 5:

[0643] Noise reduction and format conversion

[0644] The server removes background noise from the audio data and unnecessary parts from the video data to extract facial features. A noise-canceling algorithm is used to remove noise from the audio data, and a video analysis algorithm is used to extract facial features. The input to this step is the data saved in step 4, and the output is clear pre-processed audio data and facial expression data with extracted features.

[0645] Step 6:

[0646] Emotion and movement analysis

[0647] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. Specifically, the Microsoft Azure Face API classifies emotions into categories such as "happiness," "sadness," and "surprise." It also uses a gesture recognition algorithm to analyze the user's movements. The input of this step is the clear data output from Step 5, and the output is the user's emotional state and gesture identification results.

[0648] Step 7:

[0649] Audio analysis

[0650] The server uses a speech recognition engine to analyze the speech content and emotional tone from the audio data. Google Speech-to-Text is used for this process. The input is the audio data processed in step 5. For example, "Hello" and a friendly tone are identified. The output is text data of the speech content and emotional tone.

[0651] Step 8:

[0652] Line of sight analysis

[0653] The server uses eye-tracking technology to analyze the user's gaze data and recognize their intention. Specifically, it analyzes data showing the user's gaze on a specific icon on the screen. The input is the gaze data from step 2, and the output is text data about the user's gaze area and intention.

[0654] Step 9:

[0655] Generating a response

[0656] The server generates an appropriate response based on the analysis results. Using a generative AI model (e.g., OpenAI GPT-4), it generates the optimal response based on the user's emotional state and intention. For example, in response to the utterance "Hello," a response such as "Hello! How is your day?" is generated. The input is the analysis results from steps 6 to 8, and the output is the generated voice or text message.

[0657] Step 10:

[0658] Sending a Response

[0659] The server sends the generated response to the terminal. The data is transferred to the terminal in the appropriate format (audio, text, video). The input is the data generated in step 9, and the output is the response data to the terminal.

[0660] Step 11:

[0661] Providing a response

[0662] The terminal provides the generated response received from the server to the user, either playing it through the speaker if it is a voice message or displaying it on the screen if it is a text message. The input is the response data from step 10, and the output is the response the user receives.

[0663] Step 12:

[0664] Providing feedback

[0665] The user provides feedback on the system's response. Using a smartphone or PC, the user evaluates whether the response was appropriate. The input for this step is the user's impression, and the output is feedback data.

[0666] Step 13:

[0667] Collecting feedback

[0668] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, which improves the accuracy of the response. The input of this step is the feedback data from step 12, and the output is the training data.

[0669] Step 14:

[0670] Training the model

[0671] The server retrains the generative AI model with the newly collected feedback, which allows the system to self-improve. The input for this step is the training data from step 13, and the output is an updated generative AI model.

[0672] (Application example 1)

[0673] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0674] Conventional customer service systems in brick-and-mortar stores have difficulty accurately grasping customers' emotions and intentions, making smooth communication difficult. Furthermore, they rely heavily on the skills of employees, making it difficult to maintain a consistent level of service quality. When employees are dealing with a large number of customers, they need to respond efficiently and appropriately, but there has been no effective way to achieve this.

[0675] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0676] In this invention, the server includes: means for analyzing a user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to identify the content of speech and emotional tone; a generative AI model for generating appropriate responses based on the analysis results; means for providing the generated responses to the user; means for collecting feedback from the user and using it as training data for the model; means for capturing customer facial expressions and gestures using a camera; and means for collecting customer voices using a microphone and displaying the generated responses on the smart glasses based on the analysis results. This makes it possible to grasp the customer's emotions and intentions in real time and generate and provide appropriate responses. As a result, high-quality service can be provided and smooth communication can be achieved without relying on the skills of employees.

[0677] An "emotion recognition algorithm" is a technology for analyzing a user's facial expression data and identifying their emotional state.

[0678] "Means for analyzing user gesture data" refers to technology that detects the movements of the user's hands and body and recognizes the meaning of those movements.

[0679] "Means for analyzing voice data" refers to technology that identifies the content and emotional tone of a user's speech and extracts appropriate information.

[0680] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[0681] The "means for providing a response to a user" refers to a technique for transmitting the generated response to a user in the form of voice, text, or the like.

[0682] "Means for collecting feedback and using it as model training data" refers to techniques for collecting user evaluations and using that data to improve the system's performance.

[0683] "Camera-based means" refers to the device or system used to acquire video data.

[0684] "Smart glasses" are devices equipped with a camera, display, microphone, etc., that display various information in real time when worn by a user.

[0685] "Means using a microphone" refers to a device or system for capturing audio data.

[0686] This invention is a customer service support system for brick-and-mortar stores that facilitates smooth communication between store employees wearing smart glasses and customers. The system combines emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, and gaze tracking technology.

[0687] Hardware and Software Configuration

[0688] Smart glasses: Smart glasses are equipped with a camera, microphone, display, and eye-tracking sensor to collect customers' facial expressions, gestures, and voice in real time.

[0689] Server: The server receives the data sent by the smart glasses, analyzes and processes it. Specifically, the following processes are performed:

[0690] Emotion recognition: Based on the video data captured by the camera, an emotion recognition algorithm is run to identify emotions from the user's facial expressions.

[0691] Gesture recognition: Uses technology to recognize user gestures (e.g., hand movements) from video data captured by a camera.

[0692] Voice analysis: Analyzes the content and emotional tone of the user's speech based on the voice data captured by the microphone.

[0693] Eye tracking: Analyzes data acquired from an eye tracking sensor to identify the object and direction the user is looking at.

[0694] Generative AI Model: Uses an artificial intelligence model that generates the optimal response based on the analysis results. This response can be in the form of text or voice.

[0695] Operation procedure and example

[0696] As an example of use in a physical store, the system would be implemented in the following steps:

[0697] 1. User data collection:

[0698] A customer approaches an employee wearing smart glasses and begins a conversation. The camera in the glasses captures the customer's facial expressions in real time, and the microphone collects their voice.

[0699] 2. Data Analysis:

[0700] The collected data is sent to a server, where an emotion recognition algorithm analyzes the facial expression data to identify the customer's emotional state (e.g., joy, surprise), while gesture recognition technology detects hand movements and other gestures, and voice analysis technology analyzes the content and emotional tone of spoken words.

[0701] 3. Generate a response:

[0702] Based on the analysis, the generative AI model generates an appropriate response. For example, if it determines that a customer is tired, it generates a voice message saying, "Would you like to take a break?" This response can be displayed on the smart glasses' display or delivered through the audio speaker.

[0703] 4. Eye-tracking information provision:

[0704] Using eye-tracking technology, if a customer is looking at a particular product, information about that product will be displayed on the smart glasses' display and provided to the customer.

[0705] 5. Gathering feedback and self-learning:

[0706] Feedback from customers and employees is collected and sent to a server, where it is used as training data for a generative AI model, helping to improve the accuracy of the system.

[0707] Examples and prompts

[0708] Examples:

[0709] A customer begins a conversation with an employee wearing smart glasses. The camera captures the customer's face and determines their emotion as "joy." The microphone collects the voice, saying "Hello, how are you today?", which is then analyzed by the smart glasses. Based on the analysis results, the generative AI model suggests, "Hello! Would you like to check out our new promotions?"

[0710] Example prompt sentence:

[0711] A user said: Hello, how are you today?

[0712] Generate an appropriate response to it.

[0713] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0714] Step 1:

[0715] Device data collection

[0716] When a customer approaches an employee wearing smart glasses, the device (smart glasses) uses a camera to capture the user's facial expression and gesture data, and also uses a microphone to collect the user's voice data.

[0717] Input: User facial expression video, gesture video, voice

[0718] Output: Facial expression data, gesture data, voice data

[0719] Step 2:

[0720] Sending data

[0721] The terminal transmits the collected facial expression data, gesture data, and voice data to a server.

[0722] Input: facial expression data, gesture data, voice data

[0723] Output: Facial expression data, gesture data, and voice data sent to the server

[0724] Step 3:

[0725] Server data preprocessing

[0726] The server receives the transmitted data and performs noise removal and necessary data format conversion. Specifically, it removes noise from audio data and extracts facial features from video data.

[0727] Input: Facial expression data, gesture data, and voice data sent to the server

[0728] Output: Noise-removed audio data, facial expression data with extracted features

[0729] Step 4:

[0730] Performing emotion and gesture recognition

[0731] The server uses an emotion recognition algorithm to analyze the facial expression data to identify the user's emotional state, and a gesture recognition algorithm to analyze the gesture data to recognize the user's actions.

[0732] Input: Facial expression data and gesture data with extracted feature points

[0733] Output: Emotional state, recognized gestures

[0734] Step 5:

[0735] Performing audio analysis

[0736] The server uses a voice analysis engine to identify the user's vocal content and emotional tone from the transmitted voice data, filtering out background noise to ensure clear voice data.

[0737] Input: Denoised audio data

[0738] Output: Vocal content, emotional tone

[0739] Step 6:

[0740] Performing eye tracking

[0741] Data obtained from the device's eye-tracking sensor is sent to a server, which then analyzes the user's eye-gaze data to identify the object and direction the user is looking at.

[0742] Input: Gaze data

[0743] Output: Object and direction being gazed at

[0744] Step 7:

[0745] Response generation using generative AI models

[0746] The server uses a generative AI model to generate an appropriate response based on the user's emotional state, recognized gestures, vocal content, emotional tone, and gaze data, which can be in the form of text or voice.

[0747] Input: Emotional state, recognized gestures, vocal content, emotional tone, gaze data

[0748] Output: The generated response (text or audio)

[0749] Step 8:

[0750] Providing a generated response

[0751] The server sends the generated response to the terminal, which then provides the generated response to the user: if it is a voice message, it is played through the speaker, and if it is a text message, it is displayed on the display of the smart glasses.

[0752] Input: The generated response

[0753] Output: The audio or text message provided to the user.

[0754] Step 9:

[0755] Collecting feedback

[0756] Users provide feedback on the responses provided, which is collected and sent to a server that uses it as training data for a generative AI model.

[0757] Input: User feedback

[0758] Output: Feedback data, improving the generative AI model

[0759] Through the above processing steps, a customer service support system is realized that can accurately grasp the emotions and intentions of customers and provide appropriate responses in real time.

[0760] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0761] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's emotional state and intentions with high accuracy and generates appropriate responses based on that. The system functions through interactions between the terminal, the server, and the user, and its details are as follows:

[0762] Collecting input

[0763] User

[0764] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[0765] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time, and it also recognizes whether their gaze is directed in a specific direction.

[0766] Terminal

[0767] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[0768] server

[0769] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[0770] Data Preprocessing

[0771] server

[0772] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the user's face and its features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is similarly preprocessed to identify the user's gaze point.

[0773] Examples include: filtering out background noise from audio data to clarify what the user is saying and the tone of their speech; locating the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.); and determining which part of the screen the user is looking at from eye-tracking data.

[0774] Analysis and Recognition

[0775] Emotion Engine

[0776] The emotion engine in the server analyzes the collected facial expression, voice, and gesture data to comprehensively identify the user's emotional state. It also uses gaze data to complement the emotional state.

[0777] For example: If the user is smiling, the emotion engine will determine "happiness." If the user waves, the gesture recognition algorithm will recognize this as a "greeting." From the audio, it will be analyzed that the user said "hello" and that the emotional tone is friendly. From the gaze data, it will be recognized that the user is looking at a specific icon on the screen.

[0778] Analysis results and response generation

[0779] server

[0780] The server uses a generative AI model to generate an appropriate response based on the analysis results of the emotion engine. This response is encoded in the optimal format (audio, text, video, etc.) according to the user's emotional state and intention.

[0781] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[0782] Terminal

[0783] The terminal provides the generated response sent from the server to the user: if it is a voice message, it is played through the speaker, if it is a text message, it is displayed on the screen.

[0784] Gathering feedback and self-learning

[0785] User

[0786] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[0787] server

[0788] The server accumulates the feedback collected from users and stores it in a database. This feedback data is used as training data for the generative AI model and is used to improve the model's accuracy.

[0789] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[0790] The above is an embodiment of the present invention. By combining this system with an emotion engine, it is possible to solve the problems that existed in conventional communication methods, and to realize smooth communication, particularly with people with disabilities and animals.

[0791] The processing flow will be explained below.

[0792] Step 1:

[0793] User

[0794] The user speaks to the terminal, for example, saying "hello" and smiling.

[0795] Step 2:

[0796] Terminal

[0797] The device uses a camera and microphone to capture the user's facial expression and voice data in real time, and an eye-tracking sensor to capture the user's gaze data, which is temporarily stored in a buffer.

[0798] Step 3:

[0799] Terminal

[0800] The device sends the collected data, including facial expression data, voice data, and gaze data, to a server.

[0801] Step 4:

[0802] server

[0803] The server pre-processes the received data, removing background noise from the video and applying a face detection algorithm to identify facial features. The audio data is also cleared through a noise reduction filter.

[0804] Step 5:

[0805] server

[0806] The server sends the preprocessed data to the emotion engine, which analyzes the facial expression data to identify the user's emotional state. It also analyzes the gesture data to recognize the user's actions. It also analyzes the voice data using a natural language processing (NLP) engine to analyze the speech content and emotional tone. It also uses gaze data to recognize the user's gaze point and intention.

[0807] Step 6:

[0808] server

[0809] The server uses a generative AI model based on the analysis results of the emotion engine to generate an appropriate response, which can be in the form of voice, text, or video depending on the user's emotional state and intentions.

[0810] Step 7:

[0811] Terminal

[0812] The device receives the generated response sent by the server, plays it over the speaker if it is a voice message, or displays it on the screen if it is a text message.

[0813] Step 8:

[0814] User

[0815] The user provides feedback on the system's response, such as rating it "good" or "needs improvement."

[0816] Step 9:

[0817] server

[0818] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve the accuracy of the system.

[0819] Example 2

[0820] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0821] Conventional communication systems have struggled to accurately recognize a user's emotions and intentions and generate appropriate responses based on them. In particular, the technology for comprehensively analyzing various inputs, such as voice, facial expressions, and gaze, to comprehensively assess the user's state was immature. Furthermore, the functionality for collecting feedback on the generated responses and using it to improve the system was insufficient. This made it difficult to communicate smoothly, particularly with people with disabilities and animals.

[0822] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0823] In this invention, the server includes means for capturing and temporarily storing a user's facial expression data with a camera, means for collecting and temporarily storing the user's voice data with a microphone, means for acquiring and temporarily storing the user's gaze data with an eye tracking sensor, means for preprocessing the received data and identifying the position and feature points of the face using a face detection algorithm, means for applying a noise reduction filter to extract clear voice information, means for an emotion engine to analyze the facial expression data, voice data, and gesture data to identify the user's emotional state, means for generating an appropriate response based on the analysis results using a generative AI model, means for providing the generated response to the user in the form of a voice message, a text message, or a video, and means for collecting user feedback, storing it in a database, and using it as training data for the generative AI model. This makes it possible to recognize user emotions and intentions with high accuracy, generate appropriate responses, and improve the system based on the feedback.

[0824] "User" refers to an individual who operates the system and provides input data.

[0825] "Terminal" refers to a device that collects and temporarily stores a user's facial expression data, voice data, and gaze data.

[0826] "Server" refers to a central processing unit that receives data sent from terminals and performs pre-processing, analysis, and response generation.

[0827] "Facial Expression Data" refers to information about a user's facial expressions captured through a camera.

[0828] "Voice data" refers to acoustic information about a user's speech collected through a microphone.

[0829] "Gaze data" refers to information about the direction of a user's gaze and point of gaze obtained through an eye tracking sensor.

[0830] "Preprocessing" refers to the initial processing of received data, such as noise reduction and feature point extraction.

[0831] A "face detection algorithm" refers to a computational method for identifying the position and features of a face from video data.

[0832] "Noise reduction filter" refers to a processing technology for removing unwanted background noise from audio data and extracting clear audio information.

[0833] "Emotion engine" refers to a program that analyzes facial expression data, voice data, and gesture data to identify the user's emotional state.

[0834] "Gesture data" refers to information about hand and body movements made by a user.

[0835] "Generative AI model" refers to an artificial intelligence model used to generate appropriate responses based on analytical results.

[0836] "Response" refers to the output information, such as a voice message, text message, or video, generated by a generative AI model.

[0837] "Feedback" refers to information regarding evaluations and opinions provided by users to the system.

[0838] "Database" means a data storage system for storing and managing collected feedback and other data.

[0839] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's facial expression data, voice data, and gaze data with high accuracy, and generates appropriate responses based on that data. The system functions through the interaction of three parties: the terminal, the server, and the user.

[0840] Collecting input

[0841] User

[0842] The user speaks or gestures into the device, for example, by waving while saying "hello," and the camera and microphone capture this input.

[0843] Examples:

[0844] When a user turns to the device and says "hello" and waves, the camera captures the user's facial expressions and gestures, and the microphone collects audio.

[0845] Terminal

[0846] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data using an eye-tracking sensor. These input data are temporarily stored in a buffer.

[0847] Examples:

[0848] The device uses a camera and an eye-tracking sensor to capture when the user is looking at a specific part of the screen, and stores this as gaze data.

[0849] server

[0850] The server receives the data sent from the terminal and temporarily stores it in storage for preprocessing.

[0851] Data Preprocessing

[0852] server

[0853] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[0854] Examples:

[0855] The server detects the user's face from the video data and extracts the positions of the eyes and mouth as feature points, and removes background noise from the audio data to extract clear audio information.

[0856] Analysis and Recognition

[0857] Emotion engine (on the server)

[0858] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also uses gaze data to complement the emotional state.

[0859] Examples:

[0860] The emotion engine recognizes when a user is smiling as "happy," when a user is waving as "greeting," and when a user says "hello" in a friendly tone.

[0861] Generating a response

[0862] server

[0863] Based on the analysis results of the emotion engine, the generative AI model generates an appropriate response, which can be encoded in the form of audio, text, video, etc.

[0864] Examples:

[0865] If the server senses that the user is tired, it generates a voice message saying "Do you want to take a break?" If the user says "Hello", it generates a text message saying "Hello! How's your day?"

[0866] Providing a response to the user

[0867] Terminal

[0868] The device receives a response from the server and provides it to the user: a voice message is played over the speaker, a text message is displayed on the screen.

[0869] Examples:

[0870] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays text on the screen saying, "Hello! How are you today?"

[0871] Gathering feedback and self-learning

[0872] User

[0873] The user provides feedback to the system, including the appropriateness of the response and areas for improvement.

[0874] Examples:

[0875] The user gives feedback saying, "This response was appropriate."

[0876] server

[0877] The server stores the feedback in a database and uses it as training data for the generative AI model, improving the accuracy of the system.

[0878] Examples:

[0879] Based on user feedback, the server retrains the generative AI model to improve the accuracy of the next response.

[0880] The above is an embodiment of the present invention. This system enables communication that combines an emotion engine, and is expected to facilitate smooth communication, especially with people with disabilities and animals.

[0881] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0882] Step 1:

[0883] Collecting User Input

[0884] User

[0885] The user speaks or gestures into the device, and the camera and microphone capture the user's input. For example, if the user waves while saying "hello," their facial expression and voice are collected.

[0886] Input: User voice and gestures

[0887] Output: Video and audio data captured by the camera and microphone

[0888] Specific behavior:

[0889] When a user speaks or waves at the device, the camera and microphone capture the input, which is then temporarily stored in a buffer.

[0890] Step 2:

[0891] Data buffering

[0892] Terminal

[0893] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor, and temporarily stores this input data in a buffer.

[0894] Input: Captured video data, audio data, and gaze data

[0895] Output: Buffered input data

[0896] Specific behavior:

[0897] The device uses a camera, microphone, and eye-tracking sensor to collect data and temporarily store it in a buffer.

[0898] Step 3:

[0899] Data Transfer and Storage

[0900] server

[0901] The server receives the input data sent from the terminal and temporarily stores it in storage for preprocessing.

[0902] Input: Input data stored in a buffer

[0903] Output: Data temporarily saved in storage

[0904] Specific behavior:

[0905] The server receives input data from the terminal and temporarily stores it in storage, which is used for subsequent preprocessing.

[0906] Step 4:

[0907] Data Preprocessing

[0908] server

[0909] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[0910] Input: Data stored in storage

[0911] Output: Denoised and feature-pointed data

[0912] Specific behavior:

[0913] The server detects the user's face from the video data and extracts the positions of the eyes and mouth. It also removes background noise from the audio data to extract clear audio. It also identifies the user's gaze point from the gaze data.

[0914] Step 5:

[0915] Emotion analysis and recognition

[0916] Server (emotion engine)

[0917] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also analyzes gaze data to complement the emotional state.

[0918] Input: Preprocessed facial expression data, speech data, and gaze data

[0919] Output: User's emotional state

[0920] Specific behavior:

[0921] The emotion engine identifies a user's emotional state (e.g., joy, surprise, sadness, etc.) from subtle facial movements, vocal tone, and gesture patterns, and complements emotion analysis by using gaze data to understand where the user is looking.

[0922] Step 6:

[0923] Generating a response

[0924] server

[0925] Based on the emotion engine's analysis, a generative AI model generates an appropriate response, which can be encoded in the form of audio, text, or video.

[0926] Input: User's emotional state

[0927] Output: The generated response message

[0928] Specific behavior:

[0929] The generative AI model uses prompts based on the analysis results to generate responses such as "Hello! How are you today?"

[0930] Step 7:

[0931] Providing a response

[0932] Terminal

[0933] The terminal provides the user with the response sent by the server, either a voice message played over the speaker or a text message displayed on the screen.

[0934] Input: The generated response message

[0935] Output: Providing a response to the user (voice, text, etc.)

[0936] Specific behavior:

[0937] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays a text message on the screen saying, "Hello! How's your day going?"

[0938] Step 8:

[0939] Collecting feedback

[0940] User

[0941] The user provides feedback on the system's response, including whether the response was appropriate and what could be improved.

[0942] Input: User feedback (ratings and opinions)

[0943] Output: Feedback information sent to the server

[0944] Specific behavior:

[0945] The user provides feedback such as "this response was appropriate," which is sent to the server via the terminal.

[0946] Step 9:

[0947] Self-study

[0948] server

[0949] The server stores the collected feedback in a database and uses this data as training data for the generative AI model, thereby improving the accuracy of the system.

[0950] Input: User feedback data

[0951] Output: Feedback reflected as training data

[0952] Specific behavior:

[0953] The server analyzes the collected feedback and uses it to retrain the generative AI model so that it generates a better response next time.

[0954] (Application example 2)

[0955] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0956] There is a need to improve work efficiency in factories while ensuring the safety of employees. However, it is not easy for robots to quickly and accurately recognize situations in which employees are in trouble or need assistance, and generate appropriate responses. Furthermore, understanding emotional states and intentions requires integrating and analyzing multiple data sets, including facial expressions, gestures, voice, and gaze, making the process complex. There is a need for a way to solve these problems and enable smooth communication between employees and robots.

[0957] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for analyzing the user's (employee's) facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses as voice messages through a speaker; means for collecting user feedback and using it as training data for the model; and means installed in a robot that accurately identifies the emotional state and intention of the employee through interaction with the employee and generates appropriate responses. This makes it possible to quickly and accurately recognize situations in which the employee is in difficulty and provide appropriate support. This improves work efficiency in the factory and ensures employee safety.

[0958] An "emotion recognition algorithm" is a computational method for analyzing facial expression data to identify a user's emotional state.

[0959] "Gesture data" is data that includes the user's movements and hand movements, and is analyzed to recognize meaningful actions.

[0960] "Voice data" refers to data containing sounds and words uttered by a user, and is used to identify the content and emotional tone of the speech.

[0961] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analyzed data.

[0962] "Eye tracking technology" is a technology that acquires a user's gaze data and analyzes the direction of their gaze and the point of focus.

[0963] A "robot" is an automated mechanical device used in a factory to interact with workers.

[0964] "Feedback" refers to the ratings and opinions collected from users that are used as training data for generative AI models.

[0965] "Interaction" is the exchange of information between a user and a system (or robot).

[0966] This invention is a system that uses emotion recognition algorithms and generative AI models to achieve smooth communication between robots and workers in factories. This system is implemented by combining the following hardware and software.

[0967] Hardware

[0968] Camera: A device used to capture facial expressions and gestures of employees. Used to obtain video data.

[0969] Microphone: A device for collecting employee voice data. Used to analyze speech content and emotional tone.

[0970] Eye tracking sensor: Used to capture employee gaze data and analyze user intent.

[0971] Speaker: A device for providing generated audio messages to employees.

[0972] Robot: An automated mechanical device that interacts with workers in a factory.

[0973] software

[0974] Python: A programming language for implementing the entire program.

[0975] TensorFlow: A machine learning library for building emotion recognition algorithms and generative AI models.

[0976] OpenCV: A library for preprocessing video data and identifying facial positions and feature points.

[0977] PyAudio: A library for preprocessing and denoising audio data.

[0978] Eye tracking library: A library for acquiring gaze data and analyzing user intent.

[0979] System Operation

[0980] 1. Collecting input

[0981] The camera captures the employee's facial expression data and uses OpenCV to detect the face position and feature points.

[0982] The microphone collects audio data and PyAudio is used to filter out background noise.

[0983] An eye-tracking sensor captures the employee's gaze data.

[0984] 2. Data Preprocessing

[0985] Noise is removed from the video data using OpenCV, and facial feature points are identified.

[0986] The audio data is denoised using PyAudio to extract clear audio information.

[0987] The gaze data is preprocessed by an eye-tracking library to identify the gaze point.

[0988] 3. Analysis and Recognition

[0989] Emotion recognition algorithms analyze facial, voice, and gesture data to identify the employee's emotional state, and also analyze gaze data to complement the emotional state.

[0990] 4. Analysis results and response generation

[0991] The server uses a generative AI model to generate an appropriate response based on the analysis of the emotion recognition algorithm, which can be in the form of a voice message, text, or other format.

[0992] 5. Gather feedback and self-study

[0993] User feedback is sent to the server, and the generative AI model uses this feedback to self-train.

[0994] Specific examples

[0995] If an employee says with a troubled expression, "I need another person to do this task," the robot will respond by generating a voice message saying, "Would you like additional support?" This action allows the employee to quickly receive the necessary support, improving work efficiency within the factory.

[0996] Prompt Sentence Examples

[0997] "The user (employee) looks troubled and says, 'I need another person to do this task.' What would a generative AI model generate as an appropriate response to this situation?"

[0998] This system combines emotion recognition algorithms and generative AI models to enable smooth communication between workers and robots, which is expected to improve work efficiency and worker safety in factories.

[0999] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1000] Step 1:

[1001] The user provides input by speaking into the camera and microphone or waving their hand, while simultaneously capturing gaze data using an eye-tracking sensor. This allows the user to generate video, audio, and gaze data.

[1002] Step 2:

[1003] The device captures the user's facial expression data with a camera and detects the position and feature points of the face using OpenCV. The input is video data, and the output is the position and feature points of the face.

[1004] Step 3:

[1005] The device collects the user's voice data with a microphone and uses PyAudio to filter out background noise. The input is the voice data, and the output is clear, noise-removed voice information.

[1006] Step 4:

[1007] The device acquires gaze data using an eye-tracking sensor and uses an eye-tracking library to identify the gaze direction and gaze point. The input is the gaze data, and the output is the gaze direction and gaze point.

[1008] Step 5:

[1009] The data collected by the device (facial expression data, voice data, gaze data) is temporarily stored in a buffer and then sent to the server. The input is a set of preprocessed data, and the output is the data sent to the server.

[1010] Step 6:

[1011] The server preprocesses the received data and uses an emotion recognition algorithm to identify the user's emotional state. The input is the received data, and the output is the identified emotional state (e.g., "confused").

[1012] Step 7:

[1013] The server uses a generative AI model to generate an appropriate response based on the analysis results. The input is the emotional state identification result and the prompt sentence, and the output is the generated response (e.g., "Would you like to arrange additional support?").

[1014] Step 8:

[1015] The server sends the generated response to the terminal, which then provides it to the user: a voice message played through the speaker, a text message displayed on the screen. The input is the generated response, and the output is the response provided to the user.

[1016] Step 9:

[1017] The user provides feedback such as evaluation and opinion on the system's response. The input is the user's feedback, which is sent by the terminal to the server.

[1018] Step 10:

[1019] The server collects feedback from users and uses it as training data for the model. This allows the generative AI model to self-learn and improve its accuracy. The input is the feedback data, and the output is an updated generative AI model.

[1020] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1021] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1022] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[1023] [Third embodiment]

[1024] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1025] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[1026] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1027] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[1028] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1029] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1030] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1031] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1032] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1033] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1034] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1035] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[1036] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user, and its details are as follows.

[1037] Collecting input

[1038] User

[1039] The user provides input to the device, for example, by using a camera and microphone to record their facial expressions and voice.

[1040] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time.

[1041] Terminal

[1042] The device includes hardware for collecting facial expression data, gesture data, and voice data from the user, including a camera, a microphone, and an eye-tracking sensor.

[1043] server

[1044] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1045] Data Preprocessing

[1046] server

[1047] The server pre-processes the received data, which includes noise removal, necessary data format conversion, and extraction of the user's facial features.

[1048] Examples: Removing background noise from audio data to clarify what the user is saying and the tone. Identifying the position of the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.).

[1049] Analysis and Recognition

[1050] server

[1051] The server uses emotion recognition algorithms to analyze the user's facial expression data and identify their emotional state, gesture recognition algorithms to recognize meaningful actions from the user's hand and body movements, a voice analysis engine to identify the content and emotional tone of the user's speech from the voice data, and eye-tracking technology to analyze the user's gaze data and recognize their intentions.

[1052] For example: If a user smiles, an emotion recognition algorithm will determine this as "happiness." If a user waves, a gesture recognition algorithm will recognize this as "greeting." From the user's voice, it will be analyzed that the user said "hello" and that the emotional tone is friendly. Eye-tracking technology will recognize that the user is looking at a specific icon on the screen.

[1053] Communication generation

[1054] server

[1055] The server generates an appropriate response based on the analysis results, leveraging generative AI models to create the optimal response based on the user's emotional state and intent. This response can be in the form of audio, text, or video.

[1056] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[1057] Terminal

[1058] The terminal provides the generated response sent from the server to the user, either played through the speaker in the case of a voice message or displayed on the screen in the case of a text message.

[1059] Gathering feedback and self-learning

[1060] User

[1061] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[1062] server

[1063] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[1064] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1065] The above is an embodiment of the present invention. This system solves the problems that existed in conventional communication methods, and enables smooth communication, particularly with people with disabilities and animals.

[1066] The processing flow will be explained below.

[1067] Step 1:

[1068] User

[1069] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[1070] Step 2:

[1071] Terminal

[1072] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[1073] Step 3:

[1074] Terminal

[1075] The device sends the collected data to a server, where video data, audio data, and gaze data are sent together.

[1076] Step 4:

[1077] server

[1078] The server preprocesses the received data: for video data, noise reduction is performed and a face detection algorithm is used to identify the location and features of the user's face; for audio data, a noise reduction filter is applied to extract clear audio information.

[1079] Step 5:

[1080] server

[1081] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. It then uses a gesture analysis algorithm to analyze the user's movements and recognize specific actions. It then uses a voice analysis engine to analyze the voice data and identify the content and emotional tone of the speech. Based on the gaze data, it uses eye-tracking technology to analyze the user's intentions.

[1082] Step 6:

[1083] server

[1084] The server uses a generative AI model to generate an appropriate response based on the analysis results, which is encoded in the optimal format (audio, text, video, etc.) depending on the user's emotional state and intent.

[1085] Step 7:

[1086] Terminal

[1087] The terminal receives the generated response sent by the server, which is played through the speaker if it is a voice message, or displayed on the screen if it is a text message.

[1088] Step 8:

[1089] User

[1090] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[1091] Step 9:

[1092] server

[1093] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve its accuracy.

[1094] Example 1

[1095] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1096] Conventional communication systems lack the ability to comprehensively analyze multiple input data, including not only a user's facial expressions and voice, but also their gaze and gestures. This makes it difficult to accurately understand the user's emotional state and intentions and generate appropriate responses based on them. Furthermore, they lack a mechanism for effectively utilizing user feedback and continuously improving the system's response accuracy. In particular, there have been technical challenges in achieving smooth communication with people with disabilities and animals.

[1097] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1098] In this invention, the server includes: means for collecting a user's facial expression data using a camera and microphone; means for transmitting the collected data to the server; means for performing noise reduction and format conversion on the data received by the server to extract the user's facial features; means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing audio data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means for including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses to the user; and means for collecting user feedback and using it as training data for the model. This enables the system to comprehensively analyze a variety of user input information and generate more accurate responses. Furthermore, the system's accuracy can be improved using user feedback, enabling smoother communication, particularly with people with disabilities and animals.

[1099] "Facial expression data" is video information captured using a camera to show the facial expressions of a user.

[1100] "Voice data" refers to acoustic information captured using a microphone, such as the content of a user's speech and the tone of their voice.

[1101] "Noise reduction" is the process of removing background noise and unwanted sounds from audio data.

[1102] "Format conversion" is the operation of converting collected data into a format that is easy to analyze.

[1103] "Feature points" are information that indicates the main points of the user's face (eyes, mouth, eyebrows, etc.).

[1104] An "emotion recognition algorithm" is a method or program for analyzing facial expression data to identify a user's emotional state.

[1105] "Gesture data" is video information that captures the movements of the user's hands and body.

[1106] A "gesture recognition algorithm" is a method or program for analyzing gesture data and recognizing meaningful actions from user movements.

[1107] "Gaze data" is information captured using eye-tracking technology to determine where a user is looking.

[1108] "Eye tracking technology" is a technology for capturing the direction of a user's gaze and point of gaze.

[1109] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[1110] "Feedback" is the process of collecting user feedback and opinions on responses.

[1111] "Training data" is collected feedback information used to improve the performance of a model.

[1112] A "response" is a message or instruction that the generative AI model provides to the user based on the analysis results.

[1113] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user.

[1114] Collecting input

[1115] The user uses a camera and microphone to provide their facial expressions and voice to the device. For example, when the user speaks in front of the camera, their facial expressions and voice are captured in real time.

[1116] The device includes hardware for collecting facial, gesture, and voice data from the user, including a camera, microphone, and eye-tracking sensors, such as a Logitech C920 camera, a Blue Yeti microphone, and a Tobii Eye Tracker 5.

[1117] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1118] Data Preprocessing

[1119] The server then performs noise reduction and format conversion on the received data, which includes removing background noise from the audio data and making the content and tone of the user's speech clearer. It also locates the user's face from the video data and extracts feature points (eyes, mouth, eyebrows, etc.).

[1120] Analysis and Recognition

[1121] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. For example, the Microsoft Azure Face API is used. It also uses a gesture recognition algorithm to recognize meaningful actions (e.g., greeting, pointing) from the user's hand and body movements. It uses a voice analysis engine (e.g., Google Speech-to-Text) to identify the content and emotional tone of the user's speech from the voice data. Furthermore, it uses eye-tracking technology to analyze the user's gaze data and recognize the user's intentions.

[1122] Communication generation

[1123] The server generates an appropriate response based on the analysis results. It uses a generative AI model (e.g., OpenAI GPT-4) to create the optimal response (voice, text, or video) based on the user's emotional state and intent. For example, in response to the utterance "Hello," it generates the text message "Hello! How is your day?"

[1124] The terminal provides the generated response sent from the server to the user, either playing it through the speaker in the case of a voice message or displaying it on the screen in the case of a text message.

[1125] Gathering feedback and self-learning

[1126] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[1127] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[1128] For example, if a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1129] An example of a prompt sentence is a user's utterance, "What is the weather today?" The server analyzes this voice data and generates a response, "The weather today is sunny."

[1130] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1131] Step 1:

[1132] Providing user input

[1133] The user uses a camera and microphone to provide their facial expressions and voice to the device. The user speaks in front of the camera, and the video and audio are captured in real time. Specifically, a camera (Logitech C920) and a microphone (Blue Yeti) are used. The input of this step is the user's facial expressions and voice, and the output is the captured raw video and audio data.

[1134] Step 2:

[1135] Data collection

[1136] The device uses a camera, microphone, and eye-tracking sensor to collect facial expression data, voice data, and gaze data. Tobii Eye Tracker 5 is used for eye-tracking. For example, when the user looks to the right, the gaze data is captured. The input of this step is the raw data output from step 1, and the output is organized facial expression data, voice data, and gaze data.

[1137] Step 3:

[1138] Sending data

[1139] The device sends the collected data to the server in real time. A secure protocol (e.g., HTTPS) is used for communication. The input of this step is the data collected in step 2, and the output is the data transfer to the server.

[1140] Step 4:

[1141] Temporary data storage

[1142] The server temporarily stores the data received from the terminal in storage. This process allows for efficient management of large amounts of data. The input to this step is the transferred data, which is the output of step 3, and the output is the data stored in storage.

[1143] Step 5:

[1144] Noise reduction and format conversion

[1145] The server removes background noise from the audio data and unnecessary parts from the video data to extract facial features. A noise-canceling algorithm is used to remove noise from the audio data, and a video analysis algorithm is used to extract facial features. The input to this step is the data saved in step 4, and the output is clear pre-processed audio data and facial expression data with extracted features.

[1146] Step 6:

[1147] Emotion and movement analysis

[1148] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. Specifically, the Microsoft Azure Face API classifies emotions into categories such as "happiness," "sadness," and "surprise." It also uses a gesture recognition algorithm to analyze the user's movements. The input of this step is the clear data output from Step 5, and the output is the user's emotional state and gesture identification results.

[1149] Step 7:

[1150] Audio analysis

[1151] The server uses a speech recognition engine to analyze the speech content and emotional tone from the audio data. Google Speech-to-Text is used for this process. The input is the audio data processed in step 5. For example, "Hello" and a friendly tone are identified. The output is text data of the speech content and emotional tone.

[1152] Step 8:

[1153] Line of sight analysis

[1154] The server uses eye-tracking technology to analyze the user's gaze data and recognize their intention. Specifically, it analyzes data showing the user's gaze on a specific icon on the screen. The input is the gaze data from step 2, and the output is text data about the user's gaze area and intention.

[1155] Step 9:

[1156] Generating a response

[1157] The server generates an appropriate response based on the analysis results. Using a generative AI model (e.g., OpenAI GPT-4), it generates the optimal response based on the user's emotional state and intention. For example, in response to the utterance "Hello," a response such as "Hello! How is your day?" is generated. The input is the analysis results from steps 6 to 8, and the output is the generated voice or text message.

[1158] Step 10:

[1159] Sending a Response

[1160] The server sends the generated response to the terminal. The data is transferred to the terminal in the appropriate format (audio, text, video). The input is the data generated in step 9, and the output is the response data to the terminal.

[1161] Step 11:

[1162] Providing a response

[1163] The terminal provides the generated response received from the server to the user, either playing it through the speaker if it is a voice message or displaying it on the screen if it is a text message. The input is the response data from step 10, and the output is the response the user receives.

[1164] Step 12:

[1165] Providing feedback

[1166] The user provides feedback on the system's response. Using a smartphone or PC, the user evaluates whether the response was appropriate. The input for this step is the user's impression, and the output is feedback data.

[1167] Step 13:

[1168] Collecting feedback

[1169] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, which improves the accuracy of the response. The input of this step is the feedback data from step 12, and the output is the training data.

[1170] Step 14:

[1171] Training the model

[1172] The server retrains the generative AI model with the newly collected feedback, which allows the system to self-improve. The input for this step is the training data from step 13, and the output is an updated generative AI model.

[1173] (Application example 1)

[1174] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1175] Conventional customer service systems in brick-and-mortar stores have difficulty accurately grasping customers' emotions and intentions, making smooth communication difficult. Furthermore, they rely heavily on the skills of employees, making it difficult to maintain a consistent level of service quality. When employees are dealing with a large number of customers, they need to respond efficiently and appropriately, but there has been no effective way to achieve this.

[1176] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1177] In this invention, the server includes: means for analyzing a user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to identify the content of speech and emotional tone; a generative AI model for generating appropriate responses based on the analysis results; means for providing the generated responses to the user; means for collecting feedback from the user and using it as training data for the model; means for capturing customer facial expressions and gestures using a camera; and means for collecting customer voices using a microphone and displaying the generated responses on the smart glasses based on the analysis results. This makes it possible to grasp the customer's emotions and intentions in real time and generate and provide appropriate responses. As a result, high-quality service can be provided and smooth communication can be achieved without relying on the skills of employees.

[1178] An "emotion recognition algorithm" is a technology for analyzing a user's facial expression data and identifying their emotional state.

[1179] "Means for analyzing user gesture data" refers to technology that detects the movements of the user's hands and body and recognizes the meaning of those movements.

[1180] "Means for analyzing voice data" refers to technology that identifies the content and emotional tone of a user's speech and extracts appropriate information.

[1181] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[1182] The "means for providing a response to a user" refers to a technique for transmitting the generated response to a user in the form of voice, text, or the like.

[1183] "Means for collecting feedback and using it as model training data" refers to techniques for collecting user evaluations and using that data to improve the system's performance.

[1184] "Camera-based means" refers to the device or system used to acquire video data.

[1185] "Smart glasses" are devices equipped with a camera, display, microphone, etc., that display various information in real time when worn by a user.

[1186] "Means using a microphone" refers to a device or system for capturing audio data.

[1187] This invention is a customer service support system for brick-and-mortar stores that facilitates smooth communication between store employees wearing smart glasses and customers. The system combines emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, and gaze tracking technology.

[1188] Hardware and Software Configuration

[1189] Smart glasses: Smart glasses are equipped with a camera, microphone, display, and eye-tracking sensor to collect customers' facial expressions, gestures, and voice in real time.

[1190] Server: The server receives the data sent by the smart glasses, analyzes and processes it. Specifically, the following processes are performed:

[1191] Emotion recognition: Based on the video data captured by the camera, an emotion recognition algorithm is run to identify emotions from the user's facial expressions.

[1192] Gesture recognition: Uses technology to recognize user gestures (e.g., hand movements) from video data captured by a camera.

[1193] Voice analysis: Analyzes the content and emotional tone of the user's speech based on the voice data captured by the microphone.

[1194] Eye tracking: Analyzes data acquired from an eye tracking sensor to identify the object and direction the user is looking at.

[1195] Generative AI Model: Uses an artificial intelligence model that generates the optimal response based on the analysis results. This response can be in the form of text or voice.

[1196] Operation procedure and example

[1197] As an example of use in a physical store, the system would be implemented in the following steps:

[1198] 1. User data collection:

[1199] A customer approaches an employee wearing smart glasses and begins a conversation. The camera in the glasses captures the customer's facial expressions in real time, and the microphone collects their voice.

[1200] 2. Data Analysis:

[1201] The collected data is sent to a server, where an emotion recognition algorithm analyzes the facial expression data to identify the customer's emotional state (e.g., joy, surprise), while gesture recognition technology detects hand movements and other gestures, and voice analysis technology analyzes the content and emotional tone of spoken words.

[1202] 3. Generate a response:

[1203] Based on the analysis, the generative AI model generates an appropriate response. For example, if it determines that a customer is tired, it generates a voice message saying, "Would you like to take a break?" This response can be displayed on the smart glasses' display or delivered through the audio speaker.

[1204] 4. Eye-tracking information provision:

[1205] Using eye-tracking technology, if a customer is looking at a particular product, information about that product will be displayed on the smart glasses' display and provided to the customer.

[1206] 5. Gathering feedback and self-learning:

[1207] Feedback from customers and employees is collected and sent to a server, where it is used as training data for a generative AI model, helping to improve the accuracy of the system.

[1208] Examples and prompts

[1209] Examples:

[1210] A customer begins a conversation with an employee wearing smart glasses. The camera captures the customer's face and determines their emotion as "joy." The microphone collects the voice, saying "Hello, how are you today?", which is then analyzed by the smart glasses. Based on the analysis results, the generative AI model suggests, "Hello! Would you like to check out our new promotions?"

[1211] Example prompt sentence:

[1212] A user said: Hello, how are you today?

[1213] Generate an appropriate response to it.

[1214] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1215] Step 1:

[1216] Device data collection

[1217] When a customer approaches an employee wearing smart glasses, the device (smart glasses) uses a camera to capture the user's facial expression and gesture data, and also uses a microphone to collect the user's voice data.

[1218] Input: User facial expression video, gesture video, voice

[1219] Output: Facial expression data, gesture data, voice data

[1220] Step 2:

[1221] Sending data

[1222] The terminal transmits the collected facial expression data, gesture data, and voice data to a server.

[1223] Input: facial expression data, gesture data, voice data

[1224] Output: Facial expression data, gesture data, and voice data sent to the server

[1225] Step 3:

[1226] Server data preprocessing

[1227] The server receives the transmitted data and performs noise removal and necessary data format conversion. Specifically, it removes noise from audio data and extracts facial features from video data.

[1228] Input: Facial expression data, gesture data, and voice data sent to the server

[1229] Output: Noise-removed audio data, facial expression data with extracted features

[1230] Step 4:

[1231] Performing emotion and gesture recognition

[1232] The server uses an emotion recognition algorithm to analyze the facial expression data to identify the user's emotional state, and a gesture recognition algorithm to analyze the gesture data to recognize the user's actions.

[1233] Input: Facial expression data and gesture data with extracted feature points

[1234] Output: Emotional state, recognized gestures

[1235] Step 5:

[1236] Performing audio analysis

[1237] The server uses a voice analysis engine to identify the user's vocal content and emotional tone from the transmitted voice data, filtering out background noise to ensure clear voice data.

[1238] Input: Denoised audio data

[1239] Output: Vocal content, emotional tone

[1240] Step 6:

[1241] Performing eye tracking

[1242] Data obtained from the device's eye-tracking sensor is sent to a server, which then analyzes the user's eye-gaze data to identify the object and direction the user is looking at.

[1243] Input: Gaze data

[1244] Output: Object and direction being gazed at

[1245] Step 7:

[1246] Response generation using generative AI models

[1247] The server uses a generative AI model to generate an appropriate response based on the user's emotional state, recognized gestures, vocal content, emotional tone, and gaze data, which can be in the form of text or voice.

[1248] Input: Emotional state, recognized gestures, vocal content, emotional tone, gaze data

[1249] Output: The generated response (text or audio)

[1250] Step 8:

[1251] Providing a generated response

[1252] The server sends the generated response to the terminal, which then provides the generated response to the user: if it is a voice message, it is played through the speaker, and if it is a text message, it is displayed on the display of the smart glasses.

[1253] Input: The generated response

[1254] Output: The audio or text message provided to the user.

[1255] Step 9:

[1256] Collecting feedback

[1257] Users provide feedback on the responses provided, which is collected and sent to a server that uses it as training data for a generative AI model.

[1258] Input: User feedback

[1259] Output: Feedback data, improving the generative AI model

[1260] Through the above processing steps, a customer service support system is realized that can accurately grasp the emotions and intentions of customers and provide appropriate responses in real time.

[1261] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1262] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's emotional state and intentions with high accuracy and generates appropriate responses based on that. The system functions through interactions between the terminal, the server, and the user, and its details are as follows:

[1263] Collecting input

[1264] User

[1265] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[1266] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time, and it also recognizes whether their gaze is directed in a specific direction.

[1267] Terminal

[1268] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[1269] server

[1270] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1271] Data Preprocessing

[1272] server

[1273] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the user's face and its features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is similarly preprocessed to identify the user's gaze point.

[1274] Examples include: filtering out background noise from audio data to clarify what the user is saying and the tone of their speech; locating the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.); and determining which part of the screen the user is looking at from eye-tracking data.

[1275] Analysis and Recognition

[1276] Emotion Engine

[1277] The emotion engine in the server analyzes the collected facial expression, voice, and gesture data to comprehensively identify the user's emotional state. It also uses gaze data to complement the emotional state.

[1278] For example: If the user is smiling, the emotion engine will determine "happiness." If the user waves, the gesture recognition algorithm will recognize this as a "greeting." From the audio, it will be analyzed that the user said "hello" and that the emotional tone is friendly. From the gaze data, it will be recognized that the user is looking at a specific icon on the screen.

[1279] Analysis results and response generation

[1280] server

[1281] The server uses a generative AI model to generate an appropriate response based on the analysis results of the emotion engine. This response is encoded in the optimal format (audio, text, video, etc.) according to the user's emotional state and intention.

[1282] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[1283] Terminal

[1284] The terminal provides the generated response sent from the server to the user: if it is a voice message, it is played through the speaker, if it is a text message, it is displayed on the screen.

[1285] Gathering feedback and self-learning

[1286] User

[1287] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[1288] server

[1289] The server accumulates the feedback collected from users and stores it in a database. This feedback data is used as training data for the generative AI model and is used to improve the model's accuracy.

[1290] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1291] The above is an embodiment of the present invention. By combining this system with an emotion engine, it is possible to solve the problems that existed in conventional communication methods, and to realize smooth communication, particularly with people with disabilities and animals.

[1292] The processing flow will be explained below.

[1293] Step 1:

[1294] User

[1295] The user speaks to the terminal, for example, saying "hello" and smiling.

[1296] Step 2:

[1297] Terminal

[1298] The device uses a camera and microphone to capture the user's facial expression and voice data in real time, and an eye-tracking sensor to capture the user's gaze data, which is temporarily stored in a buffer.

[1299] Step 3:

[1300] Terminal

[1301] The device sends the collected data, including facial expression data, voice data, and gaze data, to a server.

[1302] Step 4:

[1303] server

[1304] The server pre-processes the received data, removing background noise from the video and applying a face detection algorithm to identify facial features. The audio data is also cleared through a noise reduction filter.

[1305] Step 5:

[1306] server

[1307] The server sends the preprocessed data to the emotion engine, which analyzes the facial expression data to identify the user's emotional state. It also analyzes the gesture data to recognize the user's actions. It also analyzes the voice data using a natural language processing (NLP) engine to analyze the speech content and emotional tone. It also uses gaze data to recognize the user's gaze point and intention.

[1308] Step 6:

[1309] server

[1310] The server uses a generative AI model based on the analysis results of the emotion engine to generate an appropriate response, which can be in the form of voice, text, or video depending on the user's emotional state and intentions.

[1311] Step 7:

[1312] Terminal

[1313] The device receives the generated response sent by the server, plays it over the speaker if it is a voice message, or displays it on the screen if it is a text message.

[1314] Step 8:

[1315] User

[1316] The user provides feedback on the system's response, such as rating it "good" or "needs improvement."

[1317] Step 9:

[1318] server

[1319] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve the accuracy of the system.

[1320] Example 2

[1321] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1322] Conventional communication systems have struggled to accurately recognize a user's emotions and intentions and generate appropriate responses based on them. In particular, the technology for comprehensively analyzing various inputs, such as voice, facial expressions, and gaze, to comprehensively assess the user's state was immature. Furthermore, the functionality for collecting feedback on the generated responses and using it to improve the system was insufficient. This made it difficult to communicate smoothly, particularly with people with disabilities and animals.

[1323] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1324] In this invention, the server includes means for capturing and temporarily storing a user's facial expression data with a camera, means for collecting and temporarily storing the user's voice data with a microphone, means for acquiring and temporarily storing the user's gaze data with an eye tracking sensor, means for preprocessing the received data and identifying the position and feature points of the face using a face detection algorithm, means for applying a noise reduction filter to extract clear voice information, means for an emotion engine to analyze the facial expression data, voice data, and gesture data to identify the user's emotional state, means for generating an appropriate response based on the analysis results using a generative AI model, means for providing the generated response to the user in the form of a voice message, a text message, or a video, and means for collecting user feedback, storing it in a database, and using it as training data for the generative AI model. This makes it possible to recognize user emotions and intentions with high accuracy, generate appropriate responses, and improve the system based on the feedback.

[1325] "User" refers to an individual who operates the system and provides input data.

[1326] "Terminal" refers to a device that collects and temporarily stores a user's facial expression data, voice data, and gaze data.

[1327] "Server" refers to a central processing unit that receives data sent from terminals and performs pre-processing, analysis, and response generation.

[1328] "Facial Expression Data" refers to information about a user's facial expressions captured through a camera.

[1329] "Voice data" refers to acoustic information about a user's speech collected through a microphone.

[1330] "Gaze data" refers to information about the direction of a user's gaze and point of gaze obtained through an eye tracking sensor.

[1331] "Preprocessing" refers to the initial processing of received data, such as noise reduction and feature point extraction.

[1332] A "face detection algorithm" refers to a computational method for identifying the position and features of a face from video data.

[1333] "Noise reduction filter" refers to a processing technology for removing unwanted background noise from audio data and extracting clear audio information.

[1334] "Emotion engine" refers to a program that analyzes facial expression data, voice data, and gesture data to identify the user's emotional state.

[1335] "Gesture data" refers to information about hand and body movements made by a user.

[1336] "Generative AI model" refers to an artificial intelligence model used to generate appropriate responses based on analytical results.

[1337] "Response" refers to the output information, such as a voice message, text message, or video, generated by a generative AI model.

[1338] "Feedback" refers to information regarding evaluations and opinions provided by users to the system.

[1339] "Database" means a data storage system for storing and managing collected feedback and other data.

[1340] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's facial expression data, voice data, and gaze data with high accuracy, and generates appropriate responses based on that data. The system functions through the interaction of three parties: the terminal, the server, and the user.

[1341] Collecting input

[1342] User

[1343] The user speaks or gestures into the device, for example, by waving while saying "hello," and the camera and microphone capture this input.

[1344] Examples:

[1345] When a user turns to the device and says "hello" and waves, the camera captures the user's facial expressions and gestures, and the microphone collects audio.

[1346] Terminal

[1347] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data using an eye-tracking sensor. These input data are temporarily stored in a buffer.

[1348] Examples:

[1349] The device uses a camera and an eye-tracking sensor to capture when the user is looking at a specific part of the screen, and stores this as gaze data.

[1350] server

[1351] The server receives the data sent from the terminal and temporarily stores it in storage for preprocessing.

[1352] Data Preprocessing

[1353] server

[1354] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[1355] Examples:

[1356] The server detects the user's face from the video data and extracts the positions of the eyes and mouth as feature points, and removes background noise from the audio data to extract clear audio information.

[1357] Analysis and Recognition

[1358] Emotion engine (on the server)

[1359] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also uses gaze data to complement the emotional state.

[1360] Examples:

[1361] The emotion engine recognizes when a user is smiling as "happy," when a user is waving as "greeting," and when a user says "hello" in a friendly tone.

[1362] Generating a response

[1363] server

[1364] Based on the analysis results of the emotion engine, the generative AI model generates an appropriate response, which can be encoded in the form of audio, text, video, etc.

[1365] Examples:

[1366] If the server senses that the user is tired, it generates a voice message saying "Do you want to take a break?" If the user says "Hello", it generates a text message saying "Hello! How's your day?"

[1367] Providing a response to the user

[1368] Terminal

[1369] The device receives a response from the server and provides it to the user: a voice message is played over the speaker, a text message is displayed on the screen.

[1370] Examples:

[1371] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays text on the screen saying, "Hello! How are you today?"

[1372] Gathering feedback and self-learning

[1373] User

[1374] The user provides feedback to the system, including the appropriateness of the response and areas for improvement.

[1375] Examples:

[1376] The user gives feedback saying, "This response was appropriate."

[1377] server

[1378] The server stores the feedback in a database and uses it as training data for the generative AI model, improving the accuracy of the system.

[1379] Examples:

[1380] Based on user feedback, the server retrains the generative AI model to improve the accuracy of the next response.

[1381] The above is an embodiment of the present invention. This system enables communication that combines an emotion engine, and is expected to facilitate smooth communication, especially with people with disabilities and animals.

[1382] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1383] Step 1:

[1384] Collecting User Input

[1385] User

[1386] The user speaks or gestures into the device, and the camera and microphone capture the user's input. For example, if the user waves while saying "hello," their facial expression and voice are collected.

[1387] Input: User voice and gestures

[1388] Output: Video and audio data captured by the camera and microphone

[1389] Specific behavior:

[1390] When a user speaks or waves at the device, the camera and microphone capture the input, which is then temporarily stored in a buffer.

[1391] Step 2:

[1392] Data buffering

[1393] Terminal

[1394] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor, and temporarily stores this input data in a buffer.

[1395] Input: Captured video data, audio data, and gaze data

[1396] Output: Buffered input data

[1397] Specific behavior:

[1398] The device uses a camera, microphone, and eye-tracking sensor to collect data and temporarily store it in a buffer.

[1399] Step 3:

[1400] Data Transfer and Storage

[1401] server

[1402] The server receives the input data sent from the terminal and temporarily stores it in storage for preprocessing.

[1403] Input: Input data stored in a buffer

[1404] Output: Data temporarily saved in storage

[1405] Specific behavior:

[1406] The server receives input data from the terminal and temporarily stores it in storage, which is used for subsequent preprocessing.

[1407] Step 4:

[1408] Data Preprocessing

[1409] server

[1410] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[1411] Input: Data stored in storage

[1412] Output: Denoised and feature-pointed data

[1413] Specific behavior:

[1414] The server detects the user's face from the video data and extracts the positions of the eyes and mouth. It also removes background noise from the audio data to extract clear audio. It also identifies the user's gaze point from the gaze data.

[1415] Step 5:

[1416] Emotion analysis and recognition

[1417] Server (emotion engine)

[1418] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also analyzes gaze data to complement the emotional state.

[1419] Input: Preprocessed facial expression data, speech data, and gaze data

[1420] Output: User's emotional state

[1421] Specific behavior:

[1422] The emotion engine identifies a user's emotional state (e.g., joy, surprise, sadness, etc.) from subtle facial movements, vocal tone, and gesture patterns, and complements emotion analysis by using gaze data to understand where the user is looking.

[1423] Step 6:

[1424] Generating a response

[1425] server

[1426] Based on the emotion engine's analysis, a generative AI model generates an appropriate response, which can be encoded in the form of audio, text, or video.

[1427] Input: User's emotional state

[1428] Output: The generated response message

[1429] Specific behavior:

[1430] The generative AI model uses prompts based on the analysis results to generate responses such as "Hello! How are you today?"

[1431] Step 7:

[1432] Providing a response

[1433] Terminal

[1434] The terminal provides the user with the response sent by the server, either a voice message played over the speaker or a text message displayed on the screen.

[1435] Input: The generated response message

[1436] Output: Providing a response to the user (voice, text, etc.)

[1437] Specific behavior:

[1438] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays a text message on the screen saying, "Hello! How's your day going?"

[1439] Step 8:

[1440] Collecting feedback

[1441] User

[1442] The user provides feedback on the system's response, including whether the response was appropriate and what could be improved.

[1443] Input: User feedback (ratings and opinions)

[1444] Output: Feedback information sent to the server

[1445] Specific behavior:

[1446] The user provides feedback such as "this response was appropriate," which is sent to the server via the terminal.

[1447] Step 9:

[1448] Self-study

[1449] server

[1450] The server stores the collected feedback in a database and uses this data as training data for the generative AI model, thereby improving the accuracy of the system.

[1451] Input: User feedback data

[1452] Output: Feedback reflected as training data

[1453] Specific behavior:

[1454] The server analyzes the collected feedback and uses it to retrain the generative AI model so that it generates a better response next time.

[1455] (Application example 2)

[1456] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1457] There is a need to improve work efficiency in factories while ensuring the safety of employees. However, it is not easy for robots to quickly and accurately recognize situations in which employees are in trouble or need assistance, and generate appropriate responses. Furthermore, understanding emotional states and intentions requires integrating and analyzing multiple data sets, including facial expressions, gestures, voice, and gaze, making the process complex. There is a need for a way to solve these problems and enable smooth communication between employees and robots.

[1458] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for analyzing the user's (employee's) facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses as voice messages through a speaker; means for collecting user feedback and using it as training data for the model; and means installed in a robot that accurately identifies the emotional state and intention of the employee through interaction with the employee and generates appropriate responses. This makes it possible to quickly and accurately recognize situations in which the employee is in difficulty and provide appropriate support. This improves work efficiency in the factory and ensures employee safety.

[1459] An "emotion recognition algorithm" is a computational method for analyzing facial expression data to identify a user's emotional state.

[1460] "Gesture data" is data that includes the user's movements and hand movements, and is analyzed to recognize meaningful actions.

[1461] "Voice data" refers to data containing sounds and words uttered by a user, and is used to identify the content and emotional tone of the speech.

[1462] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analyzed data.

[1463] "Eye tracking technology" is a technology that acquires a user's gaze data and analyzes the direction of their gaze and the point of focus.

[1464] A "robot" is an automated mechanical device used in a factory to interact with workers.

[1465] "Feedback" refers to the ratings and opinions collected from users that are used as training data for generative AI models.

[1466] "Interaction" is the exchange of information between a user and a system (or robot).

[1467] This invention is a system that uses emotion recognition algorithms and generative AI models to achieve smooth communication between robots and workers in factories. This system is implemented by combining the following hardware and software.

[1468] Hardware

[1469] Camera: A device used to capture facial expressions and gestures of employees. Used to obtain video data.

[1470] Microphone: A device for collecting employee voice data. Used to analyze speech content and emotional tone.

[1471] Eye tracking sensor: Used to capture employee gaze data and analyze user intent.

[1472] Speaker: A device for providing generated audio messages to employees.

[1473] Robot: An automated mechanical device that interacts with workers in a factory.

[1474] software

[1475] Python: A programming language for implementing the entire program.

[1476] TensorFlow: A machine learning library for building emotion recognition algorithms and generative AI models.

[1477] OpenCV: A library for preprocessing video data and identifying facial positions and feature points.

[1478] PyAudio: A library for preprocessing and denoising audio data.

[1479] Eye tracking library: A library for acquiring gaze data and analyzing user intent.

[1480] System Operation

[1481] 1. Collecting input

[1482] The camera captures the employee's facial expression data and uses OpenCV to detect the face position and feature points.

[1483] The microphone collects audio data and PyAudio is used to filter out background noise.

[1484] An eye-tracking sensor captures the employee's gaze data.

[1485] 2. Data Preprocessing

[1486] Noise is removed from the video data using OpenCV, and facial feature points are identified.

[1487] The audio data is denoised using PyAudio to extract clear audio information.

[1488] The gaze data is preprocessed by an eye-tracking library to identify the gaze point.

[1489] 3. Analysis and Recognition

[1490] Emotion recognition algorithms analyze facial, voice, and gesture data to identify the employee's emotional state, and also analyze gaze data to complement the emotional state.

[1491] 4. Analysis results and response generation

[1492] The server uses a generative AI model to generate an appropriate response based on the analysis of the emotion recognition algorithm, which can be in the form of a voice message, text, or other format.

[1493] 5. Gather feedback and self-study

[1494] User feedback is sent to the server, and the generative AI model uses this feedback to self-train.

[1495] Specific examples

[1496] If an employee says with a troubled expression, "I need another person to do this task," the robot will respond by generating a voice message saying, "Would you like additional support?" This action allows the employee to quickly receive the necessary support, improving work efficiency within the factory.

[1497] Prompt Sentence Examples

[1498] "The user (employee) looks troubled and says, 'I need another person to do this task.' What would a generative AI model generate as an appropriate response to this situation?"

[1499] This system combines emotion recognition algorithms and generative AI models to enable smooth communication between workers and robots, which is expected to improve work efficiency and worker safety in factories.

[1500] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1501] Step 1:

[1502] The user provides input by speaking into the camera and microphone or waving their hand, while simultaneously capturing gaze data using an eye-tracking sensor. This allows the user to generate video, audio, and gaze data.

[1503] Step 2:

[1504] The device captures the user's facial expression data with a camera and detects the position and feature points of the face using OpenCV. The input is video data, and the output is the position and feature points of the face.

[1505] Step 3:

[1506] The device collects the user's voice data with a microphone and uses PyAudio to filter out background noise. The input is the voice data, and the output is clear, noise-removed voice information.

[1507] Step 4:

[1508] The device acquires gaze data using an eye-tracking sensor and uses an eye-tracking library to identify the gaze direction and gaze point. The input is the gaze data, and the output is the gaze direction and gaze point.

[1509] Step 5:

[1510] The data collected by the device (facial expression data, voice data, gaze data) is temporarily stored in a buffer and then sent to the server. The input is a set of preprocessed data, and the output is the data sent to the server.

[1511] Step 6:

[1512] The server preprocesses the received data and uses an emotion recognition algorithm to identify the user's emotional state. The input is the received data, and the output is the identified emotional state (e.g., "confused").

[1513] Step 7:

[1514] The server uses a generative AI model to generate an appropriate response based on the analysis results. The input is the emotional state identification result and the prompt sentence, and the output is the generated response (e.g., "Would you like to arrange additional support?").

[1515] Step 8:

[1516] The server sends the generated response to the terminal, which then provides it to the user: a voice message played through the speaker, a text message displayed on the screen. The input is the generated response, and the output is the response provided to the user.

[1517] Step 9:

[1518] The user provides feedback such as evaluation and opinion on the system's response. The input is the user's feedback, which is sent by the terminal to the server.

[1519] Step 10:

[1520] The server collects feedback from users and uses it as training data for the model. This allows the generative AI model to self-learn and improve its accuracy. The input is the feedback data, and the output is an updated generative AI model.

[1521] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1522] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1523] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1524] [Fourth embodiment]

[1525] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1526] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1527] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1528] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1529] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1530] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1531] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1532] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1533] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1534] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1535] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1536] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1537] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1538] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user, and its details are as follows.

[1539] Collecting input

[1540] User

[1541] The user provides input to the device, for example, by using a camera and microphone to record their facial expressions and voice.

[1542] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time.

[1543] Terminal

[1544] The device includes hardware for collecting facial expression data, gesture data, and voice data from the user, including a camera, a microphone, and an eye-tracking sensor.

[1545] server

[1546] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1547] Data Preprocessing

[1548] server

[1549] The server pre-processes the received data, which includes noise removal, necessary data format conversion, and extraction of the user's facial features.

[1550] Examples: Removing background noise from audio data to clarify what the user is saying and the tone. Identifying the position of the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.).

[1551] Analysis and Recognition

[1552] server

[1553] The server uses emotion recognition algorithms to analyze the user's facial expression data and identify their emotional state, gesture recognition algorithms to recognize meaningful actions from the user's hand and body movements, a voice analysis engine to identify the content and emotional tone of the user's speech from the voice data, and eye-tracking technology to analyze the user's gaze data and recognize their intentions.

[1554] For example: If a user smiles, an emotion recognition algorithm will determine this as "happiness." If a user waves, a gesture recognition algorithm will recognize this as "greeting." From the user's voice, it will be analyzed that the user said "hello" and that the emotional tone is friendly. Eye-tracking technology will recognize that the user is looking at a specific icon on the screen.

[1555] Communication generation

[1556] server

[1557] The server generates an appropriate response based on the analysis results, leveraging generative AI models to create the optimal response based on the user's emotional state and intent. This response can be in the form of audio, text, or video.

[1558] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[1559] Terminal

[1560] The terminal provides the generated response sent from the server to the user, either played through the speaker in the case of a voice message or displayed on the screen in the case of a text message.

[1561] Gathering feedback and self-learning

[1562] User

[1563] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[1564] server

[1565] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[1566] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1567] The above is an embodiment of the present invention. This system solves the problems that existed in conventional communication methods, and enables smooth communication, particularly with people with disabilities and animals.

[1568] The processing flow will be explained below.

[1569] Step 1:

[1570] User

[1571] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[1572] Step 2:

[1573] Terminal

[1574] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[1575] Step 3:

[1576] Terminal

[1577] The device sends the collected data to a server, where video data, audio data, and gaze data are sent together.

[1578] Step 4:

[1579] server

[1580] The server preprocesses the received data: for video data, noise reduction is performed and a face detection algorithm is used to identify the location and features of the user's face; for audio data, a noise reduction filter is applied to extract clear audio information.

[1581] Step 5:

[1582] server

[1583] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. It then uses a gesture analysis algorithm to analyze the user's movements and recognize specific actions. It then uses a voice analysis engine to analyze the voice data and identify the content and emotional tone of the speech. Based on the gaze data, it uses eye-tracking technology to analyze the user's intentions.

[1584] Step 6:

[1585] server

[1586] The server uses a generative AI model to generate an appropriate response based on the analysis results, which is encoded in the optimal format (audio, text, video, etc.) depending on the user's emotional state and intent.

[1587] Step 7:

[1588] Terminal

[1589] The terminal receives the generated response sent by the server, which is played through the speaker if it is a voice message, or displayed on the screen if it is a text message.

[1590] Step 8:

[1591] User

[1592] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[1593] Step 9:

[1594] server

[1595] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve its accuracy.

[1596] Example 1

[1597] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1598] Conventional communication systems lack the ability to comprehensively analyze multiple input data, including not only a user's facial expressions and voice, but also their gaze and gestures. This makes it difficult to accurately understand the user's emotional state and intentions and generate appropriate responses based on them. Furthermore, they lack a mechanism for effectively utilizing user feedback and continuously improving the system's response accuracy. In particular, there have been technical challenges in achieving smooth communication with people with disabilities and animals.

[1599] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1600] In this invention, the server includes: means for collecting a user's facial expression data using a camera and microphone; means for transmitting the collected data to the server; means for performing noise reduction and format conversion on the data received by the server to extract the user's facial features; means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing audio data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means for including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses to the user; and means for collecting user feedback and using it as training data for the model. This enables the system to comprehensively analyze a variety of user input information and generate more accurate responses. Furthermore, the system's accuracy can be improved using user feedback, enabling smoother communication, particularly with people with disabilities and animals.

[1601] "Facial expression data" is video information captured using a camera to show the facial expressions of a user.

[1602] "Voice data" refers to acoustic information captured using a microphone, such as the content of a user's speech and the tone of their voice.

[1603] "Noise reduction" is the process of removing background noise and unwanted sounds from audio data.

[1604] "Format conversion" is the operation of converting collected data into a format that is easy to analyze.

[1605] "Feature points" are information that indicates the main points of the user's face (eyes, mouth, eyebrows, etc.).

[1606] An "emotion recognition algorithm" is a method or program for analyzing facial expression data to identify a user's emotional state.

[1607] "Gesture data" is video information that captures the movements of the user's hands and body.

[1608] A "gesture recognition algorithm" is a method or program for analyzing gesture data and recognizing meaningful actions from user movements.

[1609] "Gaze data" is information captured using eye-tracking technology to determine where a user is looking.

[1610] "Eye tracking technology" is a technology for capturing the direction of a user's gaze and point of gaze.

[1611] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[1612] "Feedback" is the process of collecting user feedback and opinions on responses.

[1613] "Training data" is collected feedback information used to improve the performance of a model.

[1614] A "response" is a message or instruction that the generative AI model provides to the user based on the analysis results.

[1615] This invention is a system that realizes smooth communication with users by combining emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, gaze tracking technology, and user feedback functions. This system functions through the interaction of three parties: the terminal, the server, and the user.

[1616] Collecting input

[1617] The user uses a camera and microphone to provide their facial expressions and voice to the device. For example, when the user speaks in front of the camera, their facial expressions and voice are captured in real time.

[1618] The device includes hardware for collecting facial, gesture, and voice data from the user, including a camera, microphone, and eye-tracking sensors, such as a Logitech C920 camera, a Blue Yeti microphone, and a Tobii Eye Tracker 5.

[1619] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1620] Data Preprocessing

[1621] The server then performs noise reduction and format conversion on the received data, which includes removing background noise from the audio data and making the content and tone of the user's speech clearer. It also locates the user's face from the video data and extracts feature points (eyes, mouth, eyebrows, etc.).

[1622] Analysis and Recognition

[1623] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. For example, the Microsoft Azure Face API is used. It also uses a gesture recognition algorithm to recognize meaningful actions (e.g., greeting, pointing) from the user's hand and body movements. It uses a voice analysis engine (e.g., Google Speech-to-Text) to identify the content and emotional tone of the user's speech from the voice data. Furthermore, it uses eye-tracking technology to analyze the user's gaze data and recognize the user's intentions.

[1624] Communication generation

[1625] The server generates an appropriate response based on the analysis results. It uses a generative AI model (e.g., OpenAI GPT-4) to create the optimal response (voice, text, or video) based on the user's emotional state and intent. For example, in response to the utterance "Hello," it generates the text message "Hello! How is your day?"

[1626] The terminal provides the generated response sent from the server to the user, either playing it through the speaker in the case of a voice message or displaying it on the screen in the case of a text message.

[1627] Gathering feedback and self-learning

[1628] The user provides feedback on the system's response, in the form of a rating of whether the response was appropriate.

[1629] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, allowing the system to continuously improve itself and generate more accurate responses.

[1630] For example, if a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1631] An example of a prompt sentence is a user's utterance, "What is the weather today?" The server analyzes this voice data and generates a response, "The weather today is sunny."

[1632] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1633] Step 1:

[1634] Providing user input

[1635] The user uses a camera and microphone to provide their facial expressions and voice to the device. The user speaks in front of the camera, and the video and audio are captured in real time. Specifically, a camera (Logitech C920) and a microphone (Blue Yeti) are used. The input of this step is the user's facial expressions and voice, and the output is the captured raw video and audio data.

[1636] Step 2:

[1637] Data collection

[1638] The device uses a camera, microphone, and eye-tracking sensor to collect facial expression data, voice data, and gaze data. Tobii Eye Tracker 5 is used for eye-tracking. For example, when the user looks to the right, the gaze data is captured. The input of this step is the raw data output from step 1, and the output is organized facial expression data, voice data, and gaze data.

[1639] Step 3:

[1640] Sending data

[1641] The device sends the collected data to the server in real time. A secure protocol (e.g., HTTPS) is used for communication. The input of this step is the data collected in step 2, and the output is the data transfer to the server.

[1642] Step 4:

[1643] Temporary data storage

[1644] The server temporarily stores the data received from the terminal in storage. This process allows for efficient management of large amounts of data. The input to this step is the transferred data, which is the output of step 3, and the output is the data stored in storage.

[1645] Step 5:

[1646] Noise reduction and format conversion

[1647] The server removes background noise from the audio data and unnecessary parts from the video data to extract facial features. A noise-canceling algorithm is used to remove noise from the audio data, and a video analysis algorithm is used to extract facial features. The input to this step is the data saved in step 4, and the output is clear pre-processed audio data and facial expression data with extracted features.

[1648] Step 6:

[1649] Emotion and movement analysis

[1650] The server uses an emotion recognition algorithm to analyze the user's facial expression data and identify their emotional state. Specifically, the Microsoft Azure Face API classifies emotions into categories such as "happiness," "sadness," and "surprise." It also uses a gesture recognition algorithm to analyze the user's movements. The input of this step is the clear data output from Step 5, and the output is the user's emotional state and gesture identification results.

[1651] Step 7:

[1652] Audio analysis

[1653] The server uses a speech recognition engine to analyze the speech content and emotional tone from the audio data. Google Speech-to-Text is used for this process. The input is the audio data processed in step 5. For example, "Hello" and a friendly tone are identified. The output is text data of the speech content and emotional tone.

[1654] Step 8:

[1655] Line of sight analysis

[1656] The server uses eye-tracking technology to analyze the user's gaze data and recognize their intention. Specifically, it analyzes data showing the user's gaze on a specific icon on the screen. The input is the gaze data from step 2, and the output is text data about the user's gaze area and intention.

[1657] Step 9:

[1658] Generating a response

[1659] The server generates an appropriate response based on the analysis results. Using a generative AI model (e.g., OpenAI GPT-4), it generates the optimal response based on the user's emotional state and intention. For example, in response to the utterance "Hello," a response such as "Hello! How is your day?" is generated. The input is the analysis results from steps 6 to 8, and the output is the generated voice or text message.

[1660] Step 10:

[1661] Sending a Response

[1662] The server sends the generated response to the terminal. The data is transferred to the terminal in the appropriate format (audio, text, video). The input is the data generated in step 9, and the output is the response data to the terminal.

[1663] Step 11:

[1664] Providing a response

[1665] The terminal provides the generated response received from the server to the user, either playing it through the speaker if it is a voice message or displaying it on the screen if it is a text message. The input is the response data from step 10, and the output is the response the user receives.

[1666] Step 12:

[1667] Providing feedback

[1668] The user provides feedback on the system's response. Using a smartphone or PC, the user evaluates whether the response was appropriate. The input for this step is the user's impression, and the output is feedback data.

[1669] Step 13:

[1670] Collecting feedback

[1671] The server accumulates the feedback collected from users and uses it as training data for the generative AI model, which improves the accuracy of the response. The input of this step is the feedback data from step 12, and the output is the training data.

[1672] Step 14:

[1673] Training the model

[1674] The server retrains the generative AI model with the newly collected feedback, which allows the system to self-improve. The input for this step is the training data from step 13, and the output is an updated generative AI model.

[1675] (Application example 1)

[1676] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1677] Conventional customer service systems in brick-and-mortar stores have difficulty accurately grasping customers' emotions and intentions, making smooth communication difficult. Furthermore, they rely heavily on the skills of employees, making it difficult to maintain a consistent level of service quality. When employees are dealing with a large number of customers, they need to respond efficiently and appropriately, but there has been no effective way to achieve this.

[1678] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1679] In this invention, the server includes: means for analyzing a user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to identify the content of speech and emotional tone; a generative AI model for generating appropriate responses based on the analysis results; means for providing the generated responses to the user; means for collecting feedback from the user and using it as training data for the model; means for capturing customer facial expressions and gestures using a camera; and means for collecting customer voices using a microphone and displaying the generated responses on the smart glasses based on the analysis results. This makes it possible to grasp the customer's emotions and intentions in real time and generate and provide appropriate responses. As a result, high-quality service can be provided and smooth communication can be achieved without relying on the skills of employees.

[1680] An "emotion recognition algorithm" is a technology for analyzing a user's facial expression data and identifying their emotional state.

[1681] "Means for analyzing user gesture data" refers to technology that detects the movements of the user's hands and body and recognizes the meaning of those movements.

[1682] "Means for analyzing voice data" refers to technology that identifies the content and emotional tone of a user's speech and extracts appropriate information.

[1683] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analysis results.

[1684] The "means for providing a response to a user" refers to a technique for transmitting the generated response to a user in the form of voice, text, or the like.

[1685] "Means for collecting feedback and using it as model training data" refers to techniques for collecting user evaluations and using that data to improve the system's performance.

[1686] "Camera-based means" refers to the device or system used to acquire video data.

[1687] "Smart glasses" are devices equipped with a camera, display, microphone, etc., that display various information in real time when worn by a user.

[1688] "Means using a microphone" refers to a device or system for capturing audio data.

[1689] This invention is a customer service support system for brick-and-mortar stores that facilitates smooth communication between store employees wearing smart glasses and customers. The system combines emotion recognition algorithms, gesture recognition technology, voice analysis technology, generative AI models, and gaze tracking technology.

[1690] Hardware and Software Configuration

[1691] Smart glasses: Smart glasses are equipped with a camera, microphone, display, and eye-tracking sensor to collect customers' facial expressions, gestures, and voice in real time.

[1692] Server: The server receives the data sent by the smart glasses, analyzes and processes it. Specifically, the following processes are performed:

[1693] Emotion recognition: Based on the video data captured by the camera, an emotion recognition algorithm is run to identify emotions from the user's facial expressions.

[1694] Gesture recognition: Uses technology to recognize user gestures (e.g., hand movements) from video data captured by a camera.

[1695] Voice analysis: Analyzes the content and emotional tone of the user's speech based on the voice data captured by the microphone.

[1696] Eye tracking: Analyzes data acquired from an eye tracking sensor to identify the object and direction the user is looking at.

[1697] Generative AI Model: Uses an artificial intelligence model that generates the optimal response based on the analysis results. This response can be in the form of text or voice.

[1698] Operation procedure and example

[1699] As an example of use in a physical store, the system would be implemented in the following steps:

[1700] 1. User data collection:

[1701] A customer approaches an employee wearing smart glasses and begins a conversation. The camera in the glasses captures the customer's facial expressions in real time, and the microphone collects their voice.

[1702] 2. Data Analysis:

[1703] The collected data is sent to a server, where an emotion recognition algorithm analyzes the facial expression data to identify the customer's emotional state (e.g., joy, surprise), while gesture recognition technology detects hand movements and other gestures, and voice analysis technology analyzes the content and emotional tone of spoken words.

[1704] 3. Generate a response:

[1705] Based on the analysis, the generative AI model generates an appropriate response. For example, if it determines that a customer is tired, it generates a voice message saying, "Would you like to take a break?" This response can be displayed on the smart glasses' display or delivered through the audio speaker.

[1706] 4. Eye-tracking information provision:

[1707] Using eye-tracking technology, if a customer is looking at a particular product, information about that product will be displayed on the smart glasses' display and provided to the customer.

[1708] 5. Gathering feedback and self-learning:

[1709] Feedback from customers and employees is collected and sent to a server, where it is used as training data for a generative AI model, helping to improve the accuracy of the system.

[1710] Examples and prompts

[1711] Examples:

[1712] A customer begins a conversation with an employee wearing smart glasses. The camera captures the customer's face and determines their emotion as "joy." The microphone collects the voice, saying "Hello, how are you today?", which is then analyzed by the smart glasses. Based on the analysis results, the generative AI model suggests, "Hello! Would you like to check out our new promotions?"

[1713] Example prompt sentence:

[1714] A user said: Hello, how are you today?

[1715] Generate an appropriate response to it.

[1716] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1717] Step 1:

[1718] Device data collection

[1719] When a customer approaches an employee wearing smart glasses, the device (smart glasses) uses a camera to capture the user's facial expression and gesture data, and also uses a microphone to collect the user's voice data.

[1720] Input: User facial expression video, gesture video, voice

[1721] Output: Facial expression data, gesture data, voice data

[1722] Step 2:

[1723] Sending data

[1724] The terminal transmits the collected facial expression data, gesture data, and voice data to a server.

[1725] Input: facial expression data, gesture data, voice data

[1726] Output: Facial expression data, gesture data, and voice data sent to the server

[1727] Step 3:

[1728] Server data preprocessing

[1729] The server receives the transmitted data and performs noise removal and necessary data format conversion. Specifically, it removes noise from audio data and extracts facial features from video data.

[1730] Input: Facial expression data, gesture data, and voice data sent to the server

[1731] Output: Noise-removed audio data, facial expression data with extracted features

[1732] Step 4:

[1733] Performing emotion and gesture recognition

[1734] The server uses an emotion recognition algorithm to analyze the facial expression data to identify the user's emotional state, and a gesture recognition algorithm to analyze the gesture data to recognize the user's actions.

[1735] Input: Facial expression data and gesture data with extracted feature points

[1736] Output: Emotional state, recognized gestures

[1737] Step 5:

[1738] Performing audio analysis

[1739] The server uses a voice analysis engine to identify the user's vocal content and emotional tone from the transmitted voice data, filtering out background noise to ensure clear voice data.

[1740] Input: Denoised audio data

[1741] Output: Vocal content, emotional tone

[1742] Step 6:

[1743] Performing eye tracking

[1744] Data obtained from the device's eye-tracking sensor is sent to a server, which then analyzes the user's eye-gaze data to identify the object and direction the user is looking at.

[1745] Input: Gaze data

[1746] Output: Object and direction being gazed at

[1747] Step 7:

[1748] Response generation using generative AI models

[1749] The server uses a generative AI model to generate an appropriate response based on the user's emotional state, recognized gestures, vocal content, emotional tone, and gaze data, which can be in the form of text or voice.

[1750] Input: Emotional state, recognized gestures, vocal content, emotional tone, gaze data

[1751] Output: The generated response (text or audio)

[1752] Step 8:

[1753] Providing a generated response

[1754] The server sends the generated response to the terminal, which then provides the generated response to the user: if it is a voice message, it is played through the speaker, and if it is a text message, it is displayed on the display of the smart glasses.

[1755] Input: The generated response

[1756] Output: The audio or text message provided to the user.

[1757] Step 9:

[1758] Collecting feedback

[1759] Users provide feedback on the responses provided, which is collected and sent to a server that uses it as training data for a generative AI model.

[1760] Input: User feedback

[1761] Output: Feedback data, improving the generative AI model

[1762] Through the above processing steps, a customer service support system is realized that can accurately grasp the emotions and intentions of customers and provide appropriate responses in real time.

[1763] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1764] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's emotional state and intentions with high accuracy and generates appropriate responses based on that. The system functions through interactions between the terminal, the server, and the user, and its details are as follows:

[1765] Collecting input

[1766] User

[1767] The user provides input to the device through a camera and microphone, for example by speaking or waving their hand.

[1768] Example: When a user speaks in front of the camera, their facial expressions and voice are captured in real time, and it also recognizes whether their gaze is directed in a specific direction.

[1769] Terminal

[1770] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor. The input data is temporarily stored in a buffer.

[1771] server

[1772] The server receives the data sent from the terminal and temporarily stores the data in storage for preprocessing.

[1773] Data Preprocessing

[1774] server

[1775] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the user's face and its features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is similarly preprocessed to identify the user's gaze point.

[1776] Examples include: filtering out background noise from audio data to clarify what the user is saying and the tone of their speech; locating the user's face from video data and extracting feature points (eyes, mouth, eyebrows, etc.); and determining which part of the screen the user is looking at from eye-tracking data.

[1777] Analysis and Recognition

[1778] Emotion Engine

[1779] The emotion engine in the server analyzes the collected facial expression, voice, and gesture data to comprehensively identify the user's emotional state. It also uses gaze data to complement the emotional state.

[1780] For example: If the user is smiling, the emotion engine will determine "happiness." If the user waves, the gesture recognition algorithm will recognize this as a "greeting." From the audio, it will be analyzed that the user said "hello" and that the emotional tone is friendly. From the gaze data, it will be recognized that the user is looking at a specific icon on the screen.

[1781] Analysis results and response generation

[1782] server

[1783] The server uses a generative AI model to generate an appropriate response based on the analysis results of the emotion engine. This response is encoded in the optimal format (audio, text, video, etc.) according to the user's emotional state and intention.

[1784] Example: If the user is determined to be tired, a voice message is generated saying "Do you want to take a break?" If the user is greeting someone, a text message is displayed saying "Hello! How's your day going?"

[1785] Terminal

[1786] The terminal provides the generated response sent from the server to the user: if it is a voice message, it is played through the speaker, if it is a text message, it is displayed on the screen.

[1787] Gathering feedback and self-learning

[1788] User

[1789] The user provides feedback such as an evaluation and opinion on the system's response, including an evaluation of the appropriate response and suggestions for improvement.

[1790] server

[1791] The server accumulates the feedback collected from users and stores it in a database. This feedback data is used as training data for the generative AI model and is used to improve the model's accuracy.

[1792] Example: When a user evaluates a response as "appropriate," that feedback is sent to the server and used as training data for the generative AI model.

[1793] The above is an embodiment of the present invention. By combining this system with an emotion engine, it is possible to solve the problems that existed in conventional communication methods, and to realize smooth communication, particularly with people with disabilities and animals.

[1794] The processing flow will be explained below.

[1795] Step 1:

[1796] User

[1797] The user speaks to the terminal, for example, saying "hello" and smiling.

[1798] Step 2:

[1799] Terminal

[1800] The device uses a camera and microphone to capture the user's facial expression and voice data in real time, and an eye-tracking sensor to capture the user's gaze data, which is temporarily stored in a buffer.

[1801] Step 3:

[1802] Terminal

[1803] The device sends the collected data, including facial expression data, voice data, and gaze data, to a server.

[1804] Step 4:

[1805] server

[1806] The server pre-processes the received data, removing background noise from the video and applying a face detection algorithm to identify facial features. The audio data is also cleared through a noise reduction filter.

[1807] Step 5:

[1808] server

[1809] The server sends the preprocessed data to the emotion engine, which analyzes the facial expression data to identify the user's emotional state. It also analyzes the gesture data to recognize the user's actions. It also analyzes the voice data using a natural language processing (NLP) engine to analyze the speech content and emotional tone. It also uses gaze data to recognize the user's gaze point and intention.

[1810] Step 6:

[1811] server

[1812] The server uses a generative AI model based on the analysis results of the emotion engine to generate an appropriate response, which can be in the form of voice, text, or video depending on the user's emotional state and intentions.

[1813] Step 7:

[1814] Terminal

[1815] The device receives the generated response sent by the server, plays it over the speaker if it is a voice message, or displays it on the screen if it is a text message.

[1816] Step 8:

[1817] User

[1818] The user provides feedback on the system's response, such as rating it "good" or "needs improvement."

[1819] Step 9:

[1820] server

[1821] The server collects user feedback and stores it in a database, which is used as training data for the generative AI model to improve the accuracy of the system.

[1822] Example 2

[1823] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1824] Conventional communication systems have struggled to accurately recognize a user's emotions and intentions and generate appropriate responses based on them. In particular, the technology for comprehensively analyzing various inputs, such as voice, facial expressions, and gaze, to comprehensively assess the user's state was immature. Furthermore, the functionality for collecting feedback on the generated responses and using it to improve the system was insufficient. This made it difficult to communicate smoothly, particularly with people with disabilities and animals.

[1825] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1826] In this invention, the server includes means for capturing and temporarily storing a user's facial expression data with a camera, means for collecting and temporarily storing the user's voice data with a microphone, means for acquiring and temporarily storing the user's gaze data with an eye tracking sensor, means for preprocessing the received data and identifying the position and feature points of the face using a face detection algorithm, means for applying a noise reduction filter to extract clear voice information, means for an emotion engine to analyze the facial expression data, voice data, and gesture data to identify the user's emotional state, means for generating an appropriate response based on the analysis results using a generative AI model, means for providing the generated response to the user in the form of a voice message, a text message, or a video, and means for collecting user feedback, storing it in a database, and using it as training data for the generative AI model. This makes it possible to recognize user emotions and intentions with high accuracy, generate appropriate responses, and improve the system based on the feedback.

[1827] "User" refers to an individual who operates the system and provides input data.

[1828] "Terminal" refers to a device that collects and temporarily stores a user's facial expression data, voice data, and gaze data.

[1829] "Server" refers to a central processing unit that receives data sent from terminals and performs pre-processing, analysis, and response generation.

[1830] "Facial Expression Data" refers to information about a user's facial expressions captured through a camera.

[1831] "Voice data" refers to acoustic information about a user's speech collected through a microphone.

[1832] "Gaze data" refers to information about the direction of a user's gaze and point of gaze obtained through an eye tracking sensor.

[1833] "Preprocessing" refers to the initial processing of received data, such as noise reduction and feature point extraction.

[1834] A "face detection algorithm" refers to a computational method for identifying the position and features of a face from video data.

[1835] "Noise reduction filter" refers to a processing technology for removing unwanted background noise from audio data and extracting clear audio information.

[1836] "Emotion engine" refers to a program that analyzes facial expression data, voice data, and gesture data to identify the user's emotional state.

[1837] "Gesture data" refers to information about hand and body movements made by a user.

[1838] "Generative AI model" refers to an artificial intelligence model used to generate appropriate responses based on analytical results.

[1839] "Response" refers to the output information, such as a voice message, text message, or video, generated by a generative AI model.

[1840] "Feedback" refers to information regarding evaluations and opinions provided by users to the system.

[1841] "Database" means a data storage system for storing and managing collected feedback and other data.

[1842] This invention provides a system that combines an emotion engine to achieve smooth communication with users. This system recognizes the user's facial expression data, voice data, and gaze data with high accuracy, and generates appropriate responses based on that data. The system functions through the interaction of three parties: the terminal, the server, and the user.

[1843] Collecting input

[1844] User

[1845] The user speaks or gestures into the device, for example, by waving while saying "hello," and the camera and microphone capture this input.

[1846] Examples:

[1847] When a user turns to the device and says "hello" and waves, the camera captures the user's facial expressions and gestures, and the microphone collects audio.

[1848] Terminal

[1849] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data using an eye-tracking sensor. These input data are temporarily stored in a buffer.

[1850] Examples:

[1851] The device uses a camera and an eye-tracking sensor to capture when the user is looking at a specific part of the screen, and stores this as gaze data.

[1852] server

[1853] The server receives the data sent from the terminal and temporarily stores it in storage for preprocessing.

[1854] Data Preprocessing

[1855] server

[1856] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[1857] Examples:

[1858] The server detects the user's face from the video data and extracts the positions of the eyes and mouth as feature points, and removes background noise from the audio data to extract clear audio information.

[1859] Analysis and Recognition

[1860] Emotion engine (on the server)

[1861] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also uses gaze data to complement the emotional state.

[1862] Examples:

[1863] The emotion engine recognizes when a user is smiling as "happy," when a user is waving as "greeting," and when a user says "hello" in a friendly tone.

[1864] Generating a response

[1865] server

[1866] Based on the analysis results of the emotion engine, the generative AI model generates an appropriate response, which can be encoded in the form of audio, text, video, etc.

[1867] Examples:

[1868] If the server senses that the user is tired, it generates a voice message saying "Do you want to take a break?" If the user says "Hello", it generates a text message saying "Hello! How's your day?"

[1869] Providing a response to the user

[1870] Terminal

[1871] The device receives a response from the server and provides it to the user: a voice message is played over the speaker, a text message is displayed on the screen.

[1872] Examples:

[1873] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays text on the screen saying, "Hello! How are you today?"

[1874] Gathering feedback and self-learning

[1875] User

[1876] The user provides feedback to the system, including the appropriateness of the response and areas for improvement.

[1877] Examples:

[1878] The user gives feedback saying, "This response was appropriate."

[1879] server

[1880] The server stores the feedback in a database and uses it as training data for the generative AI model, improving the accuracy of the system.

[1881] Examples:

[1882] Based on user feedback, the server retrains the generative AI model to improve the accuracy of the next response.

[1883] The above is an embodiment of the present invention. This system enables communication that combines an emotion engine, and is expected to facilitate smooth communication, especially with people with disabilities and animals.

[1884] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1885] Step 1:

[1886] Collecting User Input

[1887] User

[1888] The user speaks or gestures into the device, and the camera and microphone capture the user's input. For example, if the user waves while saying "hello," their facial expression and voice are collected.

[1889] Input: User voice and gestures

[1890] Output: Video and audio data captured by the camera and microphone

[1891] Specific behavior:

[1892] When a user speaks or waves at the device, the camera and microphone capture the input, which is then temporarily stored in a buffer.

[1893] Step 2:

[1894] Data buffering

[1895] Terminal

[1896] The device captures the user's facial expression data with a camera, collects audio data with a microphone, and acquires the user's gaze data with an eye-tracking sensor, and temporarily stores this input data in a buffer.

[1897] Input: Captured video data, audio data, and gaze data

[1898] Output: Buffered input data

[1899] Specific behavior:

[1900] The device uses a camera, microphone, and eye-tracking sensor to collect data and temporarily store it in a buffer.

[1901] Step 3:

[1902] Data Transfer and Storage

[1903] server

[1904] The server receives the input data sent from the terminal and temporarily stores it in storage for preprocessing.

[1905] Input: Input data stored in a buffer

[1906] Output: Data temporarily saved in storage

[1907] Specific behavior:

[1908] The server receives input data from the terminal and temporarily stores it in storage, which is used for subsequent preprocessing.

[1909] Step 4:

[1910] Data Preprocessing

[1911] server

[1912] The server preprocesses the received data. For video data, noise reduction is performed and a face detection algorithm is used to identify the location of faces and their features. For audio data, a noise reduction filter is applied to extract clear audio information. Gaze data is also preprocessed to identify the user's gaze point.

[1913] Input: Data stored in storage

[1914] Output: Denoised and feature-pointed data

[1915] Specific behavior:

[1916] The server detects the user's face from the video data and extracts the positions of the eyes and mouth. It also removes background noise from the audio data to extract clear audio. It also identifies the user's gaze point from the gaze data.

[1917] Step 5:

[1918] Emotion analysis and recognition

[1919] Server (emotion engine)

[1920] The emotion engine analyzes facial expression, voice, and gesture data to identify the user's emotional state, and also analyzes gaze data to complement the emotional state.

[1921] Input: Preprocessed facial expression data, speech data, and gaze data

[1922] Output: User's emotional state

[1923] Specific behavior:

[1924] The emotion engine identifies a user's emotional state (e.g., joy, surprise, sadness, etc.) from subtle facial movements, vocal tone, and gesture patterns, and complements emotion analysis by using gaze data to understand where the user is looking.

[1925] Step 6:

[1926] Generating a response

[1927] server

[1928] Based on the emotion engine's analysis, a generative AI model generates an appropriate response, which can be encoded in the form of audio, text, or video.

[1929] Input: User's emotional state

[1930] Output: The generated response message

[1931] Specific behavior:

[1932] The generative AI model uses prompts based on the analysis results to generate responses such as "Hello! How are you today?"

[1933] Step 7:

[1934] Providing a response

[1935] Terminal

[1936] The terminal provides the user with the response sent by the server, either a voice message played over the speaker or a text message displayed on the screen.

[1937] Input: The generated response message

[1938] Output: Providing a response to the user (voice, text, etc.)

[1939] Specific behavior:

[1940] The device plays a voice message over the speaker saying, "Would you like to take a break?" and displays a text message on the screen saying, "Hello! How's your day going?"

[1941] Step 8:

[1942] Collecting feedback

[1943] User

[1944] The user provides feedback on the system's response, including whether the response was appropriate and what could be improved.

[1945] Input: User feedback (ratings and opinions)

[1946] Output: Feedback information sent to the server

[1947] Specific behavior:

[1948] The user provides feedback such as "this response was appropriate," which is sent to the server via the terminal.

[1949] Step 9:

[1950] Self-study

[1951] server

[1952] The server stores the collected feedback in a database and uses this data as training data for the generative AI model, thereby improving the accuracy of the system.

[1953] Input: User feedback data

[1954] Output: Feedback reflected as training data

[1955] Specific behavior:

[1956] The server analyzes the collected feedback and uses it to retrain the generative AI model so that it generates a better response next time.

[1957] (Application example 2)

[1958] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1959] There is a need to improve work efficiency in factories while ensuring the safety of employees. However, it is not easy for robots to quickly and accurately recognize situations in which employees are in trouble or need assistance, and generate appropriate responses. Furthermore, understanding emotional states and intentions requires integrating and analyzing multiple data sets, including facial expressions, gestures, voice, and gaze, making the process complex. There is a need for a way to solve these problems and enable smooth communication between employees and robots.

[1960] The identification process by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes: means for analyzing the user's (employee's) facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing the user's gesture data to recognize meaningful actions; means for analyzing voice data to distinguish between spoken content and emotional tone; means for analyzing the user's gaze data using eye-tracking technology to recognize the user's intention; means including a generative AI model that generates appropriate responses based on the analysis results; means for providing the generated responses as voice messages through a speaker; means for collecting user feedback and using it as training data for the model; and means installed in a robot that accurately identifies the emotional state and intention of the employee through interaction with the employee and generates appropriate responses. This makes it possible to quickly and accurately recognize situations in which the employee is in difficulty and provide appropriate support. This improves work efficiency in the factory and ensures employee safety.

[1961] An "emotion recognition algorithm" is a computational method for analyzing facial expression data to identify a user's emotional state.

[1962] "Gesture data" is data that includes the user's movements and hand movements, and is analyzed to recognize meaningful actions.

[1963] "Voice data" refers to data containing sounds and words uttered by a user, and is used to identify the content and emotional tone of the speech.

[1964] A "generative AI model" is an artificial intelligence model that generates appropriate responses based on analyzed data.

[1965] "Eye tracking technology" is a technology that acquires a user's gaze data and analyzes the direction of their gaze and the point of focus.

[1966] A "robot" is an automated mechanical device used in a factory to interact with workers.

[1967] "Feedback" refers to the ratings and opinions collected from users that are used as training data for generative AI models.

[1968] "Interaction" is the exchange of information between a user and a system (or robot).

[1969] This invention is a system that uses emotion recognition algorithms and generative AI models to achieve smooth communication between robots and workers in factories. This system is implemented by combining the following hardware and software.

[1970] Hardware

[1971] Camera: A device used to capture facial expressions and gestures of employees. Used to obtain video data.

[1972] Microphone: A device for collecting employee voice data. Used to analyze speech content and emotional tone.

[1973] Eye tracking sensor: Used to capture employee gaze data and analyze user intent.

[1974] Speaker: A device for providing generated audio messages to employees.

[1975] Robot: An automated mechanical device that interacts with workers in a factory.

[1976] software

[1977] Python: A programming language for implementing the entire program.

[1978] TensorFlow: A machine learning library for building emotion recognition algorithms and generative AI models.

[1979] OpenCV: A library for preprocessing video data and identifying facial positions and feature points.

[1980] PyAudio: A library for preprocessing and denoising audio data.

[1981] Eye tracking library: A library for acquiring gaze data and analyzing user intent.

[1982] System Operation

[1983] 1. Collecting input

[1984] The camera captures the employee's facial expression data and uses OpenCV to detect the face position and feature points.

[1985] The microphone collects audio data and PyAudio is used to filter out background noise.

[1986] An eye-tracking sensor captures the employee's gaze data.

[1987] 2. Data Preprocessing

[1988] Noise is removed from the video data using OpenCV, and facial feature points are identified.

[1989] The audio data is denoised using PyAudio to extract clear audio information.

[1990] The gaze data is preprocessed by an eye-tracking library to identify the gaze point.

[1991] 3. Analysis and Recognition

[1992] Emotion recognition algorithms analyze facial, voice, and gesture data to identify the employee's emotional state, and also analyze gaze data to complement the emotional state.

[1993] 4. Analysis results and response generation

[1994] The server uses a generative AI model to generate an appropriate response based on the analysis of the emotion recognition algorithm, which can be in the form of a voice message, text, or other format.

[1995] 5. Gather feedback and self-study

[1996] User feedback is sent to the server, and the generative AI model uses this feedback to self-train.

[1997] Specific examples

[1998] If an employee says with a troubled expression, "I need another person to do this task," the robot will respond by generating a voice message saying, "Would you like additional support?" This action allows the employee to quickly receive the necessary support, improving work efficiency within the factory.

[1999] Prompt Sentence Examples

[2000] "The user (employee) looks troubled and says, 'I need another person to do this task.' What would a generative AI model generate as an appropriate response to this situation?"

[2001] This system combines emotion recognition algorithms and generative AI models to enable smooth communication between workers and robots, which is expected to improve work efficiency and worker safety in factories.

[2002] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[2003] Step 1:

[2004] The user provides input by speaking into the camera and microphone or waving their hand, while simultaneously capturing gaze data using an eye-tracking sensor. This allows the user to generate video, audio, and gaze data.

[2005] Step 2:

[2006] The device captures the user's facial expression data with a camera and detects the position and feature points of the face using OpenCV. The input is video data, and the output is the position and feature points of the face.

[2007] Step 3:

[2008] The device collects the user's voice data with a microphone and uses PyAudio to filter out background noise. The input is the voice data, and the output is clear, noise-removed voice information.

[2009] Step 4:

[2010] The device acquires gaze data using an eye-tracking sensor and uses an eye-tracking library to identify the gaze direction and gaze point. The input is the gaze data, and the output is the gaze direction and gaze point.

[2011] Step 5:

[2012] The data collected by the device (facial expression data, voice data, gaze data) is temporarily stored in a buffer and then sent to the server. The input is a set of preprocessed data, and the output is the data sent to the server.

[2013] Step 6:

[2014] The server preprocesses the received data and uses an emotion recognition algorithm to identify the user's emotional state. The input is the received data, and the output is the identified emotional state (e.g., "confused").

[2015] Step 7:

[2016] The server uses a generative AI model to generate an appropriate response based on the analysis results. The input is the emotional state identification result and the prompt sentence, and the output is the generated response (e.g., "Would you like to arrange additional support?").

[2017] Step 8:

[2018] The server sends the generated response to the terminal, which then provides it to the user: a voice message played through the speaker, a text message displayed on the screen. The input is the generated response, and the output is the response provided to the user.

[2019] Step 9:

[2020] The user provides feedback such as evaluation and opinion on the system's response. The input is the user's feedback, which is sent by the terminal to the server.

[2021] Step 10:

[2022] The server collects feedback from users and uses it as training data for the model. This allows the generative AI model to self-learn and improve its accuracy. The input is the feedback data, and the output is an updated generative AI model.

[2023] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transmits voice data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the voice data.

[2024] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2025] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the robot 414.

[2026] The emotion identification model 59 as an emotion engine may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to an emotion map (see FIG. 9), which is a specific mapping. Similarly, the emotion identification model 59 may determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2027] FIG. 9 is a diagram illustrating an emotion map 400 on which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. Emotions closer to the center of the concentric circles are more primitive. Emotions representing states and actions arising from a state of mind are arranged on the outer edges of the concentric circles. The concept of emotion includes both affect and mental states. Emotions generally generated from reactions occurring in the brain are arranged on the left side of the concentric circles. Emotions generally induced by situational judgment are arranged on the right side of the concentric circles. Emotions generally generated from reactions occurring in the brain and induced by situational judgment are arranged on the upper and lower sides of the concentric circles. Furthermore, the emotion of "pleasure" is arranged on the upper side of the concentric circles, and the emotion of "discomfort" is arranged on the lower side. In this way, in the emotion map 400, multiple emotions are mapped based on the structure by which emotions are generated, and emotions that tend to occur simultaneously are mapped close to each other.

[2028] These emotions are distributed in the 3 o'clock direction on emotion map 400, and typically fluctuate between relief and anxiety. In the right half of emotion map 400, situational awareness dominates over internal sensations, resulting in a sense of calm.

[2029] The inside of emotion map 400 represents what is going on in the mind, and the outside of emotion map 400 represents behavior, so the further you go outside emotion map 400, the more visible the emotions become (the more they are expressed in behavior).

[2030] Human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. Emotions can also be created for robots, automobiles, and motorcycles, based on various balances, such as posture and remaining battery life. When these balances deviate from the ideal, a state of discomfort is indicated, and when they approach the ideal, a state of pleasure is indicated. An emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on Voice Emotion Recognition and Emotional Brain Physiological Signal Analysis Systems, Tokushima University, Doctoral Dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map lists emotions belonging to the "reaction" domain, where sensation is dominant. The right half of the emotion map lists emotions belonging to the "situation" domain, where situational awareness is dominant.

[2031] The emotion map defines two emotions that promote learning. One is a negative emotion on the situation side, around the middle of "repentance" or "reflection." In other words, this occurs when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is a positive emotion on the response side, around "desire." In other words, this occurs when the robot experiences positive feelings such as "I want more" or "I want to know more."

[2032] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​indicating each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple pieces of training data that are combinations of user input and emotion values ​​indicating each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions that are located close to each other have similar values, as in the emotion map 900 shown in FIG. 10. FIG. 10 shows an example in which multiple emotions, "relieved," "calm," and "reassuring," have similar emotion values.

[2033] The system according to the present disclosure has been described above mainly with respect to the functions of the data processing device 12, but the system according to the present disclosure is not necessarily implemented on a server. The system according to the present disclosure may be implemented as a general information processing system. The present disclosure may be implemented, for example, as a software program running on a personal computer or an application running on a smartphone, etc. The method according to the present disclosure may be provided to users in the form of SaaS (Software as a Service).

[2034] In the above embodiment, an example was given in which the specific processing is performed by one computer 22, but the technology of the present disclosure is not limited to this, and the specific processing may be distributed and performed by a plurality of computers including the computer 22. For example, the data generation model 58 may be provided in an external device of the data processing device 12, and data may be generated in the external device in accordance with input data.

[2035] In the above embodiment, an example in which the specific processing program 56 is stored in the storage 32 has been described, but the technology of the present disclosure is not limited to this. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-transitory storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-transitory storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes the specific processing in accordance with the specific processing program 56.

[2036] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[2037] It is not necessary to store all of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store all of the specific processing program 56 in the storage 32; only a portion of the specific processing program 56 may be stored.

[2038] The hardware resource for executing a specific process can be any of the following processors: An example of a processor is a CPU, which is a general-purpose processor that functions as a hardware resource for executing a specific process by executing software, i.e., a program. Another example of a processor is a dedicated electrical circuit, such as an FPGA (Field-Programmable Gate Array), a PLD (Programmable Logic Device), or an ASIC (Application Specific Integrated Circuit), which is a processor with a circuit configuration designed specifically for executing a specific process. Each processor has built-in or connected memory, and each processor uses the memory to execute the specific process.

[2039] The hardware resource that executes the specific processing may be configured with one of these various processors, or may be configured with a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Also, the hardware resource that executes the specific processing may be a single processor.

[2040] As an example of a system configured with a single processor, first, one processor is configured by combining one or more CPUs and software, and this processor functions as a hardware resource that executes a specific process. Second, there is a system that uses a processor that realizes the functions of an entire system including multiple hardware resources that execute a specific process on a single IC chip, as typified by SoC (System-on-a-chip). In this way, a specific process is realized using one or more of the above-mentioned various processors as hardware resources.

[2041] Furthermore, the hardware structure of these various processors can be, more specifically, an electric circuit that combines circuit elements such as semiconductor devices. The specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps may be deleted, new steps may be added, or the processing order may be rearranged, without departing from the spirit of the invention.

[2042] The above-described description and illustrations are a detailed explanation of the parts related to the technology of the present disclosure and are merely an example of the technology of the present disclosure. For example, the above description of the configuration, functions, actions, and effects is an explanation of an example of the configuration, functions, actions, and effects of the parts related to the technology of the present disclosure. Therefore, it goes without saying that unnecessary parts may be deleted, new elements may be added, or replacements may be made to the above-described description and illustrations within the scope of the gist of the technology of the present disclosure. Furthermore, to avoid confusion and facilitate understanding of the parts related to the technology of the present disclosure, the above-described description and illustrations omit explanations of common technical knowledge that do not require particular explanation to enable the implementation of the technology of the present disclosure.

[2043] All publications, patent applications, and technical standards mentioned in this specification are herein incorporated by reference to the same extent as if each individual publication, patent application, or technical standard was specifically and individually indicated to be incorporated by reference.

[2044] The following is further disclosed regarding the above embodiment.

[2045] (Claim 1)

[2046] means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state;

[2047] means for analyzing user gesture data to recognize meaningful actions;

[2048] means for analyzing the audio data to identify speech content and emotional tone;

[2049] a means including a generative AI model for generating an appropriate response based on the analysis results;

[2050] means for providing the generated response to a user;

[2051] A means of collecting user feedback and using it as training data for the model;

[2052] A system including:

[2053] (Claim 2)

[2054] 10. The system of claim 1, further comprising means for analyzing user gaze data and recognizing user intent using eye-tracking technology.

[2055] (Claim 3)

[2056] 10. The system of claim 1, further comprising means for providing the generated response as a voice message through a speaker.

[2057] "Example 1"

[2058] (Claim 1)

[2059] A means for collecting facial expression data of a user using a camera and a microphone;

[2060] means for transmitting the collected data to a server;

[2061] A means for removing noise and converting the format of the data received by the server to extract the facial feature points of the user;

[2062] means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state;

[2063] means for analyzing user gesture data to recognize meaningful actions;

[2064] means for analyzing the audio data to identify speech content and emotional tone;

[2065] A means for analyzing a user's gaze data using gaze tracking technology and recognizing the user's intention;

[2066] a means including a generative AI model for generating an appropriate response based on the analysis results;

[2067] means for providing the generated response to a user;

[2068] A means of collecting user feedback and using it as training data for the model;

[2069] A system including:

[2070] (Claim 2)

[2071] 10. The system of claim 1, further comprising means for the generative AI model to generate an optimal response in the form of voice, text, or video based on the analysis results, according to the user's emotional state and intention.

[2072] (Claim 3)

[2073] 10. The system of claim 1, further comprising means for providing the generated response as a voice message through a speaker.

[2074] "Application Example 1"

[2075] (Claim 1)

[2076] means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state;

[2077] means for analyzing user gesture data to recognize meaningful actions;

[2078] means for analyzing the audio data to identify vocal content and emotional tone;

[2079] a means including a generative AI model for generating an appropriate response based on the analysis results;

[2080] means for providing the generated response to a user;

[2081] A means of collecting user feedback and using it as training data for the model;

[2082] A means for capturing images of facial expressions and gestures of customers using a camera;

[2083] a means for collecting the customer's voice using a microphone and displaying a generated response on the smart glasses based on the analysis result;

[2084] A system including:

[2085] (Claim 2)

[2086] 10. The system of claim 1, further comprising means for analyzing user gaze data and recognizing user intent using eye-tracking technology.

[2087] (Claim 3)

[2088] 10. The system of claim 1, further comprising means for providing the generated response as a voice message through a speaker.

[2089] "Example 2: Combining Emotion Engines"

[2090] (Claim 1)

[2091] A means for capturing and temporarily storing facial expression data of a user with a camera;

[2092] means for collecting and temporarily storing user voice data via a microphone;

[2093] A means for acquiring and temporarily storing user gaze data using an eye tracking sensor;

[2094] means for pre-processing the received data and identifying the location and features of a face using a face detection algorithm;

[2095] a means for applying a noise reduction filter to extract clear audio information;

[2096] a means for an emotion engine to analyze facial expression data, voice data, and gesture data to identify the user's emotional state;

[2097] A means for generating an appropriate response based on the analysis results using a generative AI model;

[2098] means for providing the generated response to the user in the form of an audio message, a text message, or a video;

[2099] A means to collect user feedback, store it in a database, and use it as training data for generative AI models;

[2100] A system including:

[2101] (Claim 2)

[2102] 10. The system of claim 1, further comprising means for identifying a user's point of gaze using the analyzed gaze data.

[2103] (Claim 3)

[2104] 10. The system of claim 1, further comprising means for providing the generated response as a voice message through a speaker.

[2105] "Application example 2 when combining emotion engines"

[2106] (Claim 1)

[2107] means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state;

[2108] means for analyzing user gesture data to recognize meaningful actions;

[2109] means for analyzing the audio data to identify speech content and emotional tone;

[2110] a means including a generative AI model for generating an appropriate response based on the analysis results;

[2111] means for providing the generated response to a user;

[2112] A means of collecting user feedback and using it as training data for the model;

[2113] A means for analyzing a user's gaze data and recognizing the user's intention using gaze tracking technology;

[2114] a means for providing the response generated by the generative AI model as an audio message via a speaker;

[2115] A means to be installed on the robot to identify with high accuracy the emotional state and intention of the employee through interactions with the robot and generate an appropriate response;

[2116] A system including:

[2117] (Claim 2)

[2118] 10. The system of claim 1, further comprising means for improving work efficiency in a factory and ensuring worker safety.

[2119] (Claim 3)

[2120] 10. The system of claim 1, further comprising means for identifying an emotional state of an employee and providing appropriate support or feedback to improve work efficiency. [Explanation of symbols]

[2121] 10, 210, 310, 410 Data Processing Systems 12 Data Processing Device 14 Smart Devices 214 Smart Glasses 314 Headset-type terminal 414 Robot< / url:> < / url:> < / url:> < / url:>

Claims

1. means for analyzing the user's facial expression data using an emotion recognition algorithm to identify the user's emotional state; means for analyzing user gesture data to recognize meaningful actions; means for analyzing the audio data to identify speech content and emotional tone; a means including a generative AI model for generating an appropriate response based on the analysis results; means for providing the generated response to a user; A means of collecting user feedback and using it as training data for the model; A system including:

2. The system of claim 1 , further comprising means for analyzing user gaze data and recognizing user intent using eye-tracking technology.

3. 10. The system of claim 1, further comprising means for providing the generated response as a voice message through a speaker.

Images