System

Abstract

A system is provided.SOLUTION: A system, comprising: means for performing machine learning using a dataset of a dual-license test; means for responding to questions from a driver using a model obtained by the machine learning; means for automatically generating test questions using the model; means for analyzing test results answered by the driver and evaluating proficiency of the driver; and means for generating and sending personalized follow messages based on the proficiency evaluation.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] Ride-sharing will be legalized in April 2024, allowing drivers without a Class 2 driver's license to transport passengers for a fee. This will require taxi companies to take on new responsibilities for driver training and supervision, but the existing training system makes it difficult to handle this burden efficiently. In addition, drivers themselves often have side jobs, so they need to receive training efficiently within a limited amount of time. Furthermore, individual attention will be required based on each driver's level of proficiency, but traditional training methods cannot handle this. There is a need for an training system that can solve these issues efficiently and effectively. [Means for solving the problem]

[0005] To solve the above problems, the present invention provides the following means. A system is provided that includes: a means for performing machine learning using a dataset from a Class 2 driver's license test; a means for responding to questions from drivers using a model obtained by the machine learning; a means for automatically generating test questions using the model; a means for analyzing the results of the test answers given by the driver and evaluating the driver's proficiency; and a means for generating and sending individual follow-up messages based on the proficiency evaluation. This system allows taxi companies to reduce the burden of driver education without allocating additional personnel, and allows drivers to study efficiently at their own pace. Furthermore, by adding a means for identifying areas in which the driver's proficiency is low from the answers to the generated test questions and focusing on questions related to those areas, the effectiveness of individual learning can be improved.

[0006] The "Class 2 License Examination Dataset" is a collection of data that includes questions and answers related to the Class 2 license examination required for taxi drivers, as well as related materials.

[0007] "Machine learning" is a technique that uses data to train computational models to make predictions and classifications on new data.

[0008] A "generative AI model" is an artificial intelligence model trained to perform natural language processing or generation tasks, and in this context refers to large-scale models such as GPT-3.

[0009] A "chatbot" is an automated response system that interacts with users in natural language and answers their questions.

[0010] "Automatic test question generation" is the process of mechanically creating test questions based on specific subject areas or topics.

[0011] "Response analysis" is the process of evaluating test and question responses entered by a driver to determine their accuracy.

[0012] "Proficiency" is an indicator of how well a driver understands and is able to execute specific knowledge or skills.

[0013] "Follow-up messages" are messages that provide instructions, advice, or additional learning that are sent individually based on the driver's proficiency.

[0014] A "request" is a request for a specific operation or information sent from a user terminal to a server.

[0015] "Response" refers to a response or information sent back from the server to the user terminal.

[0016] "Driver education" is the educational process by which drivers employed by taxi companies acquire the knowledge and skills necessary to perform their jobs. [Brief explanation of the drawings]

[0017] [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

[0018] 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.

[0019] First, the terms used in the following description will be explained.

[0020] 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).

[0021] 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.

[0022] 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.

[0023] 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.

[0024] 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."

[0025] [First embodiment]

[0026] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0027] 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.

[0028] 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).

[0029] 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.

[0030] 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.

[0031] 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.

[0032] 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.

[0033] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0034] 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.

[0035] 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.

[0036] 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.

[0037] 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."

[0038] This invention relates to a driver training system that uses a generative AI model on a communication platform operated by a taxi company. Specifically, the system performs machine learning using a dataset from a Class 2 driver's license test, and uses the resulting generative AI model to perform chatbots that respond to drivers' questions, automatic test generation, and proficiency analysis. Below, we provide a natural language explanation of the system's program processing and a concrete example.

[0039] Question and Answer System

[0040] 1. Enter and submit your question

[0041] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0042] Terminal: Sends the entered question to the server as an HTTP request.

[0043] 2. Receiving and analyzing questions

[0044] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0045] 3. Generate and return the answer

[0046] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[0047] Server: The generated answer is sent back to the user's device as an HTTP response.

[0048] 4. View Answers

[0049] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0050] Test Generation and Examination Systems

[0051] 1. Submit a test request

[0052] User: The driver selects "I want to take the test" in the learning application.

[0053] Device: Sends a test request to the server as an HTTP request.

[0054] 2. Generate and submit test questions

[0055] Server: Based on the received request, the test generation module automatically generates test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0056] Server: Returns the generated test questions to the user's device as an HTTP response.

[0057] 3. Display the test screen and submit your answers

[0058] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[0059] 4. Analyzing responses and sending results

[0060] Server: The answer analysis module analyzes the received answers and judges whether they are correct or incorrect. It also calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[0061] Terminal: The received results are displayed on the user's screen, allowing the driver to check their self-evaluation results.

[0062] Follow-up message system

[0063] 1. Proficiency Assessment

[0064] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0065] 2. Create and send a follow message

[0066] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0067] Server: Sends the generated follow message to the user's device as an HTTP response.

[0068] 3. Displaying follow messages

[0069] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0070] As described above, this system utilizes a communication platform operated by a taxi company to achieve effective and efficient driver training using a generative AI model. By linking each module, it is possible to provide learning support tailored to each individual's level of proficiency, significantly reducing the burden of training.

[0071] The processing flow will be explained below.

[0072] Question and Answer System

[0073] Step 1:

[0074] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[0075] Step 2:

[0076] Terminal: Sends the entered question to the server as an HTTP request.

[0077] Step 3:

[0078] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0079] Step 4:

[0080] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[0081] Step 5:

[0082] Server: The generated answer is sent back to the user's device as an HTTP response.

[0083] Step 6:

[0084] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0085] Test Generation and Examination Systems

[0086] Step 1:

[0087] User: Selects "I want to take a test" in a learning app.

[0088] Step 2:

[0089] Device: Sends a test request to the server as an HTTP request.

[0090] Step 3:

[0091] Server: Based on the received request, the test generation module automatically generates test questions, using a generative AI model to generate questions such as "Choose the best route when picking up a passenger using a taxi."

[0092] Step 4:

[0093] Server: Returns the generated test questions to the user's device as an HTTP response.

[0094] Step 5:

[0095] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[0096] Step 6:

[0097] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[0098] Step 7:

[0099] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[0100] Step 8:

[0101] Server: Returns the evaluation results to the user device as an HTTP response.

[0102] Step 9:

[0103] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[0104] Follow-up message system

[0105] Step 1:

[0106] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0107] Step 2:

[0108] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0109] Step 3:

[0110] Server: Sends the generated follow message to the user's device as an HTTP response.

[0111] Step 4:

[0112] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0113] Example 1

[0114] 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."

[0115] Conventional driver training systems have had issues with the difficulty of providing effective training tailored to individual proficiency levels, resulting in low overall training efficiency. Furthermore, question-answering and test question generation are often done manually, which requires time and effort. Furthermore, there are also issues with insufficient evaluation of driver proficiency, resulting in a lack of appropriate feedback and follow-up.

[0116] 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.

[0117] In this invention, the server includes a device that performs machine learning using a dataset of the Class 2 driver's license test, a device that responds to questions from the driver using a model obtained by the machine learning, a device that automatically generates test questions using the model, a device that analyzes the test results answered by the driver and evaluates the driver's proficiency, a device that generates and sends an individual follow-up message based on the proficiency evaluation, a device that receives the driver's input and sends it to the server as an HTTP request, a device that analyzes the content of the questions using natural language processing and generates answers using a generative AI model, and a terminal that receives and displays the generated answers to show to the driver. This makes it possible to quickly answer the driver's questions, automatically generate test questions according to the individual proficiency level, and provide appropriate evaluations and individual follow-ups.

[0118] The "Type 2 license test" is a driver's license test required to drive passenger vehicles such as taxis and buses.

[0119] A "dataset" is a collection of data collected for a specific purpose and used in machine learning and data analysis.

[0120] "Machine learning" is a technology that allows computers to learn patterns and knowledge from data and perform specific tasks automatically.

[0121] A "model" is an algorithm or function generated by machine learning that makes predictions or classifications on new data.

[0122] "Driver" refers to a professional driver who drives a vehicle such as a taxi or bus.

[0123] A "device that responds to questions" is a system that receives questions from the driver and generates and provides appropriate answers.

[0124] "Test questions" are test questions given to assess a driver's knowledge and skills.

[0125] "Analysis" is the process of examining data or information in detail to find meaning and patterns.

[0126] "Proficiency" is an indicator of how much knowledge and skill a driver has in a particular field.

[0127] A "follow-up message" is a customized message based on the proficiency assessment to encourage additional learning and improvement for the driver.

[0128] An "HTTP request" is a data request sent from a client to a server.

[0129] "Natural language processing" is a technology that allows computers to understand human language, and includes morphological analysis and semantic analysis.

[0130] A "generative AI model" is an artificial intelligence model that has been trained to perform tasks such as text generation and question answering.

[0131] A "terminal" is an electronic device that allows a user to interact with a system through an interface.

[0132] MODE FOR CARRYING OUT THE INVENTION

[0133] This invention relates to a driver training system based on machine learning techniques, specifically using a dataset from the Class 2 driver's license test. The system uses a generative AI model to respond to driver questions, automatically generate test questions, analyze test results, and provide personalized follow-up messages based on the driver's proficiency assessment.

[0134] System Configuration

[0135] 1. Hardware and Software Overview

[0136] Device: An electronic device (e.g., smartphone, tablet, computer) equipped with a learning application used by a driver.

[0137] Server: A central system for analyzing questions, generating test questions, analyzing answers, and generating follow-up messages.

[0138] Natural Language Processing (NLP) libraries, such as spaCy.

[0139] Generative AI models: Use large language models such as GPT-3.

[0140] Database: Stores past questions from the Class 2 license exam and driving knowledge materials.

[0141] System Operation

[0142] 1. Question and Answer System

[0143] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0144] Terminal: Send this input to the server as an HTTP request.

[0145] Server: Passes the received question to the chatbot module, and performs morphological analysis using a natural language processing library (e.g., spaCy) to analyze keywords and sentence structure.

[0146] Server: Enter the prompt "How often do I need to renew my driver's license?" into a generative AI model (e.g., GPT-3) and generate an appropriate answer. The generated text is "A driver's license generally needs to be renewed every five years."

[0147] Server: Sends the generated answer to the terminal as an HTTP response.

[0148] Terminal: The received answer is displayed on the user interface, and the driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0149] 2. Test Generation and Examination Systems

[0150] User: The driver selects "I want to take the test" in the learning application.

[0151] Device: Sends a test request to the server as an HTTP request.

[0152] Server: Based on the received request, the test generation module references the database and generates new test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0153] Server: Sends the generated test questions to the device as an HTTP response.

[0154] Terminal: The test questions are displayed in a user interface for the driver to answer. The driver enters the answers and sends them to the server as HTTP requests.

[0155] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, calculates the driver's proficiency, generates an evaluation result, and sends it to the user's device as an HTTP response.

[0156] Terminal: The received evaluation results are displayed on the user interface, allowing the driver to check their self-evaluation results.

[0157] 3. Follow-up message system

[0158] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects areas of low proficiency (e.g., road traffic laws).

[0159] Server: Generate a personalized follow-up message for each subject area that the follow-up module detects. For example, create a message that says, "Review your basic knowledge of road traffic laws. Additional learning materials are available here."

[0160] Server: Sends the generated follow message to the user's device as an HTTP response.

[0161] Terminal: The received follow-up message is displayed on the user interface, and the driver can receive specific learning instructions.

[0162] Examples of concrete examples and prompts

[0163] Example 1: A driver asks, "How often do I need to renew my driver's license?" and the server uses an AI model to generate and provide the answer, "Basically, a driver's license needs to be renewed every five years."

[0164] Example prompt sentence:

[0165] Question-answering system prompt: "How often do I need to renew my driver's license?"

[0166] Test generation system prompt: "Generate test questions based on the Class 2 taxi driver's license exam."

[0167] Follow-up message system prompt: "This driver failed multiple questions about road traffic laws. Please generate a follow-up message to help them study again."

[0168] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0169] Processing flow

[0170] Question and Answer System

[0171] Step 1: Enter and submit your question

[0172] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0173] Input: The driver's text input.

[0174] Terminal: Sends the entered question to the server as an HTTP request.

[0175] Output: HTTP request to the server.

[0176] Step 2: Receiving and parsing the question

[0177] Server: Receives the HTTP request and retrieves the inquiry.

[0178] Input: The HTTP request received from the device.

[0179] Server: Using an NLP library (e.g., spaCy), morphologically analyze the question and extract keywords and sentence structure.

[0180] Output: Parsed keywords and sentence structure.

[0181] Step 3: Generate an answer

[0182] Server: The analyzed question content is input as a prompt to the generative AI model (e.g., GPT-3).

[0183] Input: Parsed keywords and sentence structure.

[0184] Server: Generates an appropriate answer using the AI ​​model. It generates the text, "Driver's licenses generally need to be renewed every five years."

[0185] Output: The generated answer text.

[0186] Step 4: Return and view your responses

[0187] Server: Sends the generated answer to the user's device as an HTTP response.

[0188] Input: The generated answer text.

[0189] Output: HTTP response to the user's device.

[0190] Device: The received answer is displayed on the screen. The driver can confirm the answer: "Driver's licenses generally need to be renewed every five years."

[0191] Test Generation and Examination Systems

[0192] Step 1: Submit a test request

[0193] User: The driver selects "I want to take the test" in the learning application.

[0194] Input: Driver test request.

[0195] Terminal: Send this request to the server as an HTTP request.

[0196] Output: HTTP request to the server.

[0197] Step 2: Generate test questions

[0198] Server: Receives requests and passes instructions to the test generation module.

[0199] Input: Exam request.

[0200] Server: Refers to the database and automatically generates new test questions. For example, it generates a question such as "Choose the best route when picking up a passenger by taxi."

[0201] Output: The generated test questions.

[0202] Step 3: Submit and view the test questions

[0203] Server: Sends the generated test questions to the user's device as an HTTP response.

[0204] Input: The generated test questions.

[0205] Output: HTTP response to the user's device.

[0206] Terminal: Displays the received test questions on the screen so that the driver can answer them.

[0207] Input: Show test question.

[0208] Output: Driver answer input.

[0209] Step 4: Submit and analyze responses

[0210] Terminal: After entering the driver's answers, send them to the server as an HTTP request.

[0211] Input: Driver's answer.

[0212] Output: HTTP request to the server.

[0213] Server: The answer analysis module analyzes the received answer data, determines whether the answer is correct, and calculates the driver's proficiency level.

[0214] Input: Received response data.

[0215] Output: Evaluation result.

[0216] Step 5: Send and view results

[0217] Server: Sends the evaluation results to the user device as an HTTP response.

[0218] Input: Evaluation result.

[0219] Output: HTTP response to the user's device.

[0220] Terminal: The received evaluation results are displayed on the user's screen. Drivers can check their own evaluation results.

[0221] Input: Display evaluation results.

[0222] Output: Driver verification.

[0223] Follow-up message system

[0224] Step 1: Assess proficiency

[0225] Server: The response analysis module evaluates the driver's proficiency based on past test results.

[0226] Input: Past test results.

[0227] Server: Detect low proficiency in a specific subject (e.g., road traffic law).

[0228] Output: Low proficiency area detection results.

[0229] Step 2: Generate a follow-up message

[0230] Server: The follow-up module generates individual follow-up messages for low-proficiency areas.

[0231] Input: Information in low proficiency areas.

[0232] Server: Generate a message such as "Review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0233] Output: The generated follow message.

[0234] Step 3: Send and view follow-up messages

[0235] Server: Sends the generated follow message to the user's device as an HTTP response.

[0236] Input: The generated follow message.

[0237] Output: HTTP response to the user's device.

[0238] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0239] Input: Show follow messages.

[0240] Output: Driver verification.

[0241] (Application example 1)

[0242] 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."

[0243] Current training for robot operators in factories relies on instruction by human trainers and static teaching materials, making it difficult to provide efficient training tailored to each operator's proficiency level. It is also difficult to respond to operator questions and provide appropriate teaching materials in real time, resulting in limited training efficiency and effectiveness. This invention aims to solve these issues by using a generative AI model to provide a flexible and efficient training system tailored to each operator's proficiency level.

[0244] 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.

[0245] In this invention, the server includes a means for performing machine learning using a dataset of the Class 2 license examination, a means for responding to questions from operators using a model obtained by the machine learning, and a means for automatically generating test questions using the model. This not only enables quick and accurate responses to questions from operators, but also makes it possible to provide test questions tailored to the learning status of each individual operator and appropriately generate and send follow-up messages based on their proficiency.

[0246] The "Class 2 license examination dataset" refers to a wide range of test questions and answer data used to evaluate an operator's skills and knowledge.

[0247] "Machine learning" refers to the technology in which computers automatically learn using large amounts of data to generate algorithms and models for performing specific tasks.

[0248] A "model" refers to a data processing structure built using machine learning to predict or classify a specific outcome based on input data.

[0249] "Operator" refers to an employee who operates and manages robots and machines within a factory.

[0250] "Test Questions" refers to automatically generated test questions to evaluate an operator's knowledge and skills.

[0251] "Proficiency" refers to an indicator that evaluates how much an operator has acquired a particular skill or knowledge.

[0252] "Follow-up messages" refer to messages containing instructions or advice sent individually to encourage further learning and improvement based on the results of an operator's proficiency assessment.

[0253] "System" refers to the overall infrastructure and process for supporting training using machine learning models implemented according to this invention.

[0254] This invention is a system for supporting the training of operators of robots used in factories. It performs machine learning using a dataset from the Class 2 license examination, and uses the model to respond to the operator's questions, automatically generate tests, evaluate proficiency, and send follow-up messages.

[0255] Overall system overview

[0256] This system consists of three main functions:

[0257] 1. Question and Answer System

[0258] 2. Test Generation and Examination Systems

[0259] 3. Proficiency-based follow-up message system

[0260] Question and Answer System

[0261] The server receives a question from the operator and analyzes it using the chatbot module. Specifically, it uses natural language processing (NLP) technology to morphologically analyze the content of the question and extract its meaning. It then uses a generative AI model (e.g., GPT-3) to generate an appropriate answer and sends it back to the operator's device. The operator can then view the received answer on their device's screen.

[0262] Examples:

[0263] When an operator uses a smartphone to input a question such as, "Where is the emergency stop button on the robot?", the server receives the question, analyzes it, and generates and returns the answer, "The emergency stop button is on the left side of the robot's control panel."

[0264] Example prompt sentence:

[0265] I have a question about the emergency stop button on the robot. Can you tell me where it is?

[0266] Test Generation and Examination Systems

[0267] When the server receives a test request from an operator, it automatically generates test questions using a test generation module. The generated test questions are sent to the operator's terminal and displayed. The operator answers the questions and sends the answers to the server. The server analyzes the answers, determines whether they are correct or incorrect, evaluates the operator's proficiency, and returns the results.

[0268] Examples:

[0269] When an operator requests a test on their smartphone, the server generates a question such as "What is the normal maintenance cycle for a robot?" and sends it to the operator's device. When the operator enters and submits the answer, the answer is analyzed and the results and evaluation are sent back to the device.

[0270] Proficiency-based follow-up message system

[0271] The server evaluates the operator's proficiency based on past test results. Based on the evaluation results, follow-up messages are individually generated for specific areas and sent to the operator's terminal, allowing the operator to conduct additional learning in areas where their understanding is lacking.

[0272] Examples:

[0273] If the server detects low proficiency in "robot safety operation procedures," it generates a follow-up message such as "Please relearn the specific area. Please refer to the manual in the system for details," and sends it to the operator's terminal.

[0274] Example prompt sentence:

[0275] Your proficiency has been verified. You will need to refresh your knowledge in the following areas: Robot safety operating procedures.

[0276] Technologies and Tools Used

[0277] Hardware: Server, user device (smartphone)

[0278] Software: Natural Language Processing (NLP) module, generative AI model (e.g. GPT-3), HTTP communication protocol

[0279] The implementation of this system will not only enable effective and efficient training tailored to each operator's individual level of proficiency, but will also enable responses to questions in real time and the provision of appropriate learning instructions.

[0280] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0281] Step 1:

[0282] The user opens the learning application on their smartphone and inputs a question, for example, "Where is the emergency stop button on the robot?" This question is captured as input data on the device.

[0283] Step 2:

[0284] The device sends the entered question to the server as an HTTP request, which is in JSON format and contains the question itself.

[0285] Step 3:

[0286] The server passes the received question to the chatbot module, which uses natural language processing (NLP) to morphologically analyze the question text and extract the meaning of the question. Through this process, the question content is interpreted.

[0287] Step 4:

[0288] The server inputs the interpreted question into a generative AI model (e.g., GPT-3) and generates an appropriate answer. The generated answer might be something like, "There is an emergency stop button on the left side of the robot's control panel."

[0289] Step 5:

[0290] The server returns the generated answer to the user device as an HTTP response, which is also in JSON format and contains the answer text.

[0291] Step 6:

[0292] The device displays the received response on the user's screen, and the user can see the response on their smartphone screen: "There is an emergency stop button on the left side of the robot's control panel."

[0293] Step 7:

[0294] A user selects "I want to take a test" in a learning application. This request is received as input data on the device and sent to the server as an HTTP request.

[0295] Step 8:

[0296] The server automatically generates test questions using a test generation module based on the received test request. For example, a question such as "What is the normal maintenance cycle for a robot?" is generated.

[0297] Step 9:

[0298] The server sends the generated test questions to the user's terminal as an HTTP response. The terminal displays the received test questions on the user's screen, and the user answers the questions.

[0299] Step 10:

[0300] The user inputs the answer to the question, and the terminal sends it to the server as an HTTP request. This request contains data that includes the answer.

[0301] Step 11:

[0302] The server analyzes the received answers and judges whether they are correct or not. The answer analysis module then calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[0303] Step 12:

[0304] The device displays the received evaluation results on the user's screen, allowing the user to check their self-evaluation results on their smartphone screen.

[0305] Step 13:

[0306] The server evaluates the user's proficiency based on past test results and detects low proficiency in a specific area (e.g., robot safety operating procedures).

[0307] Step 14:

[0308] The server generates a follow-up message for the area of ​​low proficiency and sends it to the user's device as an HTTP response. For example, a follow-up message such as "Please restudy the specific area. For details, please refer to the system manual" is generated.

[0309] Step 15:

[0310] The device displays the received follow message on the user's screen, allowing the user to receive specific study instructions on the smartphone screen.

[0311] 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.

[0312] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The specific system configuration combines a machine learning model using a dataset from the Class 2 driver's license exam with an emotion engine to achieve an educational effect that takes into account the driver's emotional state.

[0313] Question and Answer System

[0314] 1. Enter and submit your question

[0315] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0316] Terminal: Sends the entered question to the server as an HTTP request.

[0317] 2. Receiving and analyzing questions

[0318] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0319] 3. Emotion Recognition and Answer Generation

[0320] Server: The emotion engine receives emotion data (e.g., facial recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[0321] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[0322] 4. Returning and displaying answers

[0323] Server: The generated answer is sent back to the user's device as an HTTP response.

[0324] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[0325] Test Generation and Examination Systems

[0326] 1. Submit a test request

[0327] User: The driver selects "I want to take the test" in the learning application.

[0328] Device: Sends a test request to the server as an HTTP request.

[0329] 2. Generate and submit test questions

[0330] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0331] Server: Returns the generated test questions to the user's device as an HTTP response.

[0332] 3. Display the test screen and submit your answers

[0333] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[0334] 4. Analyzing responses and sending results

[0335] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[0336] Server: Returns the evaluation results to the user device as an HTTP response.

[0337] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[0338] Follow-up message system

[0339] 1. Proficiency Assessment

[0340] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0341] 2. Create and send a follow message

[0342] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[0343] 3. Displaying follow messages

[0344] Server: Sends the generated follow message to the user's device as an HTTP response.

[0345] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[0346] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[0347] The processing flow will be explained below.

[0348] Question and Answer System

[0349] Step 1:

[0350] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[0351] Step 2:

[0352] Terminal: Sends the entered question to the server as an HTTP request.

[0353] Step 3:

[0354] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0355] Step 4:

[0356] Server: The emotion engine receives emotion data (e.g., facial expression recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[0357] Step 5:

[0358] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[0359] Step 6:

[0360] Server: The generated answer is sent back to the user's device as an HTTP response.

[0361] Step 7:

[0362] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[0363] Test Generation and Examination Systems

[0364] Step 1:

[0365] User: Selects "I want to take a test" in a learning app.

[0366] Step 2:

[0367] Device: Sends a test request to the server as an HTTP request.

[0368] Step 3:

[0369] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the optimal route when picking up a passenger using a taxi."

[0370] Step 4:

[0371] Server: Returns the generated test questions to the user's device as an HTTP response.

[0372] Step 5:

[0373] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[0374] Step 6:

[0375] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[0376] Step 7:

[0377] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[0378] Step 8:

[0379] Server: Returns the evaluation results to the user device as an HTTP response.

[0380] Step 9:

[0381] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[0382] Follow-up message system

[0383] Step 1:

[0384] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0385] Step 2:

[0386] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[0387] Step 3:

[0388] Server: Sends the generated follow message to the user's device as an HTTP response.

[0389] Step 4:

[0390] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[0391] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[0392] Example 2

[0393] 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."

[0394] Conventional driver training systems often provide uniform training without considering the individual needs or emotional state of each learner. Furthermore, detailed feedback based on test results is not provided, potentially reducing learning effectiveness. Furthermore, training utilizing emotional data has not been implemented, making it impossible to reduce learner stress. Therefore, a system that provides personalized training while taking into account the learner's emotional state was needed.

[0395] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending an individual follow-up message based on the proficiency evaluation, and means for analyzing emotion data from the user terminal and adding additional information based on the emotion to the answers generated by the model. This enables optimal education that takes into account the user's emotional state, thereby improving the effectiveness of the education.

[0396] "Machine learning" is a technology that allows a computer to analyze a given data set and automatically learn patterns and rules based on that data.

[0397] The "Class 2 Driver's License Examination Dataset" is a large-scale data set that includes questions and answers related to the Class 2 Driver's License Examination, as well as their trends and characteristics.

[0398] A "generative AI model" is a model that uses artificial intelligence to generate content such as text and images, and specifically includes natural language processing and generation modules.

[0399] A "chatbot" is a program that can interact with users and respond to their questions via text or voice.

[0400] "Emotion data" is data that indicates the user's emotional state, and examples include facial expression data and voice tone.

[0401] An "emotion engine" is an algorithm or system that analyzes emotional data to detect and evaluate a user's emotional state.

[0402] "Proficiency" is a measure of how well a driver understands or has mastered a particular subject or task.

[0403] A "follow-up message" is a message such as instruction or encouragement that is generated based on the user's learning progress or emotional state.

[0404] An "HTTP request" is a request for data sent by a client to a server, and is a protocol used by web browsers and applications to request data from a server.

[0405] An "HTTP response" is a data response that a server returns in response to a client's HTTP request, and includes requested data and status information.

[0406] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. Specific embodiments of the system are as follows.

[0407] Hardware and software used

[0408] This system uses the following hardware and software:

[0409] Server: A computer that performs major processing. It is equipped with a high-performance processor (e.g., Intel Xeon) and a large amount of memory (e.g., 64 GB or more).

[0410] Device: The device used by the driver, such as a smartphone or tablet (e.g., Android or iOS).

[0411] Generative AI model: An artificial intelligence model that generates text, for example using an advanced natural language processing model such as GPT-3.

[0412] Emotion engine: Software for analyzing user emotion data. Examples include Azure Cognitive Services' emotion recognition API.

[0413] NLP engine: Natural language processing software to analyze the question text, for example using spaCy or NLTK.

[0414] System processing flow

[0415] Answers to questions

[0416] A user opens a learning application and inputs a question, for example, "How often do I need to renew my driver's license?" The device sends this question to the server as an HTTP request. The server passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text. The server then uses an emotion engine to analyze emotion data (e.g., facial recognition and voice analysis data) from the user's device and determine the user's emotional state.

[0417] A generative AI model (e.g., GPT-3) is used to generate the optimal answer based on the analyzed question content and data from the emotion engine. For example, a standard answer such as "Your driver's license generally needs to be renewed every five years" is generated. If the user expresses anxiety, the server adds additional information such as "Don't worry, the process is simple."

[0418] The generated answer is then returned to the user terminal as an HTTP response, and the terminal displays the received answer on the user screen.

[0419] Test Generation and Taking

[0420] When a user selects "I want to take a test" in the learning application, the device sends a test request to the server. Based on the received request, the server uses a generative AI model to automatically generate test questions. For example, a question such as "Select the optimal route when picking up a passenger by taxi" is generated.

[0421] The generated test questions are sent back to the user's device, where the user can check them on the screen and enter their answers. The device then sends these answers to the server, which uses an answer analysis module to determine whether the answers are correct or incorrect and evaluate the user's proficiency. The results are then sent back to the user's device and displayed.

[0422] Follow message generation

[0423] The server evaluates the user's proficiency based on past test results and detects low proficiency in specific areas. It then uses an emotion engine to analyze the user's emotional state and generates appropriate follow-up messages based on that. For example, it generates specific learning instructions such as, "Please review your basic knowledge of road traffic laws. Take your time and review them carefully."

[0424] The follow message thus generated is sent to the user's terminal, and the user can check it on the screen.

[0425] Examples of concrete examples and prompts

[0426] For example, if a user asks, "How often do I need to renew my driver's license?", the following prompt is sent to the generative AI model:

[0427] User: "How many years do I need to renew my driver's license?"

[0428] Generative AI model: "Driver's licenses generally need to be renewed every five years."

[0429] Add emotion: "Don't worry, it's a simple process."

[0430] In this way, by combining a generative AI model with an emotion engine, this system is able to provide optimal training that takes into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a learning environment that is easier for drivers to understand and less stressful.

[0431] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0432] Step 1:

[0433] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0434] Input: Keyboard input data.

[0435] Output: Question text data.

[0436] What happens: The user types a question into a text input field in the application and presses the Enter key to confirm the question.

[0437] Step 2:

[0438] Terminal: Sends the entered question to the server as an HTTP request.

[0439] Input: Question text data.

[0440] Output: HTTP request packet.

[0441] Specific operation: Converts text data into HTTP packets and sends them over the network to the server.

[0442] Step 3:

[0443] Server: Passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text.

[0444] Input: HTTP request packet, text data.

[0445] Output: Parsed question semantic data.

[0446] What it does: The server decodes the HTTP packets, passes the text data to an NLP engine (e.g., spaCy or NLTK), and performs morphological analysis to extract the meaning of the question.

[0447] Step 4:

[0448] Server: Uses the emotion engine to receive and analyze emotion data (e.g., facial recognition, voice analysis data) from the user device.

[0449] Input: Emotion data, e.g., image or audio files.

[0450] Output: User's emotional state data.

[0451] Specific operation: Emotion data is passed to an emotion engine (e.g., Azure Cognitive Services emotion recognition API) for analysis, and the user's emotional state is output as numerical data.

[0452] Step 5:

[0453] Server: Using a generative AI model, it generates answers based on the analyzed question content and sentiment data.

[0454] Input: Question semantic data, emotional state data.

[0455] Output: Answer data in text format.

[0456] What it does: The semantic data and emotional state data of the question are incorporated into a prompt sentence, which is then sent to a generative AI model (e.g., GPT-3) to generate an answer, such as, "Your driver's license generally needs to be renewed every five years. Don't worry, it's a simple process."

[0457] Step 6:

[0458] Server: The generated answer is sent back to the user's device as an HTTP response.

[0459] Input: Textual response data.

[0460] Output: HTTP response packet.

[0461] Specific operation: The response data is formatted as an HTTP response and sent to the user terminal via the network.

[0462] Step 7:

[0463] Terminal: Displays the received answer on the user's screen.

[0464] Input: HTTP response packet.

[0465] Output: Screen display data.

[0466] What it does: It extracts text data from the HTTP response and displays it in a display field in the application. The user can see the answer on the screen: "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[0467] Step 8:

[0468] User: The driver selects "I want to take the test" in the learning application.

[0469] Input: A test-taking request.

[0470] Output: Test attempt selection data.

[0471] Specific behavior: A user clicks the "I want to take a test" button in the application to generate a test request.

[0472] Step 9:

[0473] Device: Sends a test request to the server as an HTTP request.

[0474] Input: Test attempt selection data.

[0475] Output: HTTP request packet.

[0476] Specific operation: The test selection data is converted into an HTTP request format and sent to the server via the network.

[0477] Step 10:

[0478] Server: Based on the received request, the server automatically generates test questions using a generative AI model.

[0479] Input: HTTP request packet.

[0480] Output: The generated test question data.

[0481] Specific operation: The request content is converted into a prompt for the generative AI model, and test questions are automatically generated by the generative AI model (e.g., GPT-3). For example, questions such as "Choose the optimal route when picking up a passenger by taxi" are generated.

[0482] Step 11:

[0483] Server: Returns the generated test questions to the user's device as an HTTP response.

[0484] Input: Generated test question data.

[0485] Output: HTTP response packet.

[0486] Specific operation: The generated test questions are formatted as an HTTP response and sent to the user's device.

[0487] Step 12:

[0488] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions.

[0489] Input: HTTP response packet.

[0490] Output: Screen display data.

[0491] Specific behavior: Analyzes received test questions, displays them on the screen, and provides an interface for users to enter answers.

[0492] Step 13:

[0493] User: The driver enters answers to the test questions and sends them to the server.

[0494] Input: User response data.

[0495] Output: HTTP request packet.

[0496] Specific operation: Converts the user's input answer into HTTP request format and sends it to the server.

[0497] Step 14:

[0498] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, and evaluates the level of proficiency.

[0499] Input: User response data.

[0500] Output: Evaluation result data.

[0501] Specific operation: The answers are passed to the analysis module, which determines whether they are correct or incorrect and generates an evaluation result. Using the generative AI model, it also properly evaluates non-standard answers.

[0502] Step 15:

[0503] Server: Returns the evaluation results to the user device as an HTTP response.

[0504] Input: Evaluation result data.

[0505] Output: HTTP response packet.

[0506] Specific operation: The evaluation results are converted into HTTP response format and sent to the user's terminal.

[0507] Step 16:

[0508] Terminal: The evaluation results are displayed on the user's screen.

[0509] Input: HTTP response packet.

[0510] Output: Screen display data.

[0511] Specific operation: The received evaluation results are displayed on the screen, allowing the user to check their own proficiency level.

[0512] (Application example 2)

[0513] 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."

[0514] Conventional driver training systems did not take into account the emotional state of the driver, resulting in ineffective learning and testing. Furthermore, with factory robots, training and proficiency management could not be carried out efficiently, limiting the improvement of robot performance. This led to robot operation errors and slower work speeds, affecting factory production efficiency.

[0515] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0516] In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending individual follow-up messages based on the proficiency evaluation, and means for analyzing the user's emotional data using an emotion engine and adding additional information to the answers based on the analysis results, thereby enabling efficient education and proficiency management that takes emotional states into account.

[0517] The "Class 2 Driver's License Examination Dataset" is a collection of data that includes the test questions, answers, and related information required to obtain a driver's license.

[0518] "Machine learning" is a technology that allows computers to automatically learn from data and improve their ability to perform specific tasks.

[0519] A "model" is a mathematical or statistical representation for data analysis obtained through machine learning.

[0520] "Users" refers to drivers and factory workers who use this system.

[0521] A "means for responding to questions" is a method or system for generating and replying to questions from users using natural language.

[0522] "Means for automatically generating test questions" refers to a method for automatically creating test questions based on a pre-prepared dataset or algorithm.

[0523] "Means for analyzing test results" refers to a method or system for analyzing the percentage of correct answers to test questions answered by users and their trends.

[0524] A "proficiency assessment tool" is a method for assessing a user's skill or knowledge level based on test results.

[0525] The "means for generating and sending follow-up messages" is a method for creating and sending additional guidance and encouraging messages based on the user's proficiency assessment.

[0526] An "emotion engine" is a technology that analyzes a user's emotional data and evaluates their emotional state.

[0527] "Emotion data" refers to data that indicates the emotional state of a user obtained from facial expressions, voice, and other biological signals.

[0528] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The method for realizing this system is described in detail below.

[0529] 1. Program Generation

[0530] The program in this system basically works as follows:

[0531] Train a machine learning model using a dataset from the Class 2 license examination.

[0532] The trained model is used to generate answers to questions posed by users.

[0533] Test questions are automatically generated using a generative AI model and provided to users.

[0534] The results of the test answered by the user are analyzed to evaluate their level of proficiency.

[0535] The emotion engine is used to analyze users' emotional data, and additional information is added to replies and follow-up messages based on the analysis results.

[0536] 2. System Configuration

[0537] Hardware

[0538] Server: A central location that receives, analyzes, and generates data. It is used to train and run models and analyze sentiment data.

[0539] Device (smartphone, tablet, etc.): A device operated by a user, used to enter questions, take tests, and receive feedback.

[0540] Factory robots: Devices that operate in factories and undergo learning and testing.

[0541] software

[0542] Natural Language Processing (NLP) models are used to analyze user questions.

[0543] Generative AI models (e.g., GPT-3): Used to generate answers to questions and automatically generate test questions.

[0544] Emotion engine: Analyzes the user's facial expressions and voice data to assess their emotional state.

[0545] HTTP request / response: A protocol for sending and receiving data between a server and a device.

[0546] 3. Specific examples of program processing

[0547] Question-answering system

[0548] User: Ask the robot, "What's the next step?"

[0549] Terminal: Sends the entered question to the server.

[0550] Server: Analyzes the question using an NLP model and generates an answer using a generative AI model. An answer such as "The next task is to connect part A to part B" is generated.

[0551] Server: The sentiment engine analyzes and adds additional information to the answer indicating that the question is easy.

[0552] Terminal: Display the generated answer to the user.

[0553] Test Generation and Evaluation System

[0554] User: The robot selects "I want to take the test."

[0555] Server: The generative AI model generates test questions such as "Select the steps to correctly connect the parts."

[0556] Terminal: Displays test questions on the user's screen and receives answers.

[0557] Server: Analyzes the answers and assesses proficiency.

[0558] Server: Based on the evaluation results, a follow-up message such as "The next step is this procedure" is generated, and if the emotion engine indicates impatience in its analysis, additional information such as "Don't rush, take your time to check it out" is added.

[0559] Device: View follow messages.

[0560] 4. Examples of prompts

[0561] Specifically, the following prompts are used:

[0562] "Question: What's the next step?\nEmotion: Relaxed\nAnswer:"

[0563] "Question: Please explain the steps to connect parts correctly.\nEmotion: stressed\nAnswer: First, prepare part A, then attach it to part B. Take your time and check carefully."

[0564] In this way, by combining a generative AI model with an emotion engine, this system can provide individualized training that takes into account the emotional state of factory robots and drivers in their learning and work, preventing operational errors and slowdowns in work speeds, and improving production efficiency.

[0565] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0566] Step 1:

[0567] The user asks the robot a question.

[0568] Specific operation: The user inputs "Tell me the next step" into the factory robot's interface.

[0569] Input: The question text from the user.

[0570] Output: The question text is typed into the terminal.

[0571] Step 2:

[0572] The terminal sends a question to the server.

[0573] Specific operation: The terminal sends the question text to the server as an HTTP request.

[0574] Input: Question text.

[0575] Output: The question text is sent to the server.

[0576] Step 3:

[0577] The server receives and parses the query.

[0578] How it works: The server receives the question text and analyzes it using a natural language processing (NLP) model to extract meaning.

[0579] Input: The submitted question text.

[0580] Output: Analysis of the question (text meaning).

[0581] Step 4:

[0582] The server uses an emotion engine to analyze the user's emotion data.

[0583] Specific operation: The server analyzes emotional data (facial expressions and voice data) obtained from the robot's sensors using an emotion engine.

[0584] Input: User emotion data.

[0585] Output: Emotional state analysis (e.g., relaxed, stressed).

[0586] Step 5:

[0587] The server generates an answer using a generative AI model.

[0588] Specific operation: The server generates an answer using a generative AI model (e.g., GPT-3) based on the analysis results of the NLP model and the emotion engine.

[0589] Input: Question content analysis results and emotional state analysis results.

[0590] Output: The generated answer (e.g., "Your next task is to connect part A to part B. Don't worry, it's easy.").

[0591] Step 6:

[0592] The server sends the generated response to the terminal.

[0593] Specific operation: The server sends the generated answer to the terminal as an HTTP response.

[0594] Input: The generated answer.

[0595] Output: The answer is sent to the terminal.

[0596] Step 7:

[0597] The device displays the answer to the user.

[0598] Specific operation: The answer received by the terminal is displayed on the screen and provided to the user.

[0599] Input: The answer sent by the server.

[0600] Output: The user confirms the answer.

[0601] Step 8:

[0602] The user requests the robot to take the test.

[0603] Specific operation: The user inputs "I would like to take a test" through the factory robot's interface.

[0604] Input: A request to take a test.

[0605] Output: The request is typed into the terminal.

[0606] Step 9:

[0607] The device sends a test request to the server.

[0608] Specific operation: The device sends a test-taking request to the server as an HTTP request.

[0609] Input: A test request.

[0610] Output: The request is sent to the server.

[0611] Step 10:

[0612] The server generates test questions using a generative AI model.

[0613] How it works: The server uses a generative AI model (e.g., GPT-3) to automatically generate appropriate test questions.

[0614] Input: A test request.

[0615] Output: The generated test questions.

[0616] Step 11:

[0617] The server sends the generated test questions to the terminal.

[0618] Specific operation: The server sends the generated test questions to the terminal as an HTTP response.

[0619] Input: Generated test questions.

[0620] Output: The test questions are sent to the terminal.

[0621] Step 12:

[0622] The terminal displays the test questions to the user.

[0623] Specific operation: The device displays the received test questions on the screen and allows the user to enter answers.

[0624] Input: The test questions sent from the server.

[0625] Output: The user checks the test questions and enters their answers.

[0626] Step 13:

[0627] The user answers the test questions.

[0628] Specific operation: The user enters answers to the test questions and sends them to the server via the terminal.

[0629] Input: User's answer.

[0630] Output: The answer is entered into the terminal and sent to the server.

[0631] Step 14:

[0632] The server analyzes the answers and assesses proficiency.

[0633] Specific operation: The server analyzes the received answers, evaluates their appropriateness using a generative AI model, and calculates the user's proficiency level.

[0634] Input: User's answer.

[0635] Output: Proficiency assessment results.

[0636] Step 15:

[0637] The server generates and sends follow-up messages based on the proficiency assessment.

[0638] Specific operation: The server generates a follow-up message based on the proficiency assessment results, adds necessary additional information based on the analysis results of the emotion engine, and sends it to the terminal.

[0639] Input: Proficiency assessment results, emotion engine analysis results.

[0640] Output: The generated follow message.

[0641] Step 16:

[0642] The device displays the follow message to the user.

[0643] Specific operation: The follow message received by the device is displayed on the screen and provided to the user.

[0644] Input: The follow message sent from the server.

[0645] Output: The user checks the follow message.

[0646] 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.

[0647] 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.

[0648] 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.

[0649] [Second embodiment]

[0650] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0651] 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.

[0652] 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).

[0653] 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.

[0654] 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.

[0655] 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).

[0656] 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.

[0657] 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.

[0658] 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.

[0659] 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.

[0660] In the smart glasses 214, the 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.

[0661] 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."

[0662] This invention relates to a driver training system that uses a generative AI model on a communication platform operated by a taxi company. Specifically, the system performs machine learning using a dataset from a Class 2 driver's license test, and uses the resulting generative AI model to perform chatbots that respond to drivers' questions, automatic test generation, and proficiency analysis. Below, we provide a natural language explanation of the system's program processing and a concrete example.

[0663] Question and Answer System

[0664] 1. Enter and submit your question

[0665] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0666] Terminal: Sends the entered question to the server as an HTTP request.

[0667] 2. Receiving and analyzing questions

[0668] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0669] 3. Generate and return the answer

[0670] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[0671] Server: The generated answer is sent back to the user's device as an HTTP response.

[0672] 4. View Answers

[0673] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0674] Test Generation and Examination Systems

[0675] 1. Submit a test request

[0676] User: The driver selects "I want to take the test" in the learning application.

[0677] Device: Sends a test request to the server as an HTTP request.

[0678] 2. Generate and submit test questions

[0679] Server: Based on the received request, the test generation module automatically generates test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0680] Server: Returns the generated test questions to the user's device as an HTTP response.

[0681] 3. Display the test screen and submit your answers

[0682] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[0683] 4. Analyzing responses and sending results

[0684] Server: The answer analysis module analyzes the received answers and judges whether they are correct or incorrect. It also calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[0685] Terminal: The received results are displayed on the user's screen, allowing the driver to check their self-evaluation results.

[0686] Follow-up message system

[0687] 1. Proficiency Assessment

[0688] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0689] 2. Create and send a follow message

[0690] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0691] Server: Sends the generated follow message to the user's device as an HTTP response.

[0692] 3. Displaying follow messages

[0693] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0694] As described above, this system utilizes a communication platform operated by a taxi company to achieve effective and efficient driver training using a generative AI model. By linking each module, it is possible to provide learning support tailored to each individual's level of proficiency, significantly reducing the burden of training.

[0695] The processing flow will be explained below.

[0696] Question and Answer System

[0697] Step 1:

[0698] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[0699] Step 2:

[0700] Terminal: Sends the entered question to the server as an HTTP request.

[0701] Step 3:

[0702] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0703] Step 4:

[0704] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[0705] Step 5:

[0706] Server: The generated answer is sent back to the user's device as an HTTP response.

[0707] Step 6:

[0708] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0709] Test Generation and Examination Systems

[0710] Step 1:

[0711] User: Selects "I want to take a test" in a learning app.

[0712] Step 2:

[0713] Device: Sends a test request to the server as an HTTP request.

[0714] Step 3:

[0715] Server: Based on the received request, the test generation module automatically generates test questions, using a generative AI model to generate questions such as "Choose the best route when picking up a passenger using a taxi."

[0716] Step 4:

[0717] Server: Returns the generated test questions to the user's device as an HTTP response.

[0718] Step 5:

[0719] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[0720] Step 6:

[0721] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[0722] Step 7:

[0723] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[0724] Step 8:

[0725] Server: Returns the evaluation results to the user device as an HTTP response.

[0726] Step 9:

[0727] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[0728] Follow-up message system

[0729] Step 1:

[0730] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0731] Step 2:

[0732] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0733] Step 3:

[0734] Server: Sends the generated follow message to the user's device as an HTTP response.

[0735] Step 4:

[0736] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0737] Example 1

[0738] 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."

[0739] Conventional driver training systems have had issues with the difficulty of providing effective training tailored to individual proficiency levels, resulting in low overall training efficiency. Furthermore, question-answering and test question generation are often done manually, which requires time and effort. Furthermore, there are also issues with insufficient evaluation of driver proficiency, resulting in a lack of appropriate feedback and follow-up.

[0740] 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.

[0741] In this invention, the server includes a device that performs machine learning using a dataset of the Class 2 driver's license test, a device that responds to questions from the driver using a model obtained by the machine learning, a device that automatically generates test questions using the model, a device that analyzes the test results answered by the driver and evaluates the driver's proficiency, a device that generates and sends an individual follow-up message based on the proficiency evaluation, a device that receives the driver's input and sends it to the server as an HTTP request, a device that analyzes the content of the questions using natural language processing and generates answers using a generative AI model, and a terminal that receives and displays the generated answers to show to the driver. This makes it possible to quickly answer the driver's questions, automatically generate test questions according to the individual proficiency level, and provide appropriate evaluations and individual follow-ups.

[0742] The "Type 2 license test" is a driver's license test required to drive passenger vehicles such as taxis and buses.

[0743] A "dataset" is a collection of data collected for a specific purpose and used in machine learning and data analysis.

[0744] "Machine learning" is a technology that allows computers to learn patterns and knowledge from data and perform specific tasks automatically.

[0745] A "model" is an algorithm or function generated by machine learning that makes predictions or classifications on new data.

[0746] "Driver" refers to a professional driver who drives a vehicle such as a taxi or bus.

[0747] A "device that responds to questions" is a system that receives questions from the driver and generates and provides appropriate answers.

[0748] "Test questions" are test questions given to assess a driver's knowledge and skills.

[0749] "Analysis" is the process of examining data or information in detail to find meaning and patterns.

[0750] "Proficiency" is an indicator of how much knowledge and skill a driver has in a particular field.

[0751] A "follow-up message" is a customized message based on the proficiency assessment to encourage additional learning and improvement for the driver.

[0752] An "HTTP request" is a data request sent from a client to a server.

[0753] "Natural language processing" is a technology that allows computers to understand human language, and includes morphological analysis and semantic analysis.

[0754] A "generative AI model" is an artificial intelligence model that has been trained to perform tasks such as text generation and question answering.

[0755] A "terminal" is an electronic device that allows a user to interact with a system through an interface.

[0756] MODE FOR CARRYING OUT THE INVENTION

[0757] This invention relates to a driver training system based on machine learning techniques, specifically using a dataset from the Class 2 driver's license test. The system uses a generative AI model to respond to driver questions, automatically generate test questions, analyze test results, and provide personalized follow-up messages based on the driver's proficiency assessment.

[0758] System Configuration

[0759] 1. Hardware and Software Overview

[0760] Device: An electronic device (e.g., smartphone, tablet, computer) equipped with a learning application used by a driver.

[0761] Server: A central system for analyzing questions, generating test questions, analyzing answers, and generating follow-up messages.

[0762] Natural Language Processing (NLP) libraries, such as spaCy.

[0763] Generative AI models: Use large language models such as GPT-3.

[0764] Database: Stores past questions from the Class 2 license exam and driving knowledge materials.

[0765] System Operation

[0766] 1. Question and Answer System

[0767] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0768] Terminal: Send this input to the server as an HTTP request.

[0769] Server: Passes the received question to the chatbot module, and performs morphological analysis using a natural language processing library (e.g., spaCy) to analyze keywords and sentence structure.

[0770] Server: Enter the prompt "How often do I need to renew my driver's license?" into a generative AI model (e.g., GPT-3) and generate an appropriate answer. The generated text is "A driver's license generally needs to be renewed every five years."

[0771] Server: Sends the generated answer to the terminal as an HTTP response.

[0772] Terminal: The received answer is displayed on the user interface, and the driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[0773] 2. Test Generation and Examination Systems

[0774] User: The driver selects "I want to take the test" in the learning application.

[0775] Device: Sends a test request to the server as an HTTP request.

[0776] Server: Based on the received request, the test generation module references the database and generates new test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0777] Server: Sends the generated test questions to the device as an HTTP response.

[0778] Terminal: The test questions are displayed in a user interface for the driver to answer. The driver enters the answers and sends them to the server as HTTP requests.

[0779] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, calculates the driver's proficiency, generates an evaluation result, and sends it to the user's device as an HTTP response.

[0780] Terminal: The received evaluation results are displayed on the user interface, allowing the driver to check their self-evaluation results.

[0781] 3. Follow-up message system

[0782] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects areas of low proficiency (e.g., road traffic laws).

[0783] Server: Generate a personalized follow-up message for each subject area that the follow-up module detects. For example, create a message that says, "Review your basic knowledge of road traffic laws. Additional learning materials are available here."

[0784] Server: Sends the generated follow message to the user's device as an HTTP response.

[0785] Terminal: The received follow-up message is displayed on the user interface, and the driver can receive specific learning instructions.

[0786] Examples of concrete examples and prompts

[0787] Example 1: A driver asks, "How often do I need to renew my driver's license?" and the server uses an AI model to generate and provide the answer, "Basically, a driver's license needs to be renewed every five years."

[0788] Example prompt sentence:

[0789] Question-answering system prompt: "How often do I need to renew my driver's license?"

[0790] Test generation system prompt: "Generate test questions based on the Class 2 taxi driver's license exam."

[0791] Follow-up message system prompt: "This driver failed multiple questions about road traffic laws. Please generate a follow-up message to help them study again."

[0792] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0793] Processing flow

[0794] Question and Answer System

[0795] Step 1: Enter and submit your question

[0796] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0797] Input: The driver's text input.

[0798] Terminal: Sends the entered question to the server as an HTTP request.

[0799] Output: HTTP request to the server.

[0800] Step 2: Receiving and parsing the question

[0801] Server: Receives the HTTP request and retrieves the inquiry.

[0802] Input: The HTTP request received from the device.

[0803] Server: Using an NLP library (e.g., spaCy), morphologically analyze the question and extract keywords and sentence structure.

[0804] Output: Parsed keywords and sentence structure.

[0805] Step 3: Generate an answer

[0806] Server: The analyzed question content is input as a prompt to the generative AI model (e.g., GPT-3).

[0807] Input: Parsed keywords and sentence structure.

[0808] Server: Generates an appropriate answer using the AI ​​model. It generates the text, "Driver's licenses generally need to be renewed every five years."

[0809] Output: The generated answer text.

[0810] Step 4: Return and view your responses

[0811] Server: Sends the generated answer to the user's device as an HTTP response.

[0812] Input: The generated answer text.

[0813] Output: HTTP response to the user's device.

[0814] Device: The received answer is displayed on the screen. The driver can confirm the answer: "Driver's licenses generally need to be renewed every five years."

[0815] Test Generation and Examination Systems

[0816] Step 1: Submit a test request

[0817] User: The driver selects "I want to take the test" in the learning application.

[0818] Input: Driver test request.

[0819] Terminal: Send this request to the server as an HTTP request.

[0820] Output: HTTP request to the server.

[0821] Step 2: Generate test questions

[0822] Server: Receives requests and passes instructions to the test generation module.

[0823] Input: Exam request.

[0824] Server: Refers to the database and automatically generates new test questions. For example, it generates a question such as "Choose the best route when picking up a passenger by taxi."

[0825] Output: The generated test questions.

[0826] Step 3: Submit and view the test questions

[0827] Server: Sends the generated test questions to the user's device as an HTTP response.

[0828] Input: The generated test questions.

[0829] Output: HTTP response to the user's device.

[0830] Terminal: Displays the received test questions on the screen so that the driver can answer them.

[0831] Input: Show test question.

[0832] Output: Driver answer input.

[0833] Step 4: Submit and analyze responses

[0834] Terminal: After entering the driver's answers, send them to the server as an HTTP request.

[0835] Input: Driver's answer.

[0836] Output: HTTP request to the server.

[0837] Server: The answer analysis module analyzes the received answer data, determines whether the answer is correct, and calculates the driver's proficiency level.

[0838] Input: Received response data.

[0839] Output: Evaluation result.

[0840] Step 5: Send and view results

[0841] Server: Sends the evaluation results to the user device as an HTTP response.

[0842] Input: Evaluation result.

[0843] Output: HTTP response to the user's device.

[0844] Terminal: The received evaluation results are displayed on the user's screen. Drivers can check their own evaluation results.

[0845] Input: Display evaluation results.

[0846] Output: Driver verification.

[0847] Follow-up message system

[0848] Step 1: Assess proficiency

[0849] Server: The response analysis module evaluates the driver's proficiency based on past test results.

[0850] Input: Past test results.

[0851] Server: Detect low proficiency in a specific subject (e.g., road traffic law).

[0852] Output: Low proficiency area detection results.

[0853] Step 2: Generate a follow-up message

[0854] Server: The follow-up module generates individual follow-up messages for low-proficiency areas.

[0855] Input: Information in low proficiency areas.

[0856] Server: Generate a message such as "Review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[0857] Output: The generated follow message.

[0858] Step 3: Send and view follow-up messages

[0859] Server: Sends the generated follow message to the user's device as an HTTP response.

[0860] Input: The generated follow message.

[0861] Output: HTTP response to the user's device.

[0862] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[0863] Input: Show follow messages.

[0864] Output: Driver verification.

[0865] (Application example 1)

[0866] 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."

[0867] Current training for robot operators in factories relies on instruction by human trainers and static teaching materials, making it difficult to provide efficient training tailored to each operator's proficiency level. It is also difficult to respond to operator questions and provide appropriate teaching materials in real time, resulting in limited training efficiency and effectiveness. This invention aims to solve these issues by using a generative AI model to provide a flexible and efficient training system tailored to each operator's proficiency level.

[0868] 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.

[0869] In this invention, the server includes a means for performing machine learning using a dataset of the Class 2 license examination, a means for responding to questions from operators using a model obtained by the machine learning, and a means for automatically generating test questions using the model. This not only enables quick and accurate responses to questions from operators, but also makes it possible to provide test questions tailored to the learning status of each individual operator and appropriately generate and send follow-up messages based on their proficiency.

[0870] The "Class 2 license examination dataset" refers to a wide range of test questions and answer data used to evaluate an operator's skills and knowledge.

[0871] "Machine learning" refers to the technology in which computers automatically learn using large amounts of data to generate algorithms and models for performing specific tasks.

[0872] A "model" refers to a data processing structure built using machine learning to predict or classify a specific outcome based on input data.

[0873] "Operator" refers to an employee who operates and manages robots and machines within a factory.

[0874] "Test Questions" refers to automatically generated test questions to evaluate an operator's knowledge and skills.

[0875] "Proficiency" refers to an indicator that evaluates how much an operator has acquired a particular skill or knowledge.

[0876] "Follow-up messages" refer to messages containing instructions or advice sent individually to encourage further learning and improvement based on the results of an operator's proficiency assessment.

[0877] "System" refers to the overall infrastructure and process for supporting training using machine learning models implemented according to this invention.

[0878] This invention is a system for supporting the training of operators of robots used in factories. It performs machine learning using a dataset from the Class 2 license examination, and uses the model to respond to the operator's questions, automatically generate tests, evaluate proficiency, and send follow-up messages.

[0879] Overall system overview

[0880] This system consists of three main functions:

[0881] 1. Question and Answer System

[0882] 2. Test Generation and Examination Systems

[0883] 3. Proficiency-based follow-up message system

[0884] Question and Answer System

[0885] The server receives a question from the operator and analyzes it using the chatbot module. Specifically, it uses natural language processing (NLP) technology to morphologically analyze the content of the question and extract its meaning. It then uses a generative AI model (e.g., GPT-3) to generate an appropriate answer and sends it back to the operator's device. The operator can then view the received answer on their device's screen.

[0886] Examples:

[0887] When an operator uses a smartphone to input a question such as, "Where is the emergency stop button on the robot?", the server receives the question, analyzes it, and generates and returns the answer, "The emergency stop button is on the left side of the robot's control panel."

[0888] Example prompt sentence:

[0889] I have a question about the emergency stop button on the robot. Can you tell me where it is?

[0890] Test Generation and Examination Systems

[0891] When the server receives a test request from an operator, it automatically generates test questions using a test generation module. The generated test questions are sent to the operator's terminal and displayed. The operator answers the questions and sends the answers to the server. The server analyzes the answers, determines whether they are correct or incorrect, evaluates the operator's proficiency, and returns the results.

[0892] Examples:

[0893] When an operator requests a test on their smartphone, the server generates a question such as "What is the normal maintenance cycle for a robot?" and sends it to the operator's device. When the operator enters and submits the answer, the answer is analyzed and the results and evaluation are sent back to the device.

[0894] Proficiency-based follow-up message system

[0895] The server evaluates the operator's proficiency based on past test results. Based on the evaluation results, follow-up messages are individually generated for specific areas and sent to the operator's terminal, allowing the operator to conduct additional learning in areas where their understanding is lacking.

[0896] Examples:

[0897] If the server detects low proficiency in "robot safety operation procedures," it generates a follow-up message such as "Please relearn the specific area. Please refer to the manual in the system for details," and sends it to the operator's terminal.

[0898] Example prompt sentence:

[0899] Your proficiency has been verified. You will need to refresh your knowledge in the following areas: Robot safety operating procedures.

[0900] Technologies and Tools Used

[0901] Hardware: Server, user device (smartphone)

[0902] Software: Natural Language Processing (NLP) module, generative AI model (e.g. GPT-3), HTTP communication protocol

[0903] The implementation of this system will not only enable effective and efficient training tailored to each operator's individual level of proficiency, but will also enable responses to questions in real time and the provision of appropriate learning instructions.

[0904] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0905] Step 1:

[0906] The user opens the learning application on their smartphone and inputs a question, for example, "Where is the emergency stop button on the robot?" This question is captured as input data on the device.

[0907] Step 2:

[0908] The device sends the entered question to the server as an HTTP request, which is in JSON format and contains the question itself.

[0909] Step 3:

[0910] The server passes the received question to the chatbot module, which uses natural language processing (NLP) to morphologically analyze the question text and extract the meaning of the question. Through this process, the question content is interpreted.

[0911] Step 4:

[0912] The server inputs the interpreted question into a generative AI model (e.g., GPT-3) and generates an appropriate answer. The generated answer might be something like, "There is an emergency stop button on the left side of the robot's control panel."

[0913] Step 5:

[0914] The server returns the generated answer to the user device as an HTTP response, which is also in JSON format and contains the answer text.

[0915] Step 6:

[0916] The device displays the received response on the user's screen, and the user can see the response on their smartphone screen: "There is an emergency stop button on the left side of the robot's control panel."

[0917] Step 7:

[0918] A user selects "I want to take a test" in a learning application. This request is received as input data on the device and sent to the server as an HTTP request.

[0919] Step 8:

[0920] The server automatically generates test questions using a test generation module based on the received test request. For example, a question such as "What is the normal maintenance cycle for a robot?" is generated.

[0921] Step 9:

[0922] The server sends the generated test questions to the user's terminal as an HTTP response. The terminal displays the received test questions on the user's screen, and the user answers the questions.

[0923] Step 10:

[0924] The user inputs the answer to the question, and the terminal sends it to the server as an HTTP request. This request contains data that includes the answer.

[0925] Step 11:

[0926] The server analyzes the received answers and judges whether they are correct or not. The answer analysis module then calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[0927] Step 12:

[0928] The device displays the received evaluation results on the user's screen, allowing the user to check their self-evaluation results on their smartphone screen.

[0929] Step 13:

[0930] The server evaluates the user's proficiency based on past test results and detects low proficiency in a specific area (e.g., robot safety operating procedures).

[0931] Step 14:

[0932] The server generates a follow-up message for the area of ​​low proficiency and sends it to the user's device as an HTTP response. For example, a follow-up message such as "Please restudy the specific area. For details, please refer to the system manual" is generated.

[0933] Step 15:

[0934] The device displays the received follow message on the user's screen, allowing the user to receive specific study instructions on the smartphone screen.

[0935] 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.

[0936] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The specific system configuration combines a machine learning model using a dataset from the Class 2 driver's license exam with an emotion engine to achieve an educational effect that takes into account the driver's emotional state.

[0937] Question and Answer System

[0938] 1. Enter and submit your question

[0939] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[0940] Terminal: Sends the entered question to the server as an HTTP request.

[0941] 2. Receiving and analyzing questions

[0942] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0943] 3. Emotion Recognition and Answer Generation

[0944] Server: The emotion engine receives emotion data (e.g., facial recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[0945] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[0946] 4. Returning and displaying answers

[0947] Server: The generated answer is sent back to the user's device as an HTTP response.

[0948] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[0949] Test Generation and Examination Systems

[0950] 1. Submit a test request

[0951] User: The driver selects "I want to take the test" in the learning application.

[0952] Device: Sends a test request to the server as an HTTP request.

[0953] 2. Generate and submit test questions

[0954] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the best route when picking up a passenger using a taxi."

[0955] Server: Returns the generated test questions to the user's device as an HTTP response.

[0956] 3. Display the test screen and submit your answers

[0957] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[0958] 4. Analyzing responses and sending results

[0959] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[0960] Server: Returns the evaluation results to the user device as an HTTP response.

[0961] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[0962] Follow-up message system

[0963] 1. Proficiency Assessment

[0964] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[0965] 2. Create and send a follow message

[0966] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[0967] 3. Displaying follow messages

[0968] Server: Sends the generated follow message to the user's device as an HTTP response.

[0969] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[0970] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[0971] The processing flow will be explained below.

[0972] Question and Answer System

[0973] Step 1:

[0974] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[0975] Step 2:

[0976] Terminal: Sends the entered question to the server as an HTTP request.

[0977] Step 3:

[0978] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[0979] Step 4:

[0980] Server: The emotion engine receives emotion data (e.g., facial expression recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[0981] Step 5:

[0982] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[0983] Step 6:

[0984] Server: The generated answer is sent back to the user's device as an HTTP response.

[0985] Step 7:

[0986] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[0987] Test Generation and Examination Systems

[0988] Step 1:

[0989] User: Selects "I want to take a test" in a learning app.

[0990] Step 2:

[0991] Device: Sends a test request to the server as an HTTP request.

[0992] Step 3:

[0993] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the optimal route when picking up a passenger using a taxi."

[0994] Step 4:

[0995] Server: Returns the generated test questions to the user's device as an HTTP response.

[0996] Step 5:

[0997] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[0998] Step 6:

[0999] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[1000] Step 7:

[1001] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[1002] Step 8:

[1003] Server: Returns the evaluation results to the user device as an HTTP response.

[1004] Step 9:

[1005] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[1006] Follow-up message system

[1007] Step 1:

[1008] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1009] Step 2:

[1010] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[1011] Step 3:

[1012] Server: Sends the generated follow message to the user's device as an HTTP response.

[1013] Step 4:

[1014] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[1015] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[1016] Example 2

[1017] 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."

[1018] Conventional driver training systems often provide uniform training without considering the individual needs or emotional state of each learner. Furthermore, detailed feedback based on test results is not provided, potentially reducing learning effectiveness. Furthermore, training utilizing emotional data has not been implemented, making it impossible to reduce learner stress. Therefore, a system that provides personalized training while taking into account the learner's emotional state was needed.

[1019] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending an individual follow-up message based on the proficiency evaluation, and means for analyzing emotion data from the user terminal and adding additional information based on the emotion to the answers generated by the model. This enables optimal education that takes into account the user's emotional state, thereby improving the effectiveness of the education.

[1020] "Machine learning" is a technology that allows a computer to analyze a given data set and automatically learn patterns and rules based on that data.

[1021] The "Class 2 Driver's License Examination Dataset" is a large-scale data set that includes questions and answers related to the Class 2 Driver's License Examination, as well as their trends and characteristics.

[1022] A "generative AI model" is a model that uses artificial intelligence to generate content such as text and images, and specifically includes natural language processing and generation modules.

[1023] A "chatbot" is a program that can interact with users and respond to their questions via text or voice.

[1024] "Emotion data" is data that indicates the user's emotional state, and examples include facial expression data and voice tone.

[1025] An "emotion engine" is an algorithm or system that analyzes emotional data to detect and evaluate a user's emotional state.

[1026] "Proficiency" is a measure of how well a driver understands or has mastered a particular subject or task.

[1027] A "follow-up message" is a message such as instruction or encouragement that is generated based on the user's learning progress or emotional state.

[1028] An "HTTP request" is a request for data sent by a client to a server, and is a protocol used by web browsers and applications to request data from a server.

[1029] An "HTTP response" is a data response that a server returns in response to a client's HTTP request, and includes requested data and status information.

[1030] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. Specific embodiments of the system are as follows.

[1031] Hardware and software used

[1032] This system uses the following hardware and software:

[1033] Server: A computer that performs major processing. It is equipped with a high-performance processor (e.g., Intel Xeon) and a large amount of memory (e.g., 64 GB or more).

[1034] Device: The device used by the driver, such as a smartphone or tablet (e.g., Android or iOS).

[1035] Generative AI model: An artificial intelligence model that generates text, for example using an advanced natural language processing model such as GPT-3.

[1036] Emotion engine: Software for analyzing user emotion data. Examples include Azure Cognitive Services' emotion recognition API.

[1037] NLP engine: Natural language processing software to analyze the question text, for example using spaCy or NLTK.

[1038] System processing flow

[1039] Answers to questions

[1040] A user opens a learning application and inputs a question, for example, "How often do I need to renew my driver's license?" The device sends this question to the server as an HTTP request. The server passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text. The server then uses an emotion engine to analyze emotion data (e.g., facial recognition and voice analysis data) from the user's device and determine the user's emotional state.

[1041] A generative AI model (e.g., GPT-3) is used to generate the optimal answer based on the analyzed question content and data from the emotion engine. For example, a standard answer such as "Your driver's license generally needs to be renewed every five years" is generated. If the user expresses anxiety, the server adds additional information such as "Don't worry, the process is simple."

[1042] The generated answer is then returned to the user terminal as an HTTP response, and the terminal displays the received answer on the user screen.

[1043] Test Generation and Taking

[1044] When a user selects "I want to take a test" in the learning application, the device sends a test request to the server. Based on the received request, the server uses a generative AI model to automatically generate test questions. For example, a question such as "Select the optimal route when picking up a passenger by taxi" is generated.

[1045] The generated test questions are sent back to the user's device, where the user can check them on the screen and enter their answers. The device then sends these answers to the server, which uses an answer analysis module to determine whether the answers are correct or incorrect and evaluate the user's proficiency. The results are then sent back to the user's device and displayed.

[1046] Follow message generation

[1047] The server evaluates the user's proficiency based on past test results and detects low proficiency in specific areas. It then uses an emotion engine to analyze the user's emotional state and generates appropriate follow-up messages based on that. For example, it generates specific learning instructions such as, "Please review your basic knowledge of road traffic laws. Take your time and review them carefully."

[1048] The follow message thus generated is sent to the user's terminal, and the user can check it on the screen.

[1049] Examples of concrete examples and prompts

[1050] For example, if a user asks, "How often do I need to renew my driver's license?", the following prompt is sent to the generative AI model:

[1051] User: "How many years do I need to renew my driver's license?"

[1052] Generative AI model: "Driver's licenses generally need to be renewed every five years."

[1053] Add emotion: "Don't worry, it's a simple process."

[1054] In this way, by combining a generative AI model with an emotion engine, this system is able to provide optimal training that takes into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a learning environment that is easier for drivers to understand and less stressful.

[1055] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1056] Step 1:

[1057] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1058] Input: Keyboard input data.

[1059] Output: Question text data.

[1060] What happens: The user types a question into a text input field in the application and presses the Enter key to confirm the question.

[1061] Step 2:

[1062] Terminal: Sends the entered question to the server as an HTTP request.

[1063] Input: Question text data.

[1064] Output: HTTP request packet.

[1065] Specific operation: Converts text data into HTTP packets and sends them over the network to the server.

[1066] Step 3:

[1067] Server: Passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text.

[1068] Input: HTTP request packet, text data.

[1069] Output: Parsed question semantic data.

[1070] What it does: The server decodes the HTTP packets, passes the text data to an NLP engine (e.g., spaCy or NLTK), and performs morphological analysis to extract the meaning of the question.

[1071] Step 4:

[1072] Server: Uses the emotion engine to receive and analyze emotion data (e.g., facial recognition, voice analysis data) from the user device.

[1073] Input: Emotion data, e.g., image or audio files.

[1074] Output: User's emotional state data.

[1075] Specific operation: Emotion data is passed to an emotion engine (e.g., Azure Cognitive Services emotion recognition API) for analysis, and the user's emotional state is output as numerical data.

[1076] Step 5:

[1077] Server: Using a generative AI model, it generates answers based on the analyzed question content and sentiment data.

[1078] Input: Question semantic data, emotional state data.

[1079] Output: Answer data in text format.

[1080] What it does: The semantic data and emotional state data of the question are incorporated into a prompt sentence, which is then sent to a generative AI model (e.g., GPT-3) to generate an answer, such as, "Your driver's license generally needs to be renewed every five years. Don't worry, it's a simple process."

[1081] Step 6:

[1082] Server: The generated answer is sent back to the user's device as an HTTP response.

[1083] Input: Textual response data.

[1084] Output: HTTP response packet.

[1085] Specific operation: The response data is formatted as an HTTP response and sent to the user terminal via the network.

[1086] Step 7:

[1087] Terminal: Displays the received answer on the user's screen.

[1088] Input: HTTP response packet.

[1089] Output: Screen display data.

[1090] What it does: It extracts text data from the HTTP response and displays it in a display field in the application. The user can see the answer on the screen: "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[1091] Step 8:

[1092] User: The driver selects "I want to take the test" in the learning application.

[1093] Input: A test-taking request.

[1094] Output: Test attempt selection data.

[1095] Specific behavior: A user clicks the "I want to take a test" button in the application to generate a test request.

[1096] Step 9:

[1097] Device: Sends a test request to the server as an HTTP request.

[1098] Input: Test attempt selection data.

[1099] Output: HTTP request packet.

[1100] Specific operation: The test selection data is converted into an HTTP request format and sent to the server via the network.

[1101] Step 10:

[1102] Server: Based on the received request, the server automatically generates test questions using a generative AI model.

[1103] Input: HTTP request packet.

[1104] Output: The generated test question data.

[1105] Specific operation: The request content is converted into a prompt for the generative AI model, and test questions are automatically generated by the generative AI model (e.g., GPT-3). For example, questions such as "Choose the optimal route when picking up a passenger by taxi" are generated.

[1106] Step 11:

[1107] Server: Returns the generated test questions to the user's device as an HTTP response.

[1108] Input: Generated test question data.

[1109] Output: HTTP response packet.

[1110] Specific operation: The generated test questions are formatted as an HTTP response and sent to the user's device.

[1111] Step 12:

[1112] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions.

[1113] Input: HTTP response packet.

[1114] Output: Screen display data.

[1115] Specific behavior: Analyzes received test questions, displays them on the screen, and provides an interface for users to enter answers.

[1116] Step 13:

[1117] User: The driver enters answers to the test questions and sends them to the server.

[1118] Input: User response data.

[1119] Output: HTTP request packet.

[1120] Specific operation: Converts the user's input answer into HTTP request format and sends it to the server.

[1121] Step 14:

[1122] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, and evaluates the level of proficiency.

[1123] Input: User response data.

[1124] Output: Evaluation result data.

[1125] Specific operation: The answers are passed to the analysis module, which determines whether they are correct or incorrect and generates an evaluation result. Using the generative AI model, it also properly evaluates non-standard answers.

[1126] Step 15:

[1127] Server: Returns the evaluation results to the user device as an HTTP response.

[1128] Input: Evaluation result data.

[1129] Output: HTTP response packet.

[1130] Specific operation: The evaluation results are converted into HTTP response format and sent to the user's terminal.

[1131] Step 16:

[1132] Terminal: The evaluation results are displayed on the user's screen.

[1133] Input: HTTP response packet.

[1134] Output: Screen display data.

[1135] Specific operation: The received evaluation results are displayed on the screen, allowing the user to check their own proficiency level.

[1136] (Application example 2)

[1137] 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."

[1138] Conventional driver training systems did not take into account the emotional state of the driver, resulting in ineffective learning and testing. Furthermore, with factory robots, training and proficiency management could not be carried out efficiently, limiting the improvement of robot performance. This led to robot operation errors and slower work speeds, affecting factory production efficiency.

[1139] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1140] In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending individual follow-up messages based on the proficiency evaluation, and means for analyzing the user's emotional data using an emotion engine and adding additional information to the answers based on the analysis results, thereby enabling efficient education and proficiency management that takes emotional states into account.

[1141] The "Class 2 Driver's License Examination Dataset" is a collection of data that includes the test questions, answers, and related information required to obtain a driver's license.

[1142] "Machine learning" is a technology that allows computers to automatically learn from data and improve their ability to perform specific tasks.

[1143] A "model" is a mathematical or statistical representation for data analysis obtained through machine learning.

[1144] "Users" refers to drivers and factory workers who use this system.

[1145] A "means for responding to questions" is a method or system for generating and replying to questions from users using natural language.

[1146] "Means for automatically generating test questions" refers to a method for automatically creating test questions based on a pre-prepared dataset or algorithm.

[1147] "Means for analyzing test results" refers to a method or system for analyzing the percentage of correct answers to test questions answered by users and their trends.

[1148] A "proficiency assessment tool" is a method for assessing a user's skill or knowledge level based on test results.

[1149] The "means for generating and sending follow-up messages" is a method for creating and sending additional guidance and encouraging messages based on the user's proficiency assessment.

[1150] An "emotion engine" is a technology that analyzes a user's emotional data and evaluates their emotional state.

[1151] "Emotion data" refers to data that indicates the emotional state of a user obtained from facial expressions, voice, and other biological signals.

[1152] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The method for realizing this system is described in detail below.

[1153] 1. Program Generation

[1154] The program in this system basically works as follows:

[1155] Train a machine learning model using a dataset from the Class 2 license examination.

[1156] The trained model is used to generate answers to questions posed by users.

[1157] Test questions are automatically generated using a generative AI model and provided to users.

[1158] The results of the test answered by the user are analyzed to evaluate their level of proficiency.

[1159] The emotion engine is used to analyze users' emotional data, and additional information is added to replies and follow-up messages based on the analysis results.

[1160] 2. System Configuration

[1161] Hardware

[1162] Server: A central location that receives, analyzes, and generates data. It is used to train and run models and analyze sentiment data.

[1163] Device (smartphone, tablet, etc.): A device operated by a user, used to enter questions, take tests, and receive feedback.

[1164] Factory robots: Devices that operate in factories and undergo learning and testing.

[1165] software

[1166] Natural Language Processing (NLP) models are used to analyze user questions.

[1167] Generative AI models (e.g., GPT-3): Used to generate answers to questions and automatically generate test questions.

[1168] Emotion engine: Analyzes the user's facial expressions and voice data to assess their emotional state.

[1169] HTTP request / response: A protocol for sending and receiving data between a server and a device.

[1170] 3. Specific examples of program processing

[1171] Question-answering system

[1172] User: Ask the robot, "What's the next step?"

[1173] Terminal: Sends the entered question to the server.

[1174] Server: Analyzes the question using an NLP model and generates an answer using a generative AI model. An answer such as "The next task is to connect part A to part B" is generated.

[1175] Server: The sentiment engine analyzes and adds additional information to the answer indicating that the question is easy.

[1176] Terminal: Display the generated answer to the user.

[1177] Test Generation and Evaluation System

[1178] User: The robot selects "I want to take the test."

[1179] Server: The generative AI model generates test questions such as "Select the steps to correctly connect the parts."

[1180] Terminal: Displays test questions on the user's screen and receives answers.

[1181] Server: Analyzes the answers and assesses proficiency.

[1182] Server: Based on the evaluation results, a follow-up message such as "The next step is this procedure" is generated, and if the emotion engine indicates impatience in its analysis, additional information such as "Don't rush, take your time to check it out" is added.

[1183] Device: View follow messages.

[1184] 4. Examples of prompts

[1185] Specifically, the following prompts are used:

[1186] "Question: What's the next step?\nEmotion: Relaxed\nAnswer:"

[1187] "Question: Please explain the steps to connect parts correctly.\nEmotion: stressed\nAnswer: First, prepare part A, then attach it to part B. Take your time and check carefully."

[1188] In this way, by combining a generative AI model with an emotion engine, this system can provide individualized training that takes into account the emotional state of factory robots and drivers in their learning and work, preventing operational errors and slowdowns in work speeds, and improving production efficiency.

[1189] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1190] Step 1:

[1191] The user asks the robot a question.

[1192] Specific operation: The user inputs "Tell me the next step" into the factory robot's interface.

[1193] Input: The question text from the user.

[1194] Output: The question text is typed into the terminal.

[1195] Step 2:

[1196] The terminal sends a question to the server.

[1197] Specific operation: The terminal sends the question text to the server as an HTTP request.

[1198] Input: Question text.

[1199] Output: The question text is sent to the server.

[1200] Step 3:

[1201] The server receives and parses the query.

[1202] How it works: The server receives the question text and analyzes it using a natural language processing (NLP) model to extract meaning.

[1203] Input: The submitted question text.

[1204] Output: Analysis of the question (text meaning).

[1205] Step 4:

[1206] The server uses an emotion engine to analyze the user's emotion data.

[1207] Specific operation: The server analyzes emotional data (facial expressions and voice data) obtained from the robot's sensors using an emotion engine.

[1208] Input: User emotion data.

[1209] Output: Emotional state analysis (e.g., relaxed, stressed).

[1210] Step 5:

[1211] The server generates an answer using a generative AI model.

[1212] Specific operation: The server generates an answer using a generative AI model (e.g., GPT-3) based on the analysis results of the NLP model and the emotion engine.

[1213] Input: Question content analysis results and emotional state analysis results.

[1214] Output: The generated answer (e.g., "Your next task is to connect part A to part B. Don't worry, it's easy.").

[1215] Step 6:

[1216] The server sends the generated response to the terminal.

[1217] Specific operation: The server sends the generated answer to the terminal as an HTTP response.

[1218] Input: The generated answer.

[1219] Output: The answer is sent to the terminal.

[1220] Step 7:

[1221] The device displays the answer to the user.

[1222] Specific operation: The answer received by the terminal is displayed on the screen and provided to the user.

[1223] Input: The answer sent by the server.

[1224] Output: The user confirms the answer.

[1225] Step 8:

[1226] The user requests the robot to take the test.

[1227] Specific operation: The user inputs "I would like to take a test" through the factory robot's interface.

[1228] Input: A request to take a test.

[1229] Output: The request is typed into the terminal.

[1230] Step 9:

[1231] The device sends a test request to the server.

[1232] Specific operation: The device sends a test-taking request to the server as an HTTP request.

[1233] Input: A test request.

[1234] Output: The request is sent to the server.

[1235] Step 10:

[1236] The server generates test questions using a generative AI model.

[1237] How it works: The server uses a generative AI model (e.g., GPT-3) to automatically generate appropriate test questions.

[1238] Input: A test request.

[1239] Output: The generated test questions.

[1240] Step 11:

[1241] The server sends the generated test questions to the terminal.

[1242] Specific operation: The server sends the generated test questions to the terminal as an HTTP response.

[1243] Input: Generated test questions.

[1244] Output: The test questions are sent to the terminal.

[1245] Step 12:

[1246] The terminal displays the test questions to the user.

[1247] Specific operation: The device displays the received test questions on the screen and allows the user to enter answers.

[1248] Input: The test questions sent from the server.

[1249] Output: The user checks the test questions and enters their answers.

[1250] Step 13:

[1251] The user answers the test questions.

[1252] Specific operation: The user enters answers to the test questions and sends them to the server via the terminal.

[1253] Input: User's answer.

[1254] Output: The answer is entered into the terminal and sent to the server.

[1255] Step 14:

[1256] The server analyzes the answers and assesses proficiency.

[1257] Specific operation: The server analyzes the received answers, evaluates their appropriateness using a generative AI model, and calculates the user's proficiency level.

[1258] Input: User's answer.

[1259] Output: Proficiency assessment results.

[1260] Step 15:

[1261] The server generates and sends follow-up messages based on the proficiency assessment.

[1262] Specific operation: The server generates a follow-up message based on the proficiency assessment results, adds necessary additional information based on the analysis results of the emotion engine, and sends it to the terminal.

[1263] Input: Proficiency assessment results, emotion engine analysis results.

[1264] Output: The generated follow message.

[1265] Step 16:

[1266] The device displays the follow message to the user.

[1267] Specific operation: The follow message received by the device is displayed on the screen and provided to the user.

[1268] Input: The follow message sent from the server.

[1269] Output: The user checks the follow message.

[1270] 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.

[1271] 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.

[1272] 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.

[1273] [Third embodiment]

[1274] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[1275] 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.

[1276] 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).

[1277] 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.

[1278] 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.

[1279] 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).

[1280] 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.

[1281] 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.

[1282] 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.

[1283] 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.

[1284] 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.

[1285] 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."

[1286] This invention relates to a driver training system that uses a generative AI model on a communication platform operated by a taxi company. Specifically, the system performs machine learning using a dataset from a Class 2 driver's license test, and uses the resulting generative AI model to perform chatbots that respond to drivers' questions, automatic test generation, and proficiency analysis. Below, we provide a natural language explanation of the system's program processing and a concrete example.

[1287] Question and Answer System

[1288] 1. Enter and submit your question

[1289] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1290] Terminal: Sends the entered question to the server as an HTTP request.

[1291] 2. Receiving and analyzing questions

[1292] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1293] 3. Generate and return the answer

[1294] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[1295] Server: The generated answer is sent back to the user's device as an HTTP response.

[1296] 4. View Answers

[1297] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[1298] Test Generation and Examination Systems

[1299] 1. Submit a test request

[1300] User: The driver selects "I want to take the test" in the learning application.

[1301] Device: Sends a test request to the server as an HTTP request.

[1302] 2. Generate and submit test questions

[1303] Server: Based on the received request, the test generation module automatically generates test questions, such as "Choose the best route when picking up a passenger using a taxi."

[1304] Server: Returns the generated test questions to the user's device as an HTTP response.

[1305] 3. Display the test screen and submit your answers

[1306] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[1307] 4. Analyzing responses and sending results

[1308] Server: The answer analysis module analyzes the received answers and judges whether they are correct or incorrect. It also calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[1309] Terminal: The received results are displayed on the user's screen, allowing the driver to check their self-evaluation results.

[1310] Follow-up message system

[1311] 1. Proficiency Assessment

[1312] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1313] 2. Create and send a follow message

[1314] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[1315] Server: Sends the generated follow message to the user's device as an HTTP response.

[1316] 3. Displaying follow messages

[1317] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[1318] As described above, this system utilizes a communication platform operated by a taxi company to achieve effective and efficient driver training using a generative AI model. By linking each module, it is possible to provide learning support tailored to each individual's level of proficiency, significantly reducing the burden of training.

[1319] The processing flow will be explained below.

[1320] Question and Answer System

[1321] Step 1:

[1322] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[1323] Step 2:

[1324] Terminal: Sends the entered question to the server as an HTTP request.

[1325] Step 3:

[1326] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1327] Step 4:

[1328] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[1329] Step 5:

[1330] Server: The generated answer is sent back to the user's device as an HTTP response.

[1331] Step 6:

[1332] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[1333] Test Generation and Examination Systems

[1334] Step 1:

[1335] User: Selects "I want to take a test" in a learning app.

[1336] Step 2:

[1337] Device: Sends a test request to the server as an HTTP request.

[1338] Step 3:

[1339] Server: Based on the received request, the test generation module automatically generates test questions, using a generative AI model to generate questions such as "Choose the best route when picking up a passenger using a taxi."

[1340] Step 4:

[1341] Server: Returns the generated test questions to the user's device as an HTTP response.

[1342] Step 5:

[1343] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[1344] Step 6:

[1345] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[1346] Step 7:

[1347] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[1348] Step 8:

[1349] Server: Returns the evaluation results to the user device as an HTTP response.

[1350] Step 9:

[1351] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[1352] Follow-up message system

[1353] Step 1:

[1354] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1355] Step 2:

[1356] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[1357] Step 3:

[1358] Server: Sends the generated follow message to the user's device as an HTTP response.

[1359] Step 4:

[1360] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[1361] Example 1

[1362] 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."

[1363] Conventional driver training systems have had issues with the difficulty of providing effective training tailored to individual proficiency levels, resulting in low overall training efficiency. Furthermore, question-answering and test question generation are often done manually, which requires time and effort. Furthermore, there are also issues with insufficient evaluation of driver proficiency, resulting in a lack of appropriate feedback and follow-up.

[1364] 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.

[1365] In this invention, the server includes a device that performs machine learning using a dataset of the Class 2 driver's license test, a device that responds to questions from the driver using a model obtained by the machine learning, a device that automatically generates test questions using the model, a device that analyzes the test results answered by the driver and evaluates the driver's proficiency, a device that generates and sends an individual follow-up message based on the proficiency evaluation, a device that receives the driver's input and sends it to the server as an HTTP request, a device that analyzes the content of the questions using natural language processing and generates answers using a generative AI model, and a terminal that receives and displays the generated answers to show to the driver. This makes it possible to quickly answer the driver's questions, automatically generate test questions according to the individual proficiency level, and provide appropriate evaluations and individual follow-ups.

[1366] The "Type 2 license test" is a driver's license test required to drive passenger vehicles such as taxis and buses.

[1367] A "dataset" is a collection of data collected for a specific purpose and used in machine learning and data analysis.

[1368] "Machine learning" is a technology that allows computers to learn patterns and knowledge from data and perform specific tasks automatically.

[1369] A "model" is an algorithm or function generated by machine learning that makes predictions or classifications on new data.

[1370] "Driver" refers to a professional driver who drives a vehicle such as a taxi or bus.

[1371] A "device that responds to questions" is a system that receives questions from the driver and generates and provides appropriate answers.

[1372] "Test questions" are test questions given to assess a driver's knowledge and skills.

[1373] "Analysis" is the process of examining data or information in detail to find meaning and patterns.

[1374] "Proficiency" is an indicator of how much knowledge and skill a driver has in a particular field.

[1375] A "follow-up message" is a customized message based on the proficiency assessment to encourage additional learning and improvement for the driver.

[1376] An "HTTP request" is a data request sent from a client to a server.

[1377] "Natural language processing" is a technology that allows computers to understand human language, and includes morphological analysis and semantic analysis.

[1378] A "generative AI model" is an artificial intelligence model that has been trained to perform tasks such as text generation and question answering.

[1379] A "terminal" is an electronic device that allows a user to interact with a system through an interface.

[1380] MODE FOR CARRYING OUT THE INVENTION

[1381] This invention relates to a driver training system based on machine learning techniques, specifically using a dataset from the Class 2 driver's license test. The system uses a generative AI model to respond to driver questions, automatically generate test questions, analyze test results, and provide personalized follow-up messages based on the driver's proficiency assessment.

[1382] System Configuration

[1383] 1. Hardware and Software Overview

[1384] Device: An electronic device (e.g., smartphone, tablet, computer) equipped with a learning application used by a driver.

[1385] Server: A central system for analyzing questions, generating test questions, analyzing answers, and generating follow-up messages.

[1386] Natural Language Processing (NLP) libraries, such as spaCy.

[1387] Generative AI models: Use large language models such as GPT-3.

[1388] Database: Stores past questions from the Class 2 license exam and driving knowledge materials.

[1389] System Operation

[1390] 1. Question and Answer System

[1391] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1392] Terminal: Send this input to the server as an HTTP request.

[1393] Server: Passes the received question to the chatbot module, and performs morphological analysis using a natural language processing library (e.g., spaCy) to analyze keywords and sentence structure.

[1394] Server: Enter the prompt "How often do I need to renew my driver's license?" into a generative AI model (e.g., GPT-3) and generate an appropriate answer. The generated text is "A driver's license generally needs to be renewed every five years."

[1395] Server: Sends the generated answer to the terminal as an HTTP response.

[1396] Terminal: The received answer is displayed on the user interface, and the driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[1397] 2. Test Generation and Examination Systems

[1398] User: The driver selects "I want to take the test" in the learning application.

[1399] Device: Sends a test request to the server as an HTTP request.

[1400] Server: Based on the received request, the test generation module references the database and generates new test questions, such as "Choose the best route when picking up a passenger using a taxi."

[1401] Server: Sends the generated test questions to the device as an HTTP response.

[1402] Terminal: The test questions are displayed in a user interface for the driver to answer. The driver enters the answers and sends them to the server as HTTP requests.

[1403] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, calculates the driver's proficiency, generates an evaluation result, and sends it to the user's device as an HTTP response.

[1404] Terminal: The received evaluation results are displayed on the user interface, allowing the driver to check their self-evaluation results.

[1405] 3. Follow-up message system

[1406] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects areas of low proficiency (e.g., road traffic laws).

[1407] Server: Generate a personalized follow-up message for each subject area that the follow-up module detects. For example, create a message that says, "Review your basic knowledge of road traffic laws. Additional learning materials are available here."

[1408] Server: Sends the generated follow message to the user's device as an HTTP response.

[1409] Terminal: The received follow-up message is displayed on the user interface, and the driver can receive specific learning instructions.

[1410] Examples of concrete examples and prompts

[1411] Example 1: A driver asks, "How often do I need to renew my driver's license?" and the server uses an AI model to generate and provide the answer, "Basically, a driver's license needs to be renewed every five years."

[1412] Example prompt sentence:

[1413] Question-answering system prompt: "How often do I need to renew my driver's license?"

[1414] Test generation system prompt: "Generate test questions based on the Class 2 taxi driver's license exam."

[1415] Follow-up message system prompt: "This driver failed multiple questions about road traffic laws. Please generate a follow-up message to help them study again."

[1416] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1417] Processing flow

[1418] Question and Answer System

[1419] Step 1: Enter and submit your question

[1420] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1421] Input: The driver's text input.

[1422] Terminal: Sends the entered question to the server as an HTTP request.

[1423] Output: HTTP request to the server.

[1424] Step 2: Receiving and parsing the question

[1425] Server: Receives the HTTP request and retrieves the inquiry.

[1426] Input: The HTTP request received from the device.

[1427] Server: Using an NLP library (e.g., spaCy), morphologically analyze the question and extract keywords and sentence structure.

[1428] Output: Parsed keywords and sentence structure.

[1429] Step 3: Generate an answer

[1430] Server: The analyzed question content is input as a prompt to the generative AI model (e.g., GPT-3).

[1431] Input: Parsed keywords and sentence structure.

[1432] Server: Generates an appropriate answer using the AI ​​model. It generates the text, "Driver's licenses generally need to be renewed every five years."

[1433] Output: The generated answer text.

[1434] Step 4: Return and view your responses

[1435] Server: Sends the generated answer to the user's device as an HTTP response.

[1436] Input: The generated answer text.

[1437] Output: HTTP response to the user's device.

[1438] Device: The received answer is displayed on the screen. The driver can confirm the answer: "Driver's licenses generally need to be renewed every five years."

[1439] Test Generation and Examination Systems

[1440] Step 1: Submit a test request

[1441] User: The driver selects "I want to take the test" in the learning application.

[1442] Input: Driver test request.

[1443] Terminal: Send this request to the server as an HTTP request.

[1444] Output: HTTP request to the server.

[1445] Step 2: Generate test questions

[1446] Server: Receives requests and passes instructions to the test generation module.

[1447] Input: Exam request.

[1448] Server: Refers to the database and automatically generates new test questions. For example, it generates a question such as "Choose the best route when picking up a passenger by taxi."

[1449] Output: The generated test questions.

[1450] Step 3: Submit and view the test questions

[1451] Server: Sends the generated test questions to the user's device as an HTTP response.

[1452] Input: The generated test questions.

[1453] Output: HTTP response to the user's device.

[1454] Terminal: Displays the received test questions on the screen so that the driver can answer them.

[1455] Input: Show test question.

[1456] Output: Driver answer input.

[1457] Step 4: Submit and analyze responses

[1458] Terminal: After entering the driver's answers, send them to the server as an HTTP request.

[1459] Input: Driver's answer.

[1460] Output: HTTP request to the server.

[1461] Server: The answer analysis module analyzes the received answer data, determines whether the answer is correct, and calculates the driver's proficiency level.

[1462] Input: Received response data.

[1463] Output: Evaluation result.

[1464] Step 5: Send and view results

[1465] Server: Sends the evaluation results to the user device as an HTTP response.

[1466] Input: Evaluation result.

[1467] Output: HTTP response to the user's device.

[1468] Terminal: The received evaluation results are displayed on the user's screen. Drivers can check their own evaluation results.

[1469] Input: Display evaluation results.

[1470] Output: Driver verification.

[1471] Follow-up message system

[1472] Step 1: Assess proficiency

[1473] Server: The response analysis module evaluates the driver's proficiency based on past test results.

[1474] Input: Past test results.

[1475] Server: Detect low proficiency in a specific subject (e.g., road traffic law).

[1476] Output: Low proficiency area detection results.

[1477] Step 2: Generate a follow-up message

[1478] Server: The follow-up module generates individual follow-up messages for low-proficiency areas.

[1479] Input: Information in low proficiency areas.

[1480] Server: Generate a message such as "Review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[1481] Output: The generated follow message.

[1482] Step 3: Send and view follow-up messages

[1483] Server: Sends the generated follow message to the user's device as an HTTP response.

[1484] Input: The generated follow message.

[1485] Output: HTTP response to the user's device.

[1486] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[1487] Input: Show follow messages.

[1488] Output: Driver verification.

[1489] (Application example 1)

[1490] 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."

[1491] Current training for robot operators in factories relies on instruction by human trainers and static teaching materials, making it difficult to provide efficient training tailored to each operator's proficiency level. It is also difficult to respond to operator questions and provide appropriate teaching materials in real time, resulting in limited training efficiency and effectiveness. This invention aims to solve these issues by using a generative AI model to provide a flexible and efficient training system tailored to each operator's proficiency level.

[1492] 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.

[1493] In this invention, the server includes a means for performing machine learning using a dataset of the Class 2 license examination, a means for responding to questions from operators using a model obtained by the machine learning, and a means for automatically generating test questions using the model. This not only enables quick and accurate responses to questions from operators, but also makes it possible to provide test questions tailored to the learning status of each individual operator and appropriately generate and send follow-up messages based on their proficiency.

[1494] The "Class 2 license examination dataset" refers to a wide range of test questions and answer data used to evaluate an operator's skills and knowledge.

[1495] "Machine learning" refers to the technology in which computers automatically learn using large amounts of data to generate algorithms and models for performing specific tasks.

[1496] A "model" refers to a data processing structure built using machine learning to predict or classify a specific outcome based on input data.

[1497] "Operator" refers to an employee who operates and manages robots and machines within a factory.

[1498] "Test Questions" refers to automatically generated test questions to evaluate an operator's knowledge and skills.

[1499] "Proficiency" refers to an indicator that evaluates how much an operator has acquired a particular skill or knowledge.

[1500] "Follow-up messages" refer to messages containing instructions or advice sent individually to encourage further learning and improvement based on the results of an operator's proficiency assessment.

[1501] "System" refers to the overall infrastructure and process for supporting training using machine learning models implemented according to this invention.

[1502] This invention is a system for supporting the training of operators of robots used in factories. It performs machine learning using a dataset from the Class 2 license examination, and uses the model to respond to the operator's questions, automatically generate tests, evaluate proficiency, and send follow-up messages.

[1503] Overall system overview

[1504] This system consists of three main functions:

[1505] 1. Question and Answer System

[1506] 2. Test Generation and Examination Systems

[1507] 3. Proficiency-based follow-up message system

[1508] Question and Answer System

[1509] The server receives a question from the operator and analyzes it using the chatbot module. Specifically, it uses natural language processing (NLP) technology to morphologically analyze the content of the question and extract its meaning. It then uses a generative AI model (e.g., GPT-3) to generate an appropriate answer and sends it back to the operator's device. The operator can then view the received answer on their device's screen.

[1510] Examples:

[1511] When an operator uses a smartphone to input a question such as, "Where is the emergency stop button on the robot?", the server receives the question, analyzes it, and generates and returns the answer, "The emergency stop button is on the left side of the robot's control panel."

[1512] Example prompt sentence:

[1513] I have a question about the emergency stop button on the robot. Can you tell me where it is?

[1514] Test Generation and Examination Systems

[1515] When the server receives a test request from an operator, it automatically generates test questions using a test generation module. The generated test questions are sent to the operator's terminal and displayed. The operator answers the questions and sends the answers to the server. The server analyzes the answers, determines whether they are correct or incorrect, evaluates the operator's proficiency, and returns the results.

[1516] Examples:

[1517] When an operator requests a test on their smartphone, the server generates a question such as "What is the normal maintenance cycle for a robot?" and sends it to the operator's device. When the operator enters and submits the answer, the answer is analyzed and the results and evaluation are sent back to the device.

[1518] Proficiency-based follow-up message system

[1519] The server evaluates the operator's proficiency based on past test results. Based on the evaluation results, follow-up messages are individually generated for specific areas and sent to the operator's terminal, allowing the operator to conduct additional learning in areas where their understanding is lacking.

[1520] Examples:

[1521] If the server detects low proficiency in "robot safety operation procedures," it generates a follow-up message such as "Please relearn the specific area. Please refer to the manual in the system for details," and sends it to the operator's terminal.

[1522] Example prompt sentence:

[1523] Your proficiency has been verified. You will need to refresh your knowledge in the following areas: Robot safety operating procedures.

[1524] Technologies and Tools Used

[1525] Hardware: Server, user device (smartphone)

[1526] Software: Natural Language Processing (NLP) module, generative AI model (e.g. GPT-3), HTTP communication protocol

[1527] The implementation of this system will not only enable effective and efficient training tailored to each operator's individual level of proficiency, but will also enable responses to questions in real time and the provision of appropriate learning instructions.

[1528] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1529] Step 1:

[1530] The user opens the learning application on their smartphone and inputs a question, for example, "Where is the emergency stop button on the robot?" This question is captured as input data on the device.

[1531] Step 2:

[1532] The device sends the entered question to the server as an HTTP request, which is in JSON format and contains the question itself.

[1533] Step 3:

[1534] The server passes the received question to the chatbot module, which uses natural language processing (NLP) to morphologically analyze the question text and extract the meaning of the question. Through this process, the question content is interpreted.

[1535] Step 4:

[1536] The server inputs the interpreted question into a generative AI model (e.g., GPT-3) and generates an appropriate answer. The generated answer might be something like, "There is an emergency stop button on the left side of the robot's control panel."

[1537] Step 5:

[1538] The server returns the generated answer to the user device as an HTTP response, which is also in JSON format and contains the answer text.

[1539] Step 6:

[1540] The device displays the received response on the user's screen, and the user can see the response on their smartphone screen: "There is an emergency stop button on the left side of the robot's control panel."

[1541] Step 7:

[1542] A user selects "I want to take a test" in a learning application. This request is received as input data on the device and sent to the server as an HTTP request.

[1543] Step 8:

[1544] The server automatically generates test questions using a test generation module based on the received test request. For example, a question such as "What is the normal maintenance cycle for a robot?" is generated.

[1545] Step 9:

[1546] The server sends the generated test questions to the user's terminal as an HTTP response. The terminal displays the received test questions on the user's screen, and the user answers the questions.

[1547] Step 10:

[1548] The user inputs the answer to the question, and the terminal sends it to the server as an HTTP request. This request contains data that includes the answer.

[1549] Step 11:

[1550] The server analyzes the received answers and judges whether they are correct or not. The answer analysis module then calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[1551] Step 12:

[1552] The device displays the received evaluation results on the user's screen, allowing the user to check their self-evaluation results on their smartphone screen.

[1553] Step 13:

[1554] The server evaluates the user's proficiency based on past test results and detects low proficiency in a specific area (e.g., robot safety operating procedures).

[1555] Step 14:

[1556] The server generates a follow-up message for the area of ​​low proficiency and sends it to the user's device as an HTTP response. For example, a follow-up message such as "Please restudy the specific area. For details, please refer to the system manual" is generated.

[1557] Step 15:

[1558] The device displays the received follow message on the user's screen, allowing the user to receive specific study instructions on the smartphone screen.

[1559] 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.

[1560] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The specific system configuration combines a machine learning model using a dataset from the Class 2 driver's license exam with an emotion engine to achieve an educational effect that takes into account the driver's emotional state.

[1561] Question and Answer System

[1562] 1. Enter and submit your question

[1563] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1564] Terminal: Sends the entered question to the server as an HTTP request.

[1565] 2. Receiving and analyzing questions

[1566] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1567] 3. Emotion Recognition and Answer Generation

[1568] Server: The emotion engine receives emotion data (e.g., facial recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[1569] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[1570] 4. Returning and displaying answers

[1571] Server: The generated answer is sent back to the user's device as an HTTP response.

[1572] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[1573] Test Generation and Examination Systems

[1574] 1. Submit a test request

[1575] User: The driver selects "I want to take the test" in the learning application.

[1576] Device: Sends a test request to the server as an HTTP request.

[1577] 2. Generate and submit test questions

[1578] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the best route when picking up a passenger using a taxi."

[1579] Server: Returns the generated test questions to the user's device as an HTTP response.

[1580] 3. Display the test screen and submit your answers

[1581] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[1582] 4. Analyzing responses and sending results

[1583] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[1584] Server: Returns the evaluation results to the user device as an HTTP response.

[1585] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[1586] Follow-up message system

[1587] 1. Proficiency Assessment

[1588] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1589] 2. Create and send a follow message

[1590] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[1591] 3. Displaying follow messages

[1592] Server: Sends the generated follow message to the user's device as an HTTP response.

[1593] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[1594] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[1595] The processing flow will be explained below.

[1596] Question and Answer System

[1597] Step 1:

[1598] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[1599] Step 2:

[1600] Terminal: Sends the entered question to the server as an HTTP request.

[1601] Step 3:

[1602] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1603] Step 4:

[1604] Server: The emotion engine receives emotion data (e.g., facial expression recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[1605] Step 5:

[1606] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[1607] Step 6:

[1608] Server: The generated answer is sent back to the user's device as an HTTP response.

[1609] Step 7:

[1610] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[1611] Test Generation and Examination Systems

[1612] Step 1:

[1613] User: Selects "I want to take a test" in a learning app.

[1614] Step 2:

[1615] Device: Sends a test request to the server as an HTTP request.

[1616] Step 3:

[1617] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the optimal route when picking up a passenger using a taxi."

[1618] Step 4:

[1619] Server: Returns the generated test questions to the user's device as an HTTP response.

[1620] Step 5:

[1621] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[1622] Step 6:

[1623] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[1624] Step 7:

[1625] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[1626] Step 8:

[1627] Server: Returns the evaluation results to the user device as an HTTP response.

[1628] Step 9:

[1629] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[1630] Follow-up message system

[1631] Step 1:

[1632] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1633] Step 2:

[1634] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[1635] Step 3:

[1636] Server: Sends the generated follow message to the user's device as an HTTP response.

[1637] Step 4:

[1638] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[1639] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[1640] Example 2

[1641] 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."

[1642] Conventional driver training systems often provide uniform training without considering the individual needs or emotional state of each learner. Furthermore, detailed feedback based on test results is not provided, potentially reducing learning effectiveness. Furthermore, training utilizing emotional data has not been implemented, making it impossible to reduce learner stress. Therefore, a system that provides personalized training while taking into account the learner's emotional state was needed.

[1643] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending an individual follow-up message based on the proficiency evaluation, and means for analyzing emotion data from the user terminal and adding additional information based on the emotion to the answers generated by the model. This enables optimal education that takes into account the user's emotional state, thereby improving the effectiveness of the education.

[1644] "Machine learning" is a technology that allows a computer to analyze a given data set and automatically learn patterns and rules based on that data.

[1645] The "Class 2 Driver's License Examination Dataset" is a large-scale data set that includes questions and answers related to the Class 2 Driver's License Examination, as well as their trends and characteristics.

[1646] A "generative AI model" is a model that uses artificial intelligence to generate content such as text and images, and specifically includes natural language processing and generation modules.

[1647] A "chatbot" is a program that can interact with users and respond to their questions via text or voice.

[1648] "Emotion data" is data that indicates the user's emotional state, and examples include facial expression data and voice tone.

[1649] An "emotion engine" is an algorithm or system that analyzes emotional data to detect and evaluate a user's emotional state.

[1650] "Proficiency" is a measure of how well a driver understands or has mastered a particular subject or task.

[1651] A "follow-up message" is a message such as instruction or encouragement that is generated based on the user's learning progress or emotional state.

[1652] An "HTTP request" is a request for data sent by a client to a server, and is a protocol used by web browsers and applications to request data from a server.

[1653] An "HTTP response" is a data response that a server returns in response to a client's HTTP request, and includes requested data and status information.

[1654] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. Specific embodiments of the system are as follows.

[1655] Hardware and software used

[1656] This system uses the following hardware and software:

[1657] Server: A computer that performs major processing. It is equipped with a high-performance processor (e.g., Intel Xeon) and a large amount of memory (e.g., 64 GB or more).

[1658] Device: The device used by the driver, such as a smartphone or tablet (e.g., Android or iOS).

[1659] Generative AI model: An artificial intelligence model that generates text, for example using an advanced natural language processing model such as GPT-3.

[1660] Emotion engine: Software for analyzing user emotion data. Examples include Azure Cognitive Services' emotion recognition API.

[1661] NLP engine: Natural language processing software to analyze the question text, for example using spaCy or NLTK.

[1662] System processing flow

[1663] Answers to questions

[1664] A user opens a learning application and inputs a question, for example, "How often do I need to renew my driver's license?" The device sends this question to the server as an HTTP request. The server passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text. The server then uses an emotion engine to analyze emotion data (e.g., facial recognition and voice analysis data) from the user's device and determine the user's emotional state.

[1665] A generative AI model (e.g., GPT-3) is used to generate the optimal answer based on the analyzed question content and data from the emotion engine. For example, a standard answer such as "Your driver's license generally needs to be renewed every five years" is generated. If the user expresses anxiety, the server adds additional information such as "Don't worry, the process is simple."

[1666] The generated answer is then returned to the user terminal as an HTTP response, and the terminal displays the received answer on the user screen.

[1667] Test Generation and Taking

[1668] When a user selects "I want to take a test" in the learning application, the device sends a test request to the server. Based on the received request, the server uses a generative AI model to automatically generate test questions. For example, a question such as "Select the optimal route when picking up a passenger by taxi" is generated.

[1669] The generated test questions are sent back to the user's device, where the user can check them on the screen and enter their answers. The device then sends these answers to the server, which uses an answer analysis module to determine whether the answers are correct or incorrect and evaluate the user's proficiency. The results are then sent back to the user's device and displayed.

[1670] Follow message generation

[1671] The server evaluates the user's proficiency based on past test results and detects low proficiency in specific areas. It then uses an emotion engine to analyze the user's emotional state and generates appropriate follow-up messages based on that. For example, it generates specific learning instructions such as, "Please review your basic knowledge of road traffic laws. Take your time and review them carefully."

[1672] The follow message thus generated is sent to the user's terminal, and the user can check it on the screen.

[1673] Examples of concrete examples and prompts

[1674] For example, if a user asks, "How often do I need to renew my driver's license?", the following prompt is sent to the generative AI model:

[1675] User: "How many years do I need to renew my driver's license?"

[1676] Generative AI model: "Driver's licenses generally need to be renewed every five years."

[1677] Add emotion: "Don't worry, it's a simple process."

[1678] In this way, by combining a generative AI model with an emotion engine, this system is able to provide optimal training that takes into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a learning environment that is easier for drivers to understand and less stressful.

[1679] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1680] Step 1:

[1681] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1682] Input: Keyboard input data.

[1683] Output: Question text data.

[1684] What happens: The user types a question into a text input field in the application and presses the Enter key to confirm the question.

[1685] Step 2:

[1686] Terminal: Sends the entered question to the server as an HTTP request.

[1687] Input: Question text data.

[1688] Output: HTTP request packet.

[1689] Specific operation: Converts text data into HTTP packets and sends them over the network to the server.

[1690] Step 3:

[1691] Server: Passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text.

[1692] Input: HTTP request packet, text data.

[1693] Output: Parsed question semantic data.

[1694] What it does: The server decodes the HTTP packets, passes the text data to an NLP engine (e.g., spaCy or NLTK), and performs morphological analysis to extract the meaning of the question.

[1695] Step 4:

[1696] Server: Uses the emotion engine to receive and analyze emotion data (e.g., facial recognition, voice analysis data) from the user device.

[1697] Input: Emotion data, e.g., image or audio files.

[1698] Output: User's emotional state data.

[1699] Specific operation: Emotion data is passed to an emotion engine (e.g., Azure Cognitive Services emotion recognition API) for analysis, and the user's emotional state is output as numerical data.

[1700] Step 5:

[1701] Server: Using a generative AI model, it generates answers based on the analyzed question content and sentiment data.

[1702] Input: Question semantic data, emotional state data.

[1703] Output: Answer data in text format.

[1704] What it does: The semantic data and emotional state data of the question are incorporated into a prompt sentence, which is then sent to a generative AI model (e.g., GPT-3) to generate an answer, such as, "Your driver's license generally needs to be renewed every five years. Don't worry, it's a simple process."

[1705] Step 6:

[1706] Server: The generated answer is sent back to the user's device as an HTTP response.

[1707] Input: Textual response data.

[1708] Output: HTTP response packet.

[1709] Specific operation: The response data is formatted as an HTTP response and sent to the user terminal via the network.

[1710] Step 7:

[1711] Terminal: Displays the received answer on the user's screen.

[1712] Input: HTTP response packet.

[1713] Output: Screen display data.

[1714] What it does: It extracts text data from the HTTP response and displays it in a display field in the application. The user can see the answer on the screen: "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[1715] Step 8:

[1716] User: The driver selects "I want to take the test" in the learning application.

[1717] Input: A test-taking request.

[1718] Output: Test attempt selection data.

[1719] Specific behavior: A user clicks the "I want to take a test" button in the application to generate a test request.

[1720] Step 9:

[1721] Device: Sends a test request to the server as an HTTP request.

[1722] Input: Test attempt selection data.

[1723] Output: HTTP request packet.

[1724] Specific operation: The test selection data is converted into an HTTP request format and sent to the server via the network.

[1725] Step 10:

[1726] Server: Based on the received request, the server automatically generates test questions using a generative AI model.

[1727] Input: HTTP request packet.

[1728] Output: The generated test question data.

[1729] Specific operation: The request content is converted into a prompt for the generative AI model, and test questions are automatically generated by the generative AI model (e.g., GPT-3). For example, questions such as "Choose the optimal route when picking up a passenger by taxi" are generated.

[1730] Step 11:

[1731] Server: Returns the generated test questions to the user's device as an HTTP response.

[1732] Input: Generated test question data.

[1733] Output: HTTP response packet.

[1734] Specific operation: The generated test questions are formatted as an HTTP response and sent to the user's device.

[1735] Step 12:

[1736] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions.

[1737] Input: HTTP response packet.

[1738] Output: Screen display data.

[1739] Specific behavior: Analyzes received test questions, displays them on the screen, and provides an interface for users to enter answers.

[1740] Step 13:

[1741] User: The driver enters answers to the test questions and sends them to the server.

[1742] Input: User response data.

[1743] Output: HTTP request packet.

[1744] Specific operation: Converts the user's input answer into HTTP request format and sends it to the server.

[1745] Step 14:

[1746] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, and evaluates the level of proficiency.

[1747] Input: User response data.

[1748] Output: Evaluation result data.

[1749] Specific operation: The answers are passed to the analysis module, which determines whether they are correct or incorrect and generates an evaluation result. Using the generative AI model, it also properly evaluates non-standard answers.

[1750] Step 15:

[1751] Server: Returns the evaluation results to the user device as an HTTP response.

[1752] Input: Evaluation result data.

[1753] Output: HTTP response packet.

[1754] Specific operation: The evaluation results are converted into HTTP response format and sent to the user's terminal.

[1755] Step 16:

[1756] Terminal: The evaluation results are displayed on the user's screen.

[1757] Input: HTTP response packet.

[1758] Output: Screen display data.

[1759] Specific operation: The received evaluation results are displayed on the screen, allowing the user to check their own proficiency level.

[1760] (Application example 2)

[1761] 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."

[1762] Conventional driver training systems did not take into account the emotional state of the driver, resulting in ineffective learning and testing. Furthermore, with factory robots, training and proficiency management could not be carried out efficiently, limiting the improvement of robot performance. This led to robot operation errors and slower work speeds, affecting factory production efficiency.

[1763] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1764] In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending individual follow-up messages based on the proficiency evaluation, and means for analyzing the user's emotional data using an emotion engine and adding additional information to the answers based on the analysis results, thereby enabling efficient education and proficiency management that takes emotional states into account.

[1765] The "Class 2 Driver's License Examination Dataset" is a collection of data that includes the test questions, answers, and related information required to obtain a driver's license.

[1766] "Machine learning" is a technology that allows computers to automatically learn from data and improve their ability to perform specific tasks.

[1767] A "model" is a mathematical or statistical representation for data analysis obtained through machine learning.

[1768] "Users" refers to drivers and factory workers who use this system.

[1769] A "means for responding to questions" is a method or system for generating and replying to questions from users using natural language.

[1770] "Means for automatically generating test questions" refers to a method for automatically creating test questions based on a pre-prepared dataset or algorithm.

[1771] "Means for analyzing test results" refers to a method or system for analyzing the percentage of correct answers to test questions answered by users and their trends.

[1772] A "proficiency assessment tool" is a method for assessing a user's skill or knowledge level based on test results.

[1773] The "means for generating and sending follow-up messages" is a method for creating and sending additional guidance and encouraging messages based on the user's proficiency assessment.

[1774] An "emotion engine" is a technology that analyzes a user's emotional data and evaluates their emotional state.

[1775] "Emotion data" refers to data that indicates the emotional state of a user obtained from facial expressions, voice, and other biological signals.

[1776] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The method for realizing this system is described in detail below.

[1777] 1. Program Generation

[1778] The program in this system basically works as follows:

[1779] Train a machine learning model using a dataset from the Class 2 license examination.

[1780] The trained model is used to generate answers to questions posed by users.

[1781] Test questions are automatically generated using a generative AI model and provided to users.

[1782] The results of the test answered by the user are analyzed to evaluate their level of proficiency.

[1783] The emotion engine is used to analyze users' emotional data, and additional information is added to replies and follow-up messages based on the analysis results.

[1784] 2. System Configuration

[1785] Hardware

[1786] Server: A central location that receives, analyzes, and generates data. It is used to train and run models and analyze sentiment data.

[1787] Device (smartphone, tablet, etc.): A device operated by a user, used to enter questions, take tests, and receive feedback.

[1788] Factory robots: Devices that operate in factories and undergo learning and testing.

[1789] software

[1790] Natural Language Processing (NLP) models are used to analyze user questions.

[1791] Generative AI models (e.g., GPT-3): Used to generate answers to questions and automatically generate test questions.

[1792] Emotion engine: Analyzes the user's facial expressions and voice data to assess their emotional state.

[1793] HTTP request / response: A protocol for sending and receiving data between a server and a device.

[1794] 3. Specific examples of program processing

[1795] Question-answering system

[1796] User: Ask the robot, "What's the next step?"

[1797] Terminal: Sends the entered question to the server.

[1798] Server: Analyzes the question using an NLP model and generates an answer using a generative AI model. An answer such as "The next task is to connect part A to part B" is generated.

[1799] Server: The sentiment engine analyzes and adds additional information to the answer indicating that the question is easy.

[1800] Terminal: Display the generated answer to the user.

[1801] Test Generation and Evaluation System

[1802] User: The robot selects "I want to take the test."

[1803] Server: The generative AI model generates test questions such as "Select the steps to correctly connect the parts."

[1804] Terminal: Displays test questions on the user's screen and receives answers.

[1805] Server: Analyzes the answers and assesses proficiency.

[1806] Server: Based on the evaluation results, a follow-up message such as "The next step is this procedure" is generated, and if the emotion engine indicates impatience in its analysis, additional information such as "Don't rush, take your time to check it out" is added.

[1807] Device: View follow messages.

[1808] 4. Examples of prompts

[1809] Specifically, the following prompts are used:

[1810] "Question: What's the next step?\nEmotion: Relaxed\nAnswer:"

[1811] "Question: Please explain the steps to connect parts correctly.\nEmotion: stressed\nAnswer: First, prepare part A, then attach it to part B. Take your time and check carefully."

[1812] In this way, by combining a generative AI model with an emotion engine, this system can provide individualized training that takes into account the emotional state of factory robots and drivers in their learning and work, preventing operational errors and slowdowns in work speeds, and improving production efficiency.

[1813] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1814] Step 1:

[1815] The user asks the robot a question.

[1816] Specific operation: The user inputs "Tell me the next step" into the factory robot's interface.

[1817] Input: The question text from the user.

[1818] Output: The question text is typed into the terminal.

[1819] Step 2:

[1820] The terminal sends a question to the server.

[1821] Specific operation: The terminal sends the question text to the server as an HTTP request.

[1822] Input: Question text.

[1823] Output: The question text is sent to the server.

[1824] Step 3:

[1825] The server receives and parses the query.

[1826] How it works: The server receives the question text and analyzes it using a natural language processing (NLP) model to extract meaning.

[1827] Input: The submitted question text.

[1828] Output: Analysis of the question (text meaning).

[1829] Step 4:

[1830] The server uses an emotion engine to analyze the user's emotion data.

[1831] Specific operation: The server analyzes emotional data (facial expressions and voice data) obtained from the robot's sensors using an emotion engine.

[1832] Input: User emotion data.

[1833] Output: Emotional state analysis (e.g., relaxed, stressed).

[1834] Step 5:

[1835] The server generates an answer using a generative AI model.

[1836] Specific operation: The server generates an answer using a generative AI model (e.g., GPT-3) based on the analysis results of the NLP model and the emotion engine.

[1837] Input: Question content analysis results and emotional state analysis results.

[1838] Output: The generated answer (e.g., "Your next task is to connect part A to part B. Don't worry, it's easy.").

[1839] Step 6:

[1840] The server sends the generated response to the terminal.

[1841] Specific operation: The server sends the generated answer to the terminal as an HTTP response.

[1842] Input: The generated answer.

[1843] Output: The answer is sent to the terminal.

[1844] Step 7:

[1845] The device displays the answer to the user.

[1846] Specific operation: The answer received by the terminal is displayed on the screen and provided to the user.

[1847] Input: The answer sent by the server.

[1848] Output: The user confirms the answer.

[1849] Step 8:

[1850] The user requests the robot to take the test.

[1851] Specific operation: The user inputs "I would like to take a test" through the factory robot's interface.

[1852] Input: A request to take a test.

[1853] Output: The request is typed into the terminal.

[1854] Step 9:

[1855] The device sends a test request to the server.

[1856] Specific operation: The device sends a test-taking request to the server as an HTTP request.

[1857] Input: A test request.

[1858] Output: The request is sent to the server.

[1859] Step 10:

[1860] The server generates test questions using a generative AI model.

[1861] How it works: The server uses a generative AI model (e.g., GPT-3) to automatically generate appropriate test questions.

[1862] Input: A test request.

[1863] Output: The generated test questions.

[1864] Step 11:

[1865] The server sends the generated test questions to the terminal.

[1866] Specific operation: The server sends the generated test questions to the terminal as an HTTP response.

[1867] Input: Generated test questions.

[1868] Output: The test questions are sent to the terminal.

[1869] Step 12:

[1870] The terminal displays the test questions to the user.

[1871] Specific operation: The device displays the received test questions on the screen and allows the user to enter answers.

[1872] Input: The test questions sent from the server.

[1873] Output: The user checks the test questions and enters their answers.

[1874] Step 13:

[1875] The user answers the test questions.

[1876] Specific operation: The user enters answers to the test questions and sends them to the server via the terminal.

[1877] Input: User's answer.

[1878] Output: The answer is entered into the terminal and sent to the server.

[1879] Step 14:

[1880] The server analyzes the answers and assesses proficiency.

[1881] Specific operation: The server analyzes the received answers, evaluates their appropriateness using a generative AI model, and calculates the user's proficiency level.

[1882] Input: User's answer.

[1883] Output: Proficiency assessment results.

[1884] Step 15:

[1885] The server generates and sends follow-up messages based on the proficiency assessment.

[1886] Specific operation: The server generates a follow-up message based on the proficiency assessment results, adds necessary additional information based on the analysis results of the emotion engine, and sends it to the terminal.

[1887] Input: Proficiency assessment results, emotion engine analysis results.

[1888] Output: The generated follow message.

[1889] Step 16:

[1890] The device displays the follow message to the user.

[1891] Specific operation: The follow message received by the device is displayed on the screen and provided to the user.

[1892] Input: The follow message sent from the server.

[1893] Output: The user checks the follow message.

[1894] 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.

[1895] 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.

[1896] 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.

[1897] [Fourth embodiment]

[1898] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1899] 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.

[1900] 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).

[1901] 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.

[1902] 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.

[1903] 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).

[1904] 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.

[1905] 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.

[1906] 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.

[1907] 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.

[1908] 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.

[1909] 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.

[1910] 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."

[1911] This invention relates to a driver training system that uses a generative AI model on a communication platform operated by a taxi company. Specifically, the system performs machine learning using a dataset from a Class 2 driver's license test, and uses the resulting generative AI model to perform chatbots that respond to drivers' questions, automatic test generation, and proficiency analysis. Below, we provide a natural language explanation of the system's program processing and a concrete example.

[1912] Question and Answer System

[1913] 1. Enter and submit your question

[1914] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[1915] Terminal: Sends the entered question to the server as an HTTP request.

[1916] 2. Receiving and analyzing questions

[1917] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1918] 3. Generate and return the answer

[1919] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[1920] Server: The generated answer is sent back to the user's device as an HTTP response.

[1921] 4. View Answers

[1922] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[1923] Test Generation and Examination Systems

[1924] 1. Submit a test request

[1925] User: The driver selects "I want to take the test" in the learning application.

[1926] Device: Sends a test request to the server as an HTTP request.

[1927] 2. Generate and submit test questions

[1928] Server: Based on the received request, the test generation module automatically generates test questions, such as "Choose the best route when picking up a passenger using a taxi."

[1929] Server: Returns the generated test questions to the user's device as an HTTP response.

[1930] 3. Display the test screen and submit your answers

[1931] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[1932] 4. Analyzing responses and sending results

[1933] Server: The answer analysis module analyzes the received answers and judges whether they are correct or incorrect. It also calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[1934] Terminal: The received results are displayed on the user's screen, allowing the driver to check their self-evaluation results.

[1935] Follow-up message system

[1936] 1. Proficiency Assessment

[1937] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1938] 2. Create and send a follow message

[1939] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[1940] Server: Sends the generated follow message to the user's device as an HTTP response.

[1941] 3. Displaying follow messages

[1942] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[1943] As described above, this system utilizes a communication platform operated by a taxi company to achieve effective and efficient driver training using a generative AI model. By linking each module, it is possible to provide learning support tailored to each individual's level of proficiency, significantly reducing the burden of training.

[1944] The processing flow will be explained below.

[1945] Question and Answer System

[1946] Step 1:

[1947] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[1948] Step 2:

[1949] Terminal: Sends the entered question to the server as an HTTP request.

[1950] Step 3:

[1951] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[1952] Step 4:

[1953] Server: Generates an answer based on the question interpreted using a generative AI model (e.g., GPT-3). Creates an answer text such as "Driver's licenses generally need to be renewed every five years."

[1954] Step 5:

[1955] Server: The generated answer is sent back to the user's device as an HTTP response.

[1956] Step 6:

[1957] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[1958] Test Generation and Examination Systems

[1959] Step 1:

[1960] User: Selects "I want to take a test" in a learning app.

[1961] Step 2:

[1962] Device: Sends a test request to the server as an HTTP request.

[1963] Step 3:

[1964] Server: Based on the received request, the test generation module automatically generates test questions, using a generative AI model to generate questions such as "Choose the best route when picking up a passenger using a taxi."

[1965] Step 4:

[1966] Server: Returns the generated test questions to the user's device as an HTTP response.

[1967] Step 5:

[1968] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[1969] Step 6:

[1970] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[1971] Step 7:

[1972] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[1973] Step 8:

[1974] Server: Returns the evaluation results to the user device as an HTTP response.

[1975] Step 9:

[1976] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[1977] Follow-up message system

[1978] Step 1:

[1979] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[1980] Step 2:

[1981] Server: The follow-up module generates individual follow-up messages for areas of low proficiency. Create a message with the following content: "Please review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[1982] Step 3:

[1983] Server: Sends the generated follow message to the user's device as an HTTP response.

[1984] Step 4:

[1985] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[1986] Example 1

[1987] 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."

[1988] Conventional driver training systems have had issues with the difficulty of providing effective training tailored to individual proficiency levels, resulting in low overall training efficiency. Furthermore, question-answering and test question generation are often done manually, which requires time and effort. Furthermore, there are also issues with insufficient evaluation of driver proficiency, resulting in a lack of appropriate feedback and follow-up.

[1989] 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.

[1990] In this invention, the server includes a device that performs machine learning using a dataset of the Class 2 driver's license test, a device that responds to questions from the driver using a model obtained by the machine learning, a device that automatically generates test questions using the model, a device that analyzes the test results answered by the driver and evaluates the driver's proficiency, a device that generates and sends an individual follow-up message based on the proficiency evaluation, a device that receives the driver's input and sends it to the server as an HTTP request, a device that analyzes the content of the questions using natural language processing and generates answers using a generative AI model, and a terminal that receives and displays the generated answers to show to the driver. This makes it possible to quickly answer the driver's questions, automatically generate test questions according to the individual proficiency level, and provide appropriate evaluations and individual follow-ups.

[1991] The "Type 2 license test" is a driver's license test required to drive passenger vehicles such as taxis and buses.

[1992] A "dataset" is a collection of data collected for a specific purpose and used in machine learning and data analysis.

[1993] "Machine learning" is a technology that allows computers to learn patterns and knowledge from data and perform specific tasks automatically.

[1994] A "model" is an algorithm or function generated by machine learning that makes predictions or classifications on new data.

[1995] "Driver" refers to a professional driver who drives a vehicle such as a taxi or bus.

[1996] A "device that responds to questions" is a system that receives questions from the driver and generates and provides appropriate answers.

[1997] "Test questions" are test questions given to assess a driver's knowledge and skills.

[1998] "Analysis" is the process of examining data or information in detail to find meaning and patterns.

[1999] "Proficiency" is an indicator of how much knowledge and skill a driver has in a particular field.

[2000] A "follow-up message" is a customized message based on the proficiency assessment to encourage additional learning and improvement for the driver.

[2001] An "HTTP request" is a data request sent from a client to a server.

[2002] "Natural language processing" is a technology that allows computers to understand human language, and includes morphological analysis and semantic analysis.

[2003] A "generative AI model" is an artificial intelligence model that has been trained to perform tasks such as text generation and question answering.

[2004] A "terminal" is an electronic device that allows a user to interact with a system through an interface.

[2005] MODE FOR CARRYING OUT THE INVENTION

[2006] This invention relates to a driver training system based on machine learning techniques, specifically using a dataset from the Class 2 driver's license test. The system uses a generative AI model to respond to driver questions, automatically generate test questions, analyze test results, and provide personalized follow-up messages based on the driver's proficiency assessment.

[2007] System Configuration

[2008] 1. Hardware and Software Overview

[2009] Device: An electronic device (e.g., smartphone, tablet, computer) equipped with a learning application used by a driver.

[2010] Server: A central system for analyzing questions, generating test questions, analyzing answers, and generating follow-up messages.

[2011] Natural Language Processing (NLP) libraries, such as spaCy.

[2012] Generative AI models: Use large language models such as GPT-3.

[2013] Database: Stores past questions from the Class 2 license exam and driving knowledge materials.

[2014] System Operation

[2015] 1. Question and Answer System

[2016] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[2017] Terminal: Send this input to the server as an HTTP request.

[2018] Server: Passes the received question to the chatbot module, and performs morphological analysis using a natural language processing library (e.g., spaCy) to analyze keywords and sentence structure.

[2019] Server: Enter the prompt "How often do I need to renew my driver's license?" into a generative AI model (e.g., GPT-3) and generate an appropriate answer. The generated text is "A driver's license generally needs to be renewed every five years."

[2020] Server: Sends the generated answer to the terminal as an HTTP response.

[2021] Terminal: The received answer is displayed on the user interface, and the driver can see the answer, "Driver's licenses generally need to be renewed every five years."

[2022] 2. Test Generation and Examination Systems

[2023] User: The driver selects "I want to take the test" in the learning application.

[2024] Device: Sends a test request to the server as an HTTP request.

[2025] Server: Based on the received request, the test generation module references the database and generates new test questions, such as "Choose the best route when picking up a passenger using a taxi."

[2026] Server: Sends the generated test questions to the device as an HTTP response.

[2027] Terminal: The test questions are displayed in a user interface for the driver to answer. The driver enters the answers and sends them to the server as HTTP requests.

[2028] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, calculates the driver's proficiency, generates an evaluation result, and sends it to the user's device as an HTTP response.

[2029] Terminal: The received evaluation results are displayed on the user interface, allowing the driver to check their self-evaluation results.

[2030] 3. Follow-up message system

[2031] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects areas of low proficiency (e.g., road traffic laws).

[2032] Server: Generate a personalized follow-up message for each subject area that the follow-up module detects. For example, create a message that says, "Review your basic knowledge of road traffic laws. Additional learning materials are available here."

[2033] Server: Sends the generated follow message to the user's device as an HTTP response.

[2034] Terminal: The received follow-up message is displayed on the user interface, and the driver can receive specific learning instructions.

[2035] Examples of concrete examples and prompts

[2036] Example 1: A driver asks, "How often do I need to renew my driver's license?" and the server uses an AI model to generate and provide the answer, "Basically, a driver's license needs to be renewed every five years."

[2037] Example prompt sentence:

[2038] Question-answering system prompt: "How often do I need to renew my driver's license?"

[2039] Test generation system prompt: "Generate test questions based on the Class 2 taxi driver's license exam."

[2040] Follow-up message system prompt: "This driver failed multiple questions about road traffic laws. Please generate a follow-up message to help them study again."

[2041] The flow of the identification process in the first embodiment will be described with reference to FIG.

[2042] Processing flow

[2043] Question and Answer System

[2044] Step 1: Enter and submit your question

[2045] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[2046] Input: The driver's text input.

[2047] Terminal: Sends the entered question to the server as an HTTP request.

[2048] Output: HTTP request to the server.

[2049] Step 2: Receiving and parsing the question

[2050] Server: Receives the HTTP request and retrieves the inquiry.

[2051] Input: The HTTP request received from the device.

[2052] Server: Using an NLP library (e.g., spaCy), morphologically analyze the question and extract keywords and sentence structure.

[2053] Output: Parsed keywords and sentence structure.

[2054] Step 3: Generate an answer

[2055] Server: The analyzed question content is input as a prompt to the generative AI model (e.g., GPT-3).

[2056] Input: Parsed keywords and sentence structure.

[2057] Server: Generates an appropriate answer using the AI ​​model. It generates the text, "Driver's licenses generally need to be renewed every five years."

[2058] Output: The generated answer text.

[2059] Step 4: Return and view your responses

[2060] Server: Sends the generated answer to the user's device as an HTTP response.

[2061] Input: The generated answer text.

[2062] Output: HTTP response to the user's device.

[2063] Device: The received answer is displayed on the screen. The driver can confirm the answer: "Driver's licenses generally need to be renewed every five years."

[2064] Test Generation and Examination Systems

[2065] Step 1: Submit a test request

[2066] User: The driver selects "I want to take the test" in the learning application.

[2067] Input: Driver test request.

[2068] Terminal: Send this request to the server as an HTTP request.

[2069] Output: HTTP request to the server.

[2070] Step 2: Generate test questions

[2071] Server: Receives requests and passes instructions to the test generation module.

[2072] Input: Exam request.

[2073] Server: Refers to the database and automatically generates new test questions. For example, it generates a question such as "Choose the best route when picking up a passenger by taxi."

[2074] Output: The generated test questions.

[2075] Step 3: Submit and view the test questions

[2076] Server: Sends the generated test questions to the user's device as an HTTP response.

[2077] Input: The generated test questions.

[2078] Output: HTTP response to the user's device.

[2079] Terminal: Displays the received test questions on the screen so that the driver can answer them.

[2080] Input: Show test question.

[2081] Output: Driver answer input.

[2082] Step 4: Submit and analyze responses

[2083] Terminal: After entering the driver's answers, send them to the server as an HTTP request.

[2084] Input: Driver's answer.

[2085] Output: HTTP request to the server.

[2086] Server: The answer analysis module analyzes the received answer data, determines whether the answer is correct, and calculates the driver's proficiency level.

[2087] Input: Received response data.

[2088] Output: Evaluation result.

[2089] Step 5: Send and view results

[2090] Server: Sends the evaluation results to the user device as an HTTP response.

[2091] Input: Evaluation result.

[2092] Output: HTTP response to the user's device.

[2093] Terminal: The received evaluation results are displayed on the user's screen. Drivers can check their own evaluation results.

[2094] Input: Display evaluation results.

[2095] Output: Driver verification.

[2096] Follow-up message system

[2097] Step 1: Assess proficiency

[2098] Server: The response analysis module evaluates the driver's proficiency based on past test results.

[2099] Input: Past test results.

[2100] Server: Detect low proficiency in a specific subject (e.g., road traffic law).

[2101] Output: Low proficiency area detection results.

[2102] Step 2: Generate a follow-up message

[2103] Server: The follow-up module generates individual follow-up messages for low-proficiency areas.

[2104] Input: Information in low proficiency areas.

[2105] Server: Generate a message such as "Review your basic knowledge of road traffic laws. Additional learning materials can be found here."

[2106] Output: The generated follow message.

[2107] Step 3: Send and view follow-up messages

[2108] Server: Sends the generated follow message to the user's device as an HTTP response.

[2109] Input: The generated follow message.

[2110] Output: HTTP response to the user's device.

[2111] Device: The received follow-up message is displayed on the user's screen, allowing the driver to receive specific learning instructions.

[2112] Input: Show follow messages.

[2113] Output: Driver verification.

[2114] (Application example 1)

[2115] 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."

[2116] Current training for robot operators in factories relies on instruction by human trainers and static teaching materials, making it difficult to provide efficient training tailored to each operator's proficiency level. It is also difficult to respond to operator questions and provide appropriate teaching materials in real time, resulting in limited training efficiency and effectiveness. This invention aims to solve these issues by using a generative AI model to provide a flexible and efficient training system tailored to each operator's proficiency level.

[2117] 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.

[2118] In this invention, the server includes a means for performing machine learning using a dataset of the Class 2 license examination, a means for responding to questions from operators using a model obtained by the machine learning, and a means for automatically generating test questions using the model. This not only enables quick and accurate responses to questions from operators, but also makes it possible to provide test questions tailored to the learning status of each individual operator and appropriately generate and send follow-up messages based on their proficiency.

[2119] The "Class 2 license examination dataset" refers to a wide range of test questions and answer data used to evaluate an operator's skills and knowledge.

[2120] "Machine learning" refers to the technology in which computers automatically learn using large amounts of data to generate algorithms and models for performing specific tasks.

[2121] A "model" refers to a data processing structure built using machine learning to predict or classify a specific outcome based on input data.

[2122] "Operator" refers to an employee who operates and manages robots and machines within a factory.

[2123] "Test Questions" refers to automatically generated test questions to evaluate an operator's knowledge and skills.

[2124] "Proficiency" refers to an indicator that evaluates how much an operator has acquired a particular skill or knowledge.

[2125] "Follow-up messages" refer to messages containing instructions or advice sent individually to encourage further learning and improvement based on the results of an operator's proficiency assessment.

[2126] "System" refers to the overall infrastructure and process for supporting training using machine learning models implemented according to this invention.

[2127] This invention is a system for supporting the training of operators of robots used in factories. It performs machine learning using a dataset from the Class 2 license examination, and uses the model to respond to the operator's questions, automatically generate tests, evaluate proficiency, and send follow-up messages.

[2128] Overall system overview

[2129] This system consists of three main functions:

[2130] 1. Question and Answer System

[2131] 2. Test Generation and Examination Systems

[2132] 3. Proficiency-based follow-up message system

[2133] Question and Answer System

[2134] The server receives a question from the operator and analyzes it using the chatbot module. Specifically, it uses natural language processing (NLP) technology to morphologically analyze the content of the question and extract its meaning. It then uses a generative AI model (e.g., GPT-3) to generate an appropriate answer and sends it back to the operator's device. The operator can then view the received answer on their device's screen.

[2135] Examples:

[2136] When an operator uses a smartphone to input a question such as, "Where is the emergency stop button on the robot?", the server receives the question, analyzes it, and generates and returns the answer, "The emergency stop button is on the left side of the robot's control panel."

[2137] Example prompt sentence:

[2138] I have a question about the emergency stop button on the robot. Can you tell me where it is?

[2139] Test Generation and Examination Systems

[2140] When the server receives a test request from an operator, it automatically generates test questions using a test generation module. The generated test questions are sent to the operator's terminal and displayed. The operator answers the questions and sends the answers to the server. The server analyzes the answers, determines whether they are correct or incorrect, evaluates the operator's proficiency, and returns the results.

[2141] Examples:

[2142] When an operator requests a test on their smartphone, the server generates a question such as "What is the normal maintenance cycle for a robot?" and sends it to the operator's device. When the operator enters and submits the answer, the answer is analyzed and the results and evaluation are sent back to the device.

[2143] Proficiency-based follow-up message system

[2144] The server evaluates the operator's proficiency based on past test results. Based on the evaluation results, follow-up messages are individually generated for specific areas and sent to the operator's terminal, allowing the operator to conduct additional learning in areas where their understanding is lacking.

[2145] Examples:

[2146] If the server detects low proficiency in "robot safety operation procedures," it generates a follow-up message such as "Please relearn the specific area. Please refer to the manual in the system for details," and sends it to the operator's terminal.

[2147] Example prompt sentence:

[2148] Your proficiency has been verified. You will need to refresh your knowledge in the following areas: Robot safety operating procedures.

[2149] Technologies and Tools Used

[2150] Hardware: Server, user device (smartphone)

[2151] Software: Natural Language Processing (NLP) module, generative AI model (e.g. GPT-3), HTTP communication protocol

[2152] The implementation of this system will not only enable effective and efficient training tailored to each operator's individual level of proficiency, but will also enable responses to questions in real time and the provision of appropriate learning instructions.

[2153] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[2154] Step 1:

[2155] The user opens the learning application on their smartphone and inputs a question, for example, "Where is the emergency stop button on the robot?" This question is captured as input data on the device.

[2156] Step 2:

[2157] The device sends the entered question to the server as an HTTP request, which is in JSON format and contains the question itself.

[2158] Step 3:

[2159] The server passes the received question to the chatbot module, which uses natural language processing (NLP) to morphologically analyze the question text and extract the meaning of the question. Through this process, the question content is interpreted.

[2160] Step 4:

[2161] The server inputs the interpreted question into a generative AI model (e.g., GPT-3) and generates an appropriate answer. The generated answer might be something like, "There is an emergency stop button on the left side of the robot's control panel."

[2162] Step 5:

[2163] The server returns the generated answer to the user device as an HTTP response, which is also in JSON format and contains the answer text.

[2164] Step 6:

[2165] The device displays the received response on the user's screen, and the user can see the response on their smartphone screen: "There is an emergency stop button on the left side of the robot's control panel."

[2166] Step 7:

[2167] A user selects "I want to take a test" in a learning application. This request is received as input data on the device and sent to the server as an HTTP request.

[2168] Step 8:

[2169] The server automatically generates test questions using a test generation module based on the received test request. For example, a question such as "What is the normal maintenance cycle for a robot?" is generated.

[2170] Step 9:

[2171] The server sends the generated test questions to the user's terminal as an HTTP response. The terminal displays the received test questions on the user's screen, and the user answers the questions.

[2172] Step 10:

[2173] The user inputs the answer to the question, and the terminal sends it to the server as an HTTP request. This request contains data that includes the answer.

[2174] Step 11:

[2175] The server analyzes the received answers and judges whether they are correct or not. The answer analysis module then calculates the driver's proficiency and returns the evaluation result to the user's device as an HTTP response.

[2176] Step 12:

[2177] The device displays the received evaluation results on the user's screen, allowing the user to check their self-evaluation results on their smartphone screen.

[2178] Step 13:

[2179] The server evaluates the user's proficiency based on past test results and detects low proficiency in a specific area (e.g., robot safety operating procedures).

[2180] Step 14:

[2181] The server generates a follow-up message for the area of ​​low proficiency and sends it to the user's device as an HTTP response. For example, a follow-up message such as "Please restudy the specific area. For details, please refer to the system manual" is generated.

[2182] Step 15:

[2183] The device displays the received follow message on the user's screen, allowing the user to receive specific study instructions on the smartphone screen.

[2184] 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.

[2185] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The specific system configuration combines a machine learning model using a dataset from the Class 2 driver's license exam with an emotion engine to achieve an educational effect that takes into account the driver's emotional state.

[2186] Question and Answer System

[2187] 1. Enter and submit your question

[2188] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[2189] Terminal: Sends the entered question to the server as an HTTP request.

[2190] 2. Receiving and analyzing questions

[2191] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[2192] 3. Emotion Recognition and Answer Generation

[2193] Server: The emotion engine receives emotion data (e.g., facial recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[2194] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[2195] 4. Returning and displaying answers

[2196] Server: The generated answer is sent back to the user's device as an HTTP response.

[2197] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[2198] Test Generation and Examination Systems

[2199] 1. Submit a test request

[2200] User: The driver selects "I want to take the test" in the learning application.

[2201] Device: Sends a test request to the server as an HTTP request.

[2202] 2. Generate and submit test questions

[2203] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the best route when picking up a passenger using a taxi."

[2204] Server: Returns the generated test questions to the user's device as an HTTP response.

[2205] 3. Display the test screen and submit your answers

[2206] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver enters the answers and sends them to the server as HTTP requests.

[2207] 4. Analyzing responses and sending results

[2208] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[2209] Server: Returns the evaluation results to the user device as an HTTP response.

[2210] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[2211] Follow-up message system

[2212] 1. Proficiency Assessment

[2213] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[2214] 2. Create and send a follow message

[2215] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[2216] 3. Displaying follow messages

[2217] Server: Sends the generated follow message to the user's device as an HTTP response.

[2218] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[2219] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[2220] The processing flow will be explained below.

[2221] Question and Answer System

[2222] Step 1:

[2223] User: Opens the learning app and types the question, "How often do I need to renew my driver's license?"

[2224] Step 2:

[2225] Terminal: Sends the entered question to the server as an HTTP request.

[2226] Step 3:

[2227] Server: Passes the received question to the chatbot module and analyzes the question text. Specifically, it performs morphological analysis of the question content using natural language processing (NLP) to extract meaning.

[2228] Step 4:

[2229] Server: The emotion engine receives emotion data (e.g., facial expression recognition, voice analysis) from the user device and analyzes the driver's emotional state.

[2230] Step 5:

[2231] Server: Generates an answer based on the question content interpreted using a generative AI model (e.g., GPT-3) and the analysis results of the emotion engine. If the driver shows signs of anxiety, the server adds additional information such as "Don't worry, the process is simple" to the standard answer, "Your driver's license generally needs to be renewed every five years."

[2232] Step 6:

[2233] Server: The generated answer is sent back to the user's device as an HTTP response.

[2234] Step 7:

[2235] Terminal: The received answer is displayed on the user's screen. The driver can see the answer, "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[2236] Test Generation and Examination Systems

[2237] Step 1:

[2238] User: Selects "I want to take a test" in a learning app.

[2239] Step 2:

[2240] Device: Sends a test request to the server as an HTTP request.

[2241] Step 3:

[2242] Server: Based on the received request, the test generation module uses the generative AI model to automatically generate test questions, such as "Choose the optimal route when picking up a passenger using a taxi."

[2243] Step 4:

[2244] Server: Returns the generated test questions to the user's device as an HTTP response.

[2245] Step 5:

[2246] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions. The driver types in the answers.

[2247] Step 6:

[2248] Terminal: The answers entered by the driver are sent to the server as an HTTP request.

[2249] Step 7:

[2250] Server: The answer analysis module analyzes the received answers and determines whether they are correct. Using a generative AI model, the appropriateness of the answers is evaluated and the driver's proficiency is calculated.

[2251] Step 8:

[2252] Server: Returns the evaluation results to the user device as an HTTP response.

[2253] Step 9:

[2254] Terminal: The results are displayed on the user's screen, allowing the driver to check the results of their self-assessment.

[2255] Follow-up message system

[2256] Step 1:

[2257] Server: The answer analysis module evaluates the driver's proficiency based on past test results and detects low proficiency in specific areas (e.g., road traffic laws).

[2258] Step 2:

[2259] Server: The emotion engine analyzes the driver's emotional state and adjusts the content of follow-up messages accordingly. While the standard message might say, "Please review the basics of road traffic laws. Additional learning materials are available here," if the driver shows signs of stress, the message might include additional information like, "Don't rush, take your time to review."

[2260] Step 3:

[2261] Server: Sends the generated follow message to the user's device as an HTTP response.

[2262] Step 4:

[2263] Device: The received follow-up message is displayed on the user's screen. The driver can receive specific learning instructions such as, "Please review the basics of the Road Traffic Act. Take your time and don't rush."

[2264] In this way, by combining a generative AI model with an emotion engine, the system can provide personalized training while taking into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a more understandable and less stressful learning environment for drivers.

[2265] Example 2

[2266] 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."

[2267] Conventional driver training systems often provide uniform training without considering the individual needs or emotional state of each learner. Furthermore, detailed feedback based on test results is not provided, potentially reducing learning effectiveness. Furthermore, training utilizing emotional data has not been implemented, making it impossible to reduce learner stress. Therefore, a system that provides personalized training while taking into account the learner's emotional state was needed.

[2268] The identification processing by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means. In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending an individual follow-up message based on the proficiency evaluation, and means for analyzing emotion data from the user terminal and adding additional information based on the emotion to the answers generated by the model. This enables optimal education that takes into account the user's emotional state, thereby improving the effectiveness of the education.

[2269] "Machine learning" is a technology that allows a computer to analyze a given data set and automatically learn patterns and rules based on that data.

[2270] The "Class 2 Driver's License Examination Dataset" is a large-scale data set that includes questions and answers related to the Class 2 Driver's License Examination, as well as their trends and characteristics.

[2271] A "generative AI model" is a model that uses artificial intelligence to generate content such as text and images, and specifically includes natural language processing and generation modules.

[2272] A "chatbot" is a program that can interact with users and respond to their questions via text or voice.

[2273] "Emotion data" is data that indicates the user's emotional state, and examples include facial expression data and voice tone.

[2274] An "emotion engine" is an algorithm or system that analyzes emotional data to detect and evaluate a user's emotional state.

[2275] "Proficiency" is a measure of how well a driver understands or has mastered a particular subject or task.

[2276] A "follow-up message" is a message such as instruction or encouragement that is generated based on the user's learning progress or emotional state.

[2277] An "HTTP request" is a request for data sent by a client to a server, and is a protocol used by web browsers and applications to request data from a server.

[2278] An "HTTP response" is a data response that a server returns in response to a client's HTTP request, and includes requested data and status information.

[2279] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. Specific embodiments of the system are as follows.

[2280] Hardware and software used

[2281] This system uses the following hardware and software:

[2282] Server: A computer that performs major processing. It is equipped with a high-performance processor (e.g., Intel Xeon) and a large amount of memory (e.g., 64 GB or more).

[2283] Device: The device used by the driver, such as a smartphone or tablet (e.g., Android or iOS).

[2284] Generative AI model: An artificial intelligence model that generates text, for example using an advanced natural language processing model such as GPT-3.

[2285] Emotion engine: Software for analyzing user emotion data. Examples include Azure Cognitive Services' emotion recognition API.

[2286] NLP engine: Natural language processing software to analyze the question text, for example using spaCy or NLTK.

[2287] System processing flow

[2288] Answers to questions

[2289] A user opens a learning application and inputs a question, for example, "How often do I need to renew my driver's license?" The device sends this question to the server as an HTTP request. The server passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text. The server then uses an emotion engine to analyze emotion data (e.g., facial recognition and voice analysis data) from the user's device and determine the user's emotional state.

[2290] A generative AI model (e.g., GPT-3) is used to generate the optimal answer based on the analyzed question content and data from the emotion engine. For example, a standard answer such as "Your driver's license generally needs to be renewed every five years" is generated. If the user expresses anxiety, the server adds additional information such as "Don't worry, the process is simple."

[2291] The generated answer is then returned to the user terminal as an HTTP response, and the terminal displays the received answer on the user screen.

[2292] Test Generation and Taking

[2293] When a user selects "I want to take a test" in the learning application, the device sends a test request to the server. Based on the received request, the server uses a generative AI model to automatically generate test questions. For example, a question such as "Select the optimal route when picking up a passenger by taxi" is generated.

[2294] The generated test questions are sent back to the user's device, where the user can check them on the screen and enter their answers. The device then sends these answers to the server, which uses an answer analysis module to determine whether the answers are correct or incorrect and evaluate the user's proficiency. The results are then sent back to the user's device and displayed.

[2295] Follow message generation

[2296] The server evaluates the user's proficiency based on past test results and detects low proficiency in specific areas. It then uses an emotion engine to analyze the user's emotional state and generates appropriate follow-up messages based on that. For example, it generates specific learning instructions such as, "Please review your basic knowledge of road traffic laws. Take your time and review them carefully."

[2297] The follow message thus generated is sent to the user's terminal, and the user can check it on the screen.

[2298] Examples of concrete examples and prompts

[2299] For example, if a user asks, "How often do I need to renew my driver's license?", the following prompt is sent to the generative AI model:

[2300] User: "How many years do I need to renew my driver's license?"

[2301] Generative AI model: "Driver's licenses generally need to be renewed every five years."

[2302] Add emotion: "Don't worry, it's a simple process."

[2303] In this way, by combining a generative AI model with an emotion engine, this system is able to provide optimal training that takes into account the driver's emotional state, improving the efficiency and effectiveness of training and providing a learning environment that is easier for drivers to understand and less stressful.

[2304] The flow of the identification process in the second embodiment will be described with reference to FIG.

[2305] Step 1:

[2306] User: A driver opens a learning application and types the question, "How often do I need to renew my driver's license?"

[2307] Input: Keyboard input data.

[2308] Output: Question text data.

[2309] What happens: The user types a question into a text input field in the application and presses the Enter key to confirm the question.

[2310] Step 2:

[2311] Terminal: Sends the entered question to the server as an HTTP request.

[2312] Input: Question text data.

[2313] Output: HTTP request packet.

[2314] Specific operation: Converts text data into HTTP packets and sends them over the network to the server.

[2315] Step 3:

[2316] Server: Passes the received question to the chatbot module, which uses an NLP engine to perform morphological analysis and extract the meaning of the text.

[2317] Input: HTTP request packet, text data.

[2318] Output: Parsed question semantic data.

[2319] What it does: The server decodes the HTTP packets, passes the text data to an NLP engine (e.g., spaCy or NLTK), and performs morphological analysis to extract the meaning of the question.

[2320] Step 4:

[2321] Server: Uses the emotion engine to receive and analyze emotion data (e.g., facial recognition, voice analysis data) from the user device.

[2322] Input: Emotion data, e.g., image or audio files.

[2323] Output: User's emotional state data.

[2324] Specific operation: Emotion data is passed to an emotion engine (e.g., Azure Cognitive Services emotion recognition API) for analysis, and the user's emotional state is output as numerical data.

[2325] Step 5:

[2326] Server: Using a generative AI model, it generates answers based on the analyzed question content and sentiment data.

[2327] Input: Question semantic data, emotional state data.

[2328] Output: Answer data in text format.

[2329] What it does: The semantic data and emotional state data of the question are incorporated into a prompt sentence, which is then sent to a generative AI model (e.g., GPT-3) to generate an answer, such as, "Your driver's license generally needs to be renewed every five years. Don't worry, it's a simple process."

[2330] Step 6:

[2331] Server: The generated answer is sent back to the user's device as an HTTP response.

[2332] Input: Textual response data.

[2333] Output: HTTP response packet.

[2334] Specific operation: The response data is formatted as an HTTP response and sent to the user terminal via the network.

[2335] Step 7:

[2336] Terminal: Displays the received answer on the user's screen.

[2337] Input: HTTP response packet.

[2338] Output: Screen display data.

[2339] What it does: It extracts text data from the HTTP response and displays it in a display field in the application. The user can see the answer on the screen: "Driver's licenses generally need to be renewed every five years. Don't worry, the process is simple."

[2340] Step 8:

[2341] User: The driver selects "I want to take the test" in the learning application.

[2342] Input: A test-taking request.

[2343] Output: Test attempt selection data.

[2344] Specific behavior: A user clicks the "I want to take a test" button in the application to generate a test request.

[2345] Step 9:

[2346] Device: Sends a test request to the server as an HTTP request.

[2347] Input: Test attempt selection data.

[2348] Output: HTTP request packet.

[2349] Specific operation: The test selection data is converted into an HTTP request format and sent to the server via the network.

[2350] Step 10:

[2351] Server: Based on the received request, the server automatically generates test questions using a generative AI model.

[2352] Input: HTTP request packet.

[2353] Output: The generated test question data.

[2354] Specific operation: The request content is converted into a prompt for the generative AI model, and test questions are automatically generated by the generative AI model (e.g., GPT-3). For example, questions such as "Choose the optimal route when picking up a passenger by taxi" are generated.

[2355] Step 11:

[2356] Server: Returns the generated test questions to the user's device as an HTTP response.

[2357] Input: Generated test question data.

[2358] Output: HTTP response packet.

[2359] Specific operation: The generated test questions are formatted as an HTTP response and sent to the user's device.

[2360] Step 12:

[2361] Terminal: Displays test questions on the user's screen and allows the driver to answer the questions.

[2362] Input: HTTP response packet.

[2363] Output: Screen display data.

[2364] Specific behavior: Analyzes received test questions, displays them on the screen, and provides an interface for users to enter answers.

[2365] Step 13:

[2366] User: The driver enters answers to the test questions and sends them to the server.

[2367] Input: User response data.

[2368] Output: HTTP request packet.

[2369] Specific operation: Converts the user's input answer into HTTP request format and sends it to the server.

[2370] Step 14:

[2371] Server: The answer analysis module analyzes the received answers, determines whether they are correct or incorrect, and evaluates the level of proficiency.

[2372] Input: User response data.

[2373] Output: Evaluation result data.

[2374] Specific operation: The answers are passed to the analysis module, which determines whether they are correct or incorrect and generates an evaluation result. Using the generative AI model, it also properly evaluates non-standard answers.

[2375] Step 15:

[2376] Server: Returns the evaluation results to the user device as an HTTP response.

[2377] Input: Evaluation result data.

[2378] Output: HTTP response packet.

[2379] Specific operation: The evaluation results are converted into HTTP response format and sent to the user's terminal.

[2380] Step 16:

[2381] Terminal: The evaluation results are displayed on the user's screen.

[2382] Input: HTTP response packet.

[2383] Output: Screen display data.

[2384] Specific operation: The received evaluation results are displayed on the screen, allowing the user to check their own proficiency level.

[2385] (Application example 2)

[2386] 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."

[2387] Conventional driver training systems did not take into account the emotional state of the driver, resulting in ineffective learning and testing. Furthermore, with factory robots, training and proficiency management could not be carried out efficiently, limiting the improvement of robot performance. This led to robot operation errors and slower work speeds, affecting factory production efficiency.

[2388] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[2389] In this invention, the server includes means for performing machine learning using a dataset of a Class 2 driver's license examination, means for responding to questions from users using a model obtained by the machine learning, means for automatically generating test questions using the model, means for analyzing the results of the test answered by the user and evaluating the user's proficiency, means for generating and sending individual follow-up messages based on the proficiency evaluation, and means for analyzing the user's emotional data using an emotion engine and adding additional information to the answers based on the analysis results, thereby enabling efficient education and proficiency management that takes emotional states into account.

[2390] The "Class 2 Driver's License Examination Dataset" is a collection of data that includes the test questions, answers, and related information required to obtain a driver's license.

[2391] "Machine learning" is a technology that allows computers to automatically learn from data and improve their ability to perform specific tasks.

[2392] A "model" is a mathematical or statistical representation for data analysis obtained through machine learning.

[2393] "Users" refers to drivers and factory workers who use this system.

[2394] A "means for responding to questions" is a method or system for generating and replying to questions from users using natural language.

[2395] "Means for automatically generating test questions" refers to a method for automatically creating test questions based on a pre-prepared dataset or algorithm.

[2396] "Means for analyzing test results" refers to a method or system for analyzing the percentage of correct answers to test questions answered by users and their trends.

[2397] A "proficiency assessment tool" is a method for assessing a user's skill or knowledge level based on test results.

[2398] The "means for generating and sending follow-up messages" is a method for creating and sending additional guidance and encouraging messages based on the user's proficiency assessment.

[2399] An "emotion engine" is a technology that analyzes a user's emotional data and evaluates their emotional state.

[2400] "Emotion data" refers to data that indicates the emotional state of a user obtained from facial expressions, voice, and other biological signals.

[2401] This invention relates to a driver training system that combines a generative AI model and an emotion engine on a communication platform operated by a taxi company. The method for realizing this system is described in detail below.

[2402] 1. Program Generation

[2403] The program in this system basically works as follows:

[2404] Train a machine learning model using a dataset from the Class 2 license examination.

[2405] The trained model is used to generate answers to questions posed by users.

[2406] Test questions are automatically generated using a generative AI model and provided to users.

[2407] The results of the test answered by the user are analyzed to evaluate their level of proficiency.

[2408] The emotion engine is used to analyze users' emotional data, and additional information is added to replies and follow-up messages based on the analysis results.

[2409] 2. System Configuration

[2410] Hardware

[2411] Server: A central location that receives, analyzes, and generates data. It is used to train and run models and analyze sentiment data.

[2412] Device (smartphone, tablet, etc.): A device operated by a user, used to enter questions, take tests, and receive feedback.

[2413] Factory robots: Devices that operate in factories and undergo learning and testing.

[2414] software

[2415] Natural Language Processing (NLP) models are used to analyze user questions.

[2416] Generative AI models (e.g., GPT-3): Used to generate answers to questions and automatically generate test questions.

[2417] Emotion engine: Analyzes the user's facial expressions and voice data to assess their emotional state.

[2418] HTTP request / response: A protocol for sending and receiving data between a server and a device.

[2419] 3. Specific examples of program processing

[2420] Question-answering system

[2421] User: Ask the robot, "What's the next step?"

[2422] Terminal: Sends the entered question to the server.

[2423] Server: Analyzes the question using an NLP model and generates an answer using a generative AI model. An answer such as "The next task is to connect part A to part B" is generated.

[2424] Server: The sentiment engine analyzes and adds additional information to the answer indicating that the question is easy.

[2425] Terminal: Display the generated answer to the user.

[2426] Test Generation and Evaluation System

[2427] User: The robot selects "I want to take the test."

[2428] Server: The generative AI model generates test questions such as "Select the steps to correctly connect the parts."

[2429] Terminal: Displays test questions on the user's screen and receives answers.

[2430] Server: Analyzes the answers and assesses proficiency.

[2431] Server: Based on the evaluation results, a follow-up message such as "The next step is this procedure" is generated, and if the emotion engine indicates impatience in its analysis, additional information such as "Don't rush, take your time to check it out" is added.

[2432] Device: View follow messages.

[2433] 4. Examples of prompts

[2434] Specifically, the following prompts are used:

[2435] "Question: What's the next step?\nEmotion: Relaxed\nAnswer:"

[2436] "Question: Please explain the steps to connect parts correctly.\nEmotion: stressed\nAnswer: First, prepare part A, then attach it to part B. Take your time and check carefully."

[2437] In this way, by combining a generative AI model with an emotion e...

Claims

1. A means for performing machine learning using a dataset of the second-class license examination; a means for responding to questions from a driver using the model obtained by machine learning; means for automatically generating test questions using the model; A means for analyzing the results of the test answered by the driver and evaluating the driver's proficiency; means for generating and sending a personalized follow-up message based on the proficiency assessment; A system including:

2. The system of claim 1 , wherein the system uses a chatbot to accept questions from a driver, uses the model to answer questions, and provides the answers to the driver.

3. 2. The system according to claim 1, further comprising means for identifying an area in which the driver has low proficiency from the answers to the generated test questions, and for providing questions related to that area with a focus on the area.

Images