System

Abstract

A system is provided.SOLUTION: A system for predicting traffic accidents, the system comprising: means for generating a prediction map of traffic accidents; means for analyzing the prediction map using a generative AI model to generate a prediction of traffic accidents; means for plotting the prediction map on a map; and means for distributing the generated prediction map to a plurality of devices in real time.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] Conventional traffic accident information was only provided as local location information, and most of it was presented as text, making it difficult to grasp the detailed trends of traffic accident occurrence. Furthermore, police station reports were mainly text-based and did not include map information, making it difficult to obtain useful information that could lead to the prediction and prevention of traffic accidents. Furthermore, local governments' mapping and posting of images of fatal accident information was limited to certain areas, making it difficult to grasp the overall accident trends. There is a need to improve this current situation and provide a system that can efficiently predict and prevent traffic accidents from a holistic perspective, thereby preventing traffic accidents before they occur. [Means for solving the problem]

[0005] The present invention provides a traffic accident forecasting system that includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, and a data distribution means for distributing the generated prediction map to multiple devices in real time. Specifically, the data collection means acquires traffic accident information from the police and prosecutors, and the analysis means predicts the risk of traffic accidents by learning from past traffic accident data and weather data. The data distribution means distributes the prediction map via an API or WebSocket, and the mapping means visually displays high-risk locations using GIS, thereby providing predictive information to devices such as automobile navigation systems and providing users with practical accident prevention information.

[0006] The "traffic accident forecast system" is a system that collects and analyzes traffic accident data, weather data, and road surface condition data, and predicts and displays the risk of accidents occurring.

[0007] "Data collection means" refers to functions and devices for acquiring traffic accident data, and is responsible for collecting information from the police and prosecutors.

[0008] The "additional data acquisition means" is a means for acquiring additional data that may affect the occurrence of a traffic accident, such as weather data and road surface condition data.

[0009] "Data cleaning methods" are processes for preparing collected data in an analyzable format, such as filling in missing values, removing unnecessary data, and converting it into a unified format.

[0010] "Analysis means" refers to functions or devices that analyze preprocessed data using a generative AI model and predict traffic accidents.

[0011] A "generative AI model" is an artificial intelligence model that learns from past data and generates predictive results.

[0012] "Mapping means" refers to a function or device that visually displays the analysis results on a map.

[0013] The "data distribution means" is a means for distributing the generated predictive map to multiple devices in real time.

[0014] A "GIS (Geographic Information System)" is a system that handles, analyzes, and displays geospatial data.

[0015] An "automobile navigation system" is a device installed in an automobile that displays route guidance and traffic accident prediction information. [Brief explanation of the drawings]

[0016] [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

[0017] 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.

[0018] First, the terms used in the following description will be explained.

[0019] 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).

[0020] 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.

[0021] 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.

[0022] 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.

[0023] 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."

[0024] [First embodiment]

[0025] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0026] 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.

[0027] 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).

[0028] 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.

[0029] 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.

[0030] 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.

[0031] 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.

[0032] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0033] 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.

[0034] 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.

[0035] 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.

[0036] 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."

[0037] The traffic accident forecasting system of the present invention is implemented as follows.

[0038] Server Processing

[0039] Data collection

[0040] The server collects traffic accident data from police, prosecutors, local governments, etc. via API or database connection. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0041] Data Preprocessing

[0042] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[0043] Analysis by generative AI models

[0044] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[0045] Predictive Map Generation

[0046] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[0047] Real-time streaming

[0048] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[0049] Terminal handling

[0050] Receiving information

[0051] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0052] Information display

[0053] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[0054] Alert delivery

[0055] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0056] User Action

[0057] Information confirmation

[0058] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[0059] behavior adjustment

[0060] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[0061] Alert Response

[0062] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0063] A traffic accident forecast system is realized by this series of processes.

[0064] The processing flow will be explained below.

[0065] Server Processing

[0066] Step 1:

[0067] The server obtains traffic accident data from police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the status of the victims.

[0068] Step 2:

[0069] The server retrieves weather and road condition data from the Japan Meteorological Agency and weather forecast sites, including date, time, temperature, precipitation, wind speed, etc.

[0070] Step 3:

[0071] The server stores the collected traffic accident data and weather data in a database and manages them centrally.

[0072] Step 4:

[0073] The server uses data cleaning techniques to fill in missing values, remove unnecessary information, and standardize formats, for example, filling in missing latitude and longitude information in accident data.

[0074] Step 5:

[0075] The server inputs the preprocessed data into the generative AI model, which then learns traffic accident occurrence patterns.The AI ​​model then analyzes past data and learns the risk of accidents occurring under specific conditions.

[0076] Step 6:

[0077] The server uses a trained generative AI model to input the day's weather data and predict the risk of traffic accidents.

[0078] Step 7:

[0079] The server plots the prediction results on a map using a GIS (geographic information system), which visually displays high-risk locations.

[0080] Step 8:

[0081] The server delivers the created predictive maps to various devices in real time via API or WebSocket.

[0082] Terminal handling

[0083] Step 1:

[0084] A terminal (e.g., a car navigation system) receives forecast information sent from the server via an API.

[0085] Step 2:

[0086] The terminal integrates the received forecast data with map data and stores it in its internal memory.

[0087] Step 3:

[0088] The terminal displays risk areas on the map in different colors (e.g., red), allowing the user to visually identify the risk areas.

[0089] Step 4:

[0090] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[0091] User Action

[0092] Step 1:

[0093] The user checks the traffic accident prediction information displayed on the device, which allows them to understand risk areas and high-risk time periods.

[0094] Step 2:

[0095] The user can adjust their driving route and speed based on the displayed forecast information, for example, by choosing a detour route to avoid risk areas.

[0096] Step 3:

[0097] Users will be able to drive more carefully when they receive alerts and warnings from their devices, for example by slowing down and remaining vigilant when an alert is issued.

[0098] Example 1

[0099] 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."

[0100] Conventional traffic accident prevention systems have had difficulty predicting traffic accidents in advance. In particular, they were unable to adequately consider external factors such as weather and road conditions, limiting the accuracy of their predictions. Furthermore, they were unable to provide real-time information or visually indicate high-risk areas, preventing them from providing useful information to users.

[0101] 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.

[0102] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model to learn traffic accident occurrence patterns and predict traffic accidents, a mapping means for plotting the prediction results on a map using a GIS, and a data distribution means for distributing the generated prediction map to multiple devices in real time via an API or WebSocket. This makes it possible to predict traffic accidents with high accuracy and provide useful information to users in real time.

[0103] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides the results to users.

[0104] A "data collection means" is a method or device for obtaining traffic accident data.

[0105] "Additional data acquisition means" is a method or device for collecting weather and road condition data.

[0106] A "data cleaning means" is a method or device for preprocessing collected data, such as filling in missing values ​​and removing outliers.

[0107] "Analysis means" refers to a method or device that uses a generative AI model to analyze preprocessed data, learn patterns of traffic accident occurrence, and make predictions.

[0108] A "generative AI model" is an artificial intelligence model that learns from collected data and analyzes and predicts traffic accident patterns.

[0109] "Mapping means" refers to a method or device for plotting the analysis results on a map using a geographic information system (GIS).

[0110] "Data distribution means" refers to a method or apparatus for distributing the generated predictive maps to multiple devices in real time via an API or WebSocket.

[0111] The traffic accident forecasting system of the present invention is constructed to accurately predict the occurrence of traffic accidents and provide the information to users in real time. This system includes a data collection means, an additional data acquisition means, a data cleaning means, an analysis means, a mapping means, and a data distribution means.

[0112] Server Processing

[0113] Data collection

[0114] The server collects traffic accident data from police and local governments via API. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast websites. As a specific example, the server obtains traffic accident data from a city's police station from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0115] Data Preprocessing

[0116] The server cleanses the collected data, fills in missing and outlier values, and converts it into a unified format. It also accurately maps date, time, and location information. For example, the server fills in missing latitude and longitude information in traffic accident data, integrates it with weather data, and unifies the date and time format.

[0117] Analysis by generative AI models

[0118] The server inputs the preprocessed data into a generative AI model and analyzes traffic accident occurrence patterns. The generative AI model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model and makes it learn the pattern that "accidents occur frequently during the daytime on rainy days." An example of a prompt is "Analyze the risk of traffic accidents occurring on rainy days based on past traffic accident data."

[0119] Predictive Map Generation

[0120] The server plots the prediction results on a map using a GIS (geographic information system) based on the prediction results from the generative AI model. High-risk locations are displayed in a different color (for example, red). As a specific example, the server displays areas in a city where traffic accidents are frequent in red on a map based on the output of the generative AI model.

[0121] Real-time streaming

[0122] The server distributes the generated predictive maps to various devices (e.g., automobile navigation systems) in real time. Distribution is typically performed via API or WebSocket. For example, the server transmits the generated predictive maps to the navigation system's API in real time.

[0123] Terminal handling

[0124] Receiving information

[0125] The terminal (for example, a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0126] Information display

[0127] The device displays the received forecast information on a map or in an interface. It provides a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data it receives.

[0128] Alert delivery

[0129] The device generates an alert under certain conditions to warn the user. Specifically, it provides audio guidance and visual warnings when approaching a high-risk area. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0130] User Action

[0131] Information confirmation

[0132] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[0133] behavior adjustment

[0134] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[0135] Alert Response

[0136] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0137] In this way, the traffic accident forecasting system can predict the risk of traffic accidents occurring in real time and provide users with immediately useful information.

[0138] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0139] Step 1: Data collection

[0140] Input: Requests to police and local government APIs, the Japan Meteorological Agency, and weather forecast site APIs

[0141] Output: Traffic accident data and weather data are stored on the server.

[0142] Specific operation: The server sends a request to an endpoint such as "https: / / example-police-api.com / data / accidents" or "https: / / example-weather-api.com / data / weather", receives the JSON-formatted data as a response, and then saves it to local storage as " / data / accidents / 2023-01-01_to_2023-07-31.json" or " / data / weather / 2023-01-01_to_2023-07-31.json".

[0143] Step 2: Data Preprocessing

[0144] Input: Traffic accident data and weather data collected in Step 1

[0145] Output: Preprocessed data in a unified format

[0146] What it does: The server cleanses the data and imputes missing and outlier values. For example, it imputes the mean or median for fields marked as "NA." Next, it converts all data to "YYYY-MM-DD HH:MM:SS" format and imputes missing latitude and longitude information using address information.

[0147] Step 3: Analysis by generative AI model

[0148] Input: Preprocessed data from step 2

[0149] Output: Traffic accident prediction results

[0150] Specific operation: The server generates a file called "ai_model_input.csv" and inputs it into the generative AI model. The AI ​​model runs the "train_model()" method and learns patterns from past accident data, such as "accidents tend to occur more frequently during the daytime on rainy days." The prediction results are saved as "predicted_accidents.csv."

[0151] Step 4: Generate a prediction map

[0152] Input: Prediction results obtained in step 3

[0153] Output: A map with risk areas plotted

[0154] Specific operation: The server reads "predicted_accidents.csv" and plots risk areas in red on a map using a GIS (geographic information system). It then executes the function "plot_map('predicted_accidents.csv')" to generate a map image file showing high-risk areas.

[0155] Step 5: Real-time delivery

[0156] Input: Prediction map generated in step 4

[0157] Output: Predictive maps sent to navigation systems via API or WebSocket

[0158] Specific operation: The server executes "send_data_to_api('https: / / example-navigation-api.com / predicted_map', 'predicted_map_data')" and sends the data to the navigation system.

[0159] Step 6: Receiving information (terminal)

[0160] Input: Predictive map data sent from the server

[0161] Output: Predictive map data stored in the device's internal memory

[0162] Specific operation: The device periodically sends a request to the server, executes "fetch(' / predicted_map')" to receive data, and stores it in its internal memory.

[0163] Step 7: Display Information (Terminal)

[0164] Input: Predicted map data received in step 6

[0165] Output: Risk areas displayed in a visually friendly interface

[0166] Specific operation: The device executes the function "display_map(data)" and displays the risk area in red on the map.

[0167] Step 8: Alert delivery (terminal)

[0168] Input: Your device's current location and the risk area information displayed in step 7

[0169] Output: Audio prompts and visual warning alerts

[0170] Specific operation: The device executes "if(current_location in high_risk_area): trigger_alert('There is a risk of an accident at the next intersection. Be careful.')" and issues an audio alert.

[0171] Step 9: User Confirmation

[0172] Input: Risk Area information displayed in step 7

[0173] Output: User who understands risk area and time period

[0174] Specific operation: The user checks the navigation screen and identifies the location of the risk area displayed in red.

[0175] Step 10: Behavioral Adjustment (User)

[0176] Input: Risk area information confirmed in step 9

[0177] Output: Adjusted driving route and speed

[0178] Specific operation: The user follows the re-route guidance provided by the navigation system and selects a safe detour.

[0179] Step 11: Alert response (user)

[0180] Input: The alert received from the terminal in step 8

[0181] Output: More careful driving

[0182] Specific actions: The user hears the alert sound, slows down, and drives safely.

[0183] Through each step in this way, the traffic accident forecasting system is able to predict the risk of traffic accidents occurring in real time and provide users with useful information immediately.

[0184] (Application example 1)

[0185] 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."

[0186] Traffic accidents are an unavoidable problem that seriously impacts human life and property. Even now, as autonomous vehicles become more widespread, the risk of traffic accidents still exists, posing a challenge to the reliability of autonomous driving systems. Conventional navigation systems only provide static map and traffic information, and are unable to predict and respond immediately to changing traffic accident risks in real time. Therefore, there is a need for a method to predict traffic accident risks in real time and efficiently incorporate them into autonomous driving systems.

[0187] 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.

[0188] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for displaying the prediction map information on the navigation system of the autonomous vehicle, an alert generation means for issuing an audio guide or a visual warning when approaching the risk area, and an interface means for allowing the user to check the risk area and adjust their behavior accordingly. This makes it possible to predict traffic accident risks in real time and efficiently reflect the results in the autonomous vehicle.

[0189] The "traffic accident forecast system" is a system that predicts the risk of traffic accidents and encourages appropriate responses.

[0190] "Data collection means" refers to a means for collecting traffic accident data.

[0191] The "additional data acquisition means" is a means for collecting weather and road surface condition data.

[0192] "Data cleaning means" refers to means for preprocessing collected traffic accident data and additional data to make them suitable for analysis.

[0193] "Analysis means" refers to a means for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[0194] "Mapping means" is a means for plotting the analysis results on a map.

[0195] The "data distribution means" is a means for distributing the generated predictive map to multiple devices in real time.

[0196] The "means for displaying on a navigation system" refers to a means for displaying the predicted map information on a navigation system of an autonomous driving vehicle.

[0197] The "alert generating means" is a means for issuing an audio guide or a visual warning when approaching the risk area.

[0198] The "interface means" is a means by which users can check risk areas and adjust their behavior accordingly.

[0199] As an embodiment of the present invention, a traffic accident prediction system includes the following steps: A server collects traffic accident data and acquires weather and road condition data as additional data; the data is preprocessed by a data cleaning means and analyzed using a generative AI model; the analyzed results are plotted on a map by a mapping means; and the predicted map is distributed to multiple devices in real time by a data distribution means.

[0200] The server uses the following specific hardware and software: A REST API is used to collect data, and data collection scripts are executed using the Python language. Weather and road condition data is obtained from external data sources via the API. For data preprocessing, the Python Pandas library is used to fill in missing values ​​and clean the data. For the generative AI model, machine learning libraries such as TensorFlow and PyTorch are used to train the model on past traffic accident data and weather data. For plotting on a map, the Folium library is used, and the prediction results are displayed on the map via the Google Maps API.

[0201] The terminal side acquires the predictive map information using a means to display it on the navigation system and provides the information visually to the user. This interface has an alert generation means that issues audio guidance and visual warnings when approaching a risk area. This alert generation uses WebSocket and API to receive data distribution in real time.

[0202] The user checks the displayed predictive map information to understand the relationship between their current location and the risk area. When approaching a risk area, they take appropriate action based on audio guidance and visual warnings. The user can select a detour or adjust their driving speed based on the information displayed on the navigation system.

[0203] As a concrete example, traffic accident data and weather data are collected on a server, and analyzed by a generative AI model, resulting in the learning result that "accidents occur frequently during the daytime on rainy days." This information is input to the generative AI model as the following prompt:

[0204] Example prompt sentence:

[0205] Analyze traffic accident patterns based on traffic accident data from January 1, 2023 to July 31, 2023, and weather data for the same period. Learn the tendency for accidents to occur more frequently during the daytime on rainy days, and build a prediction model.

[0206] In this way, predicted traffic accident risk information is delivered to user terminals in real time, which can contribute to reducing traffic accident risks in actual usage environments.

[0207] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0208] Step 1:

[0209] The server collects traffic accident data. This is done using APIs or database connections provided by the police or local government. The server retrieves traffic accident data from, for example, January 1, 2023 to July 31, 2023. The input is traffic accident data, and the output is the collected raw data. This data includes the date, time, location, and details of the accident.

[0210] Step 2:

[0211] The server collects additional weather and road condition data. For this, it uses the API of the Japan Meteorological Agency or weather forecast site. For example, it obtains weather data for the same period. The input is the weather and road condition data, and the output is the collected raw data. This data includes temperature, precipitation, road condition, etc.

[0212] Step 3:

[0213] The server preprocesses the collected traffic accident data and additional data. First, it completes missing values ​​and removes inappropriate data. Next, it unifies the data format and accurately maps date, time, and location information. The input is the collected raw data, and the output is the cleansed data. Specifically, it uses the Python Pandas library to cleanse the data.

[0214] Step 4:

[0215] The server inputs the preprocessed data into a generative AI model to predict the risk of traffic accidents. The generative AI model learns accident patterns from past traffic accident data and weather data and builds a predictive model. The input is the preprocessed data, and the output is risk prediction data. Specifically, the AI ​​model is run using TensorFlow or PyTorch.

[0216] Step 5:

[0217] The server plots the predicted risk data on a map. Specifically, it uses GIS technology to color-code risk areas. The input is risk prediction data, and the output is a prediction map. Specifically, it generates a map using the Folium library and displays it on the map using the Google Maps API.

[0218] Step 6:

[0219] The server delivers the generated predicted map to the device in real time. This is done using an API or WebSocket. The input is the predicted map, and the output is real-time delivery data. Specifically, data is delivered using a web framework such as Flask.

[0220] Step 7:

[0221] The device receives predictive map information from the server via API. The input is real-time data delivered from the server, and the output is predictive map information stored inside the device. Specifically, the navigation system stores the received data in memory.

[0222] Step 8:

[0223] The device displays the received predictive map information on a map or interface. The input is the predictive map information stored in the device, and the output is the visually displayed map information. Specifically, the navigation system displays risk areas in red on the map.

[0224] Step 9:

[0225] The device generates audio guidance and visual warnings when approaching a risk area. The input is the predicted map information and current location information stored in the device, and the output is the audio guidance and visual warning. Specifically, the navigation system issues an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0226] Step 10:

[0227] The user checks the forecast information displayed on the device and understands the risk areas near their current location. The input is the visually displayed map information, and the output is the user's perception. Specifically, the driver checks the navigation screen and confirms the location of the risk area displayed in red.

[0228] Step 11:

[0229] The user adjusts the driving route and speed based on the confirmed information. The input is the risk information recognized by the user, and the output is the adjusted driving behavior. Specifically, the driver follows the navigation system's instructions to select a detour to avoid the risk area.

[0230] Step 12:

[0231] Users receive alerts from their devices and drive more carefully. The input is audio guidance and visual warnings, and the output is safe driving behavior. Specifically, when a driver hears an alert and receives information that there is a risk of an accident at the next intersection, they slow down and continue driving vigilantly.

[0232] 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.

[0233] The traffic accident prediction system of the present invention provides more effective accident prevention information by combining the user's emotion engine. This system is implemented as follows.

[0234] Server Processing

[0235] Data collection

[0236] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0237] Data Preprocessing

[0238] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[0239] Analysis by generative AI models

[0240] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[0241] Predictive Map Generation

[0242] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[0243] Real-time streaming

[0244] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[0245] Terminal handling

[0246] Receiving information

[0247] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0248] Information display

[0249] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[0250] Alert delivery

[0251] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0252] Emotion engine processing

[0253] Emotional Data Collection

[0254] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera captures the user's face, and the microphone records their voice.

[0255] Emotion analysis

[0256] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. For example, the device may determine that the user is in a "tense state" based on changes in their facial expression.

[0257] Adjusting alert content

[0258] The displayed predictive information and alert content are adjusted based on the results of emotion analysis. For example, if the user is in a tense state, the alert wording will be softer. For example, a message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" will be displayed.

[0259] User Action

[0260] Information confirmation

[0261] Users can check the traffic accident prediction information and adjusted alert messages displayed on the device, which allows them to accurately identify risk areas and high-risk time periods.

[0262] behavior adjustment

[0263] The user can adjust their driving route and speed based on the displayed forecast information and alert messages. For example, to avoid risk areas, the user can select a detour route according to the navigation system's instructions.

[0264] Alert Response

[0265] Users can drive more carefully when they receive alerts and warnings from their devices. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0266] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[0267] The processing flow will be explained below.

[0268] Server Processing

[0269] Step 1:

[0270] The server receives traffic accident data from the police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the victim's condition.

[0271] Step 2:

[0272] The server obtains weather and road condition data from the Japan Meteorological Agency and weather forecast websites via API. Weather data includes date, time, temperature, precipitation, wind speed, etc.

[0273] Step 3:

[0274] The server stores the acquired traffic accident data and weather data in a database, allowing each data item to be managed in an integrated manner.

[0275] Step 4:

[0276] The server uses data cleaning means to fill in missing values ​​in the collected data, remove unnecessary information, and standardize the format.

[0277] Step 5:

[0278] The server inputs the preprocessed data into a generative AI model, which analyzes and learns patterns of traffic accident occurrence.

[0279] Step 6:

[0280] The server uses a trained generative AI model to predict the risk of traffic accidents based on the day's weather data.

[0281] Step 7:

[0282] The server plots the prediction results on a map using GIS (geographic information system), visually displaying high-risk locations.

[0283] Step 8:

[0284] The server delivers the generated predictive maps to various devices in real time via API or WebSocket.

[0285] Terminal handling

[0286] Step 1:

[0287] The terminal receives forecast information sent in real time from the server via an API.

[0288] Step 2:

[0289] The terminal integrates the received forecast information with map data and stores it in its internal memory.

[0290] Step 3:

[0291] Based on the received data, the terminal displays risk areas on a map, color-coding them (for example, red).

[0292] Step 4:

[0293] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[0294] Emotion engine processing

[0295] Step 1:

[0296] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors.

[0297] Step 2:

[0298] The device inputs the collected data into an emotion engine to analyze the user's emotional state in real time.

[0299] Step 3:

[0300] The device adjusts the content of the displayed forecast information and alerts based on the user's emotional state as determined by the emotion engine.

[0301] User Action

[0302] Step 1:

[0303] The user checks the traffic accident prediction information and the adjusted alert message displayed on the terminal.

[0304] Step 2:

[0305] The user can adjust their driving route and speed based on the displayed forecast information and alert messages.

[0306] Step 3:

[0307] Users will receive alerts and warnings from their devices and drive more carefully.

[0308] Example 2

[0309] 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."

[0310] Conventional traffic accident forecasting systems only identify risk areas based on past traffic accident data and weather data. As a result, they do not reflect the user's emotional state, and therefore may not provide accurate alerts. In particular, when emotional states such as tension and fatigue contribute to accident risk, not taking these data into account is a serious drawback.

[0311] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0312] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for checking traffic accident prediction information and alert messages displayed by the terminal, a driving adjustment means for adjusting the driving route and speed based on the prediction information and alert messages displayed by the terminal, an emotion analysis means for collecting and analyzing user emotion data by the terminal, and an alert adjustment means for adjusting the content of the alert based on the emotion analysis results. This enables the provision of more accurate traffic accident forecasts and alerts that take the user's emotional state into consideration.

[0313] A "traffic accident forecast system" is a system that predicts the risk of traffic accidents and provides users with that information.

[0314] "Data collection means" refers to devices and methods for collecting traffic accident data.

[0315] "Additional data acquisition means" refers to devices and methods for collecting weather and road condition data.

[0316] "Data cleaning means" refers to devices and methods for organizing collected data and removing unnecessary information.

[0317] "Analysis means" refers to devices or methods for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[0318] "Mapping means" refers to a device or method for plotting the analysis results on a map.

[0319] "Data distribution means" refers to an apparatus or method for distributing the generated predictive map to multiple devices in real time.

[0320] "Terminal" refers to a device that displays prediction information and collects user emotion data.

[0321] "Emotion analysis means" refers to a device or method for collecting and analyzing user emotion data.

[0322] "Alert adjustment means" refers to a device or method for adjusting the content of an alert based on the results of emotion analysis.

[0323] "Driving adjustment means" refers to a device or method for adjusting the driving route and speed based on the forecast information and alert messages displayed on the terminal.

[0324] The traffic accident forecasting system of this invention is a system that combines user emotional information to provide more effective accident prevention information. This system functions in cooperation with three parties: a server, a terminal, and a user.

[0325] Server Processing

[0326] Data collection

[0327] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, a detailed dataset will be created by collecting traffic accident and weather data in Tokyo from January 1, 2023 to July 31, 2023.

[0328] Data Preprocessing

[0329] Collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, missing latitude and longitude information in traffic accident data is filled, and weather data is integrated to unify the date and time format.

[0330] Analysis by generative AI models

[0331] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, past accident data is input into the generative AI model, and it learns the pattern that "accidents occur frequently during the daytime on rainy days."

[0332] Predictive Map Generation

[0333] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (e.g., red). For example, based on the output of the generative AI model, areas in Tokyo with a high incidence of traffic accidents are displayed in red on a map.

[0334] Real-time streaming

[0335] The generated predictive map is distributed in real time to various devices (e.g., automobile navigation systems). Distribution is performed via API or WebSocket. For example, the generated predictive map is sent in real time to the navigation system's API.

[0336] Terminal handling

[0337] Receiving information

[0338] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. The navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0339] Information display

[0340] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface. For example, based on the data received by the navigation system, risk areas are displayed in red on a map.

[0341] Alert delivery

[0342] When approaching a high-risk area, audio guidance and visual warnings are provided, such as generating an audio alert saying, "There is an accident risk at the next intersection. Be careful."

[0343] Emotion engine processing

[0344] Emotional Data Collection

[0345] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera scans the user's face and the microphone records their voice.

[0346] Emotion analysis

[0347] The collected user data is input into an emotion engine to analyze the user's emotional state in real time, thereby determining whether the user is tense or relaxed.

[0348] Adjusting alert content

[0349] The displayed forecast information and alerts are adjusted based on the results of emotion analysis. For example, if the user is feeling nervous, a gentler message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" is used.

[0350] User Action

[0351] Information confirmation

[0352] The user checks the traffic accident prediction information and coordinated alert messages displayed on the device, which allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action.

[0353] behavior adjustment

[0354] The user can adjust their driving route and speed based on the displayed forecast information and alert messages, and can select a detour route to avoid risk areas by following the navigation system's instructions.

[0355] Alert Response

[0356] Users can drive more carefully by receiving alerts and warnings from their devices. For example, a driver may hear an alert and receive information that there is a risk of an accident at the next intersection, so they can slow down and continue driving vigilantly.

[0357] Examples of specific examples and prompts

[0358] Specific examples

[0359] The server collects traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, and analyzes the data using a generative AI model. From the analysis results, it learns patterns of accidents occurring frequently during the daytime on rainy days, and identifies and plots risk areas within Tokyo. This information is then sent to the navigation system in real time, where the user can confirm it and take appropriate measures.

[0360] Prompt Sentence Examples

[0361] "Based on traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, the generative AI model identifies areas where accidents are expected to occur frequently during the daytime on rainy days, and displays the results on a map. If the user is in a tense state, the alert message will be softened."

[0362] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[0363] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0364] Step 1: Data collection

[0365] The server collects traffic accident data from police, prosecutors, local governments, etc. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. Specifically, it uses API or database connections to obtain traffic accident and weather data for Tokyo from January 1, 2023 to July 31, 2023. It receives traffic accident data and weather data as input and builds a detailed dataset as output.

[0366] Step 2: Data Preprocessing

[0367] The server preprocesses the collected data. Specifically, it removes unnecessary information and fills in missing values. For example, it fills in missing latitude and longitude information, integrates it with weather data, and standardizes the date and time format. It receives the dataset constructed in step 1 as input and generates cleansed and unified data as output.

[0368] Step 3: Analysis by generative AI model

[0369] The server inputs the preprocessed data into the generative AI model and analyzes traffic accident occurrence patterns. Specifically, it inputs past accident data and weather data into the AI ​​model, and has it learn patterns such as "accidents occur frequently during the daytime on rainy days." It receives the preprocessed data as input and generates a predicted result of the risk of traffic accidents as output.

[0370] Step 4: Generate a prediction map

[0371] The server uses a GIS (geographic information system) to plot the prediction results on a map based on the predictions from the generative AI model. Specifically, it colors the results by, for example, showing high-risk areas in red. It receives the prediction results from the generative AI model as input and generates a visualized prediction map as output.

[0372] Step 5: Real-time delivery

[0373] The server delivers the generated predictive map to the device in real time. Specifically, it sends the predictive map to a navigation system or other device using an API or WebSocket. It receives the predictive map as input and delivers it to the device in real time as output.

[0374] Step 6: Receiving information

[0375] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. Specifically, it receives the forecast data through the API and stores it in its internal memory. It receives the forecast information from the server as input and generates the forecast information stored in its internal memory as output.

[0376] Step 7: Display information

[0377] The device displays the received forecast information on a map or interface. Specifically, it displays risk areas in red on the map to allow the user to visually recognize them. It receives forecast information stored in its internal memory as input and displays visualized forecast information as output.

[0378] Step 8: Alert Delivery

[0379] The device provides audio guidance and visual warnings when approaching a high-risk area. Specifically, it generates an audio alert saying, "There is an accident risk at the next intersection. Please be careful." It receives visualized prediction information as input and generates an alert message as output.

[0380] Step 9: Emotional Data Collection

[0381] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the camera scans the user's face and the microphone records their voice. The device receives the user's emotional data as input and collects emotional data as output.

[0382] Step 10: Sentiment Analysis

[0383] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. Specifically, it determines whether the user is tense or relaxed. It receives emotional data as input and generates analysis results as output.

[0384] Step 11: Adjust the alert content

[0385] The device adjusts the content of the alert based on the emotion analysis results. Specifically, if the user is in a tense state, it displays a message such as, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly." It receives the emotion analysis results as input and generates a tailored alert message as output.

[0386] Step 12: Verify the information

[0387] The user checks the traffic accident prediction information and the adjusted alert message displayed on the device. This allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action. The user receives visualized prediction information and alert messages as input, and checks the information as output.

[0388] Step 13: Behavioral Adjustments

[0389] The user adjusts the driving route and speed based on the displayed forecast information and alert messages. Specifically, the user selects a detour route to avoid risk areas according to the navigation instructions. The user receives confirmed information as input and performs adjusted driving behavior as output.

[0390] Step 14: Respond to alerts

[0391] Users receive alerts and warnings from their devices and drive more carefully. Specifically, when a user hears an alert, they recognize that there is a risk of an accident at the next intersection, slow down, and continue driving vigilantly. The alert message is received as input, and the behavior of driving carefully is output.

[0392] (Application example 2)

[0393] 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."

[0394] There is a need to provide automated driving vehicles and drivers with appropriate and timely information on traffic accident risks. In particular, it is necessary to more effectively promote safe driving by providing alerts and guidance that take into account the driver's emotional state. However, existing systems are insufficient in providing such comprehensive information, and integrating risk information and alerts that respond to the driver's emotional state is a challenge.

[0395] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, an emotion data collection means for collecting user emotion data, an emotion analysis means for analyzing the collected emotion data, and an alert adjustment means for adjusting the display and alert content based on the emotion analysis results. This makes it possible to provide integrated traffic accident risk information and the driver's emotional state.

[0396] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides that information to users.

[0397] The "data collection means" is a device or program that has the function of acquiring traffic accident data.

[0398] The "additional data acquisition means" is a device or program that acquires weather and road condition data.

[0399] The "data cleaning means" is a device or program that has the function of preprocessing the acquired traffic accident data and additional data.

[0400] A "generative AI model" is an artificial intelligence model that performs predictive analysis of traffic accident occurrences based on collected data.

[0401] "Analysis means" refers to a device or program that analyzes preprocessed data using a generative AI model and predicts traffic accidents.

[0402] "Mapping means" refers to a device or program that has the function of plotting the analysis results on a map.

[0403] The "data distribution means" is a device or program that has the function of distributing the generated predictive map to multiple devices in real time.

[0404] "Emotion data collection means" refers to a device or program that has the function of acquiring user emotion data.

[0405] The "emotion analysis means" is a device or program that has the function of analyzing collected emotion data and determining the user's emotional state.

[0406] The "alert adjustment means" is a device or program that has the function of adjusting the display and alert content based on the emotion analysis results.

[0407] "Traffic Accident Data" refers to information relating to past and current traffic accidents.

[0408] "Weather Data" means information regarding current and forecast weather.

[0409] "Road Condition Data" refers to information regarding current and predicted road surface conditions.

[0410] "Plotting on a map" refers to visually displaying the analysis results using a geographic information system (GIS).

[0411] "Delivering in real time" refers to immediately transmitting the generated prediction map to the user device.

[0412] "Adjusting alert content" refers to changing the content and intensity of a warning message taking into account the user's emotional state.

[0413] The traffic accident forecasting system of this invention effectively utilizes various data to provide information useful for predicting and preventing traffic accidents. This system is mainly composed of three main components: a server, a terminal, and a user. Each component and its specific processing are explained below.

[0414] Server Processing

[0415] Data collection

[0416] The server collects traffic accident data and weather and road condition data. Traffic accident data is obtained from police and local government databases, while weather and road condition data is obtained from weather information websites and the Japan Meteorological Agency. These data are collected through APIs and database connections.

[0417] Data Preprocessing

[0418] The collected data is first preprocessed using data cleaning methods, such as imputing missing values, removing unnecessary information, and standardizing the format, to prepare the dataset for analysis.

[0419] Analysis by generative AI models

[0420] The preprocessed data is input into a generative AI model to predict traffic accidents. The generative AI model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents.

[0421] Predictive map generation and delivery

[0422] The prediction results of the generative AI model are plotted on a map using GIS, generating a predictive map. This predictive map is then distributed in real time to multiple devices, such as navigation systems and smartphone applications, via a data distribution method.

[0423] Terminal handling

[0424] Receiving information

[0425] The device receives the forecast information sent from the server, typically via an API.

[0426] Information display

[0427] The received forecast information is displayed on a map or in the UI, providing a visually easy-to-understand interface that makes it easy for users to identify risk areas.

[0428] Emotional data collection and analysis

[0429] The device is equipped with a camera, microphone, and sensors to collect the user's facial expressions, voice, and vital data. This data is analyzed in real time by emotion analysis. For example, if the user is in a tense state, this information can be obtained as an analysis result.

[0430] User Action

[0431] Information confirmation

[0432] The device displays traffic accident prediction information and alert messages tailored to the user's emotional state, allowing the user to instantly identify risk areas and dangerous time periods.

[0433] behavior adjustment

[0434] Users can adjust their driving route and speed based on the displayed forecast information and alert messages, for example by choosing a detour route to avoid risk areas.

[0435] Alert Response

[0436] Users receive alerts and warnings from their devices and strive to drive safely. For example, if they receive a message such as "There is a risk of an accident at the next intersection. Please be careful," they should slow down and be vigilant.

[0437] Specific examples

[0438] The server collects traffic accident data and weather data over a certain period of time and performs data cleaning. A generative AI model is used to learn a predictive pattern, such as "high accident rates during the daytime on rainy days." A risk map is then generated using GIS and sent to the device in real time. The device receives this information and displays risk areas in red on the map. In addition, the camera captures the user's face, and a microphone collects audio data for emotion analysis. If the device determines that the user is in a tense state, it will provide an audio message saying, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly."

[0439] Prompt Sentence Examples

[0440] "This driver is on edge. What kind of caution would work in the next high-accident area?"

[0441] "The current weather is rain, and the location is XX intersection. Please predict the accident risk under these conditions."

[0442] In this way, the traffic accident forecast system incorporates the user's emotional information to provide more effective accident prevention information.

[0443] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0444] Step 1: Collecting traffic accident and weather data

[0445] The server collects traffic accident data and weather data. It obtains traffic accident data from police and local government databases, and weather data from weather information websites and the Japan Meteorological Agency. It collects data through APIs and database connections and stores it. The input is traffic accident data and weather data, and the output is raw data.

[0446] Step 2: Preprocessing the data

[0447] The server preprocesses the collected data by completing missing values, removing unnecessary information, and standardizing the format. The input is raw data, and the output is preprocessed data. Specific operations include cleaning the data, completing missing values, and correcting improper formats.

[0448] Step 3: Analysis by generative AI model

[0449] The server inputs the preprocessed data into the generative AI model to predict traffic accidents. The AI ​​model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents. The input is the preprocessed data, and the output is the traffic accident risk prediction results. Specific operations include inputting data into the AI ​​model, running the model, and obtaining the results.

[0450] Step 4: Generate prediction maps

[0451] The server uses GIS to plot the prediction results of the generative AI model on a map and generate a prediction map. The input is the traffic accident risk prediction result, and the output is the prediction map. Specifically, the server uses GIS software to overlay geographical information and risk information.

[0452] Step 5: Serving the predicted map

[0453] The server distributes the generated predictive map to terminals in real time via a data distribution means. The input is the predictive map, and the output is the data distributed to each terminal. Specifically, the server uses an API or WebSocket to send the predictive map to multiple terminals.

[0454] Step 6: Collect emotion data

[0455] To collect the user's emotional data, the device uses a camera, microphone, and sensors to acquire the user's facial expressions, voice, and vital data. The input is the user's biometric information, and the output is emotional data. Specifically, the device performs processes to capture the user's face, record their voice, and record their vital data.

[0456] Step 7: Analyze the sentiment data

[0457] The device inputs the collected emotional data into an emotion analysis means to determine the user's emotional state. The input is emotional data, and the output is the emotion analysis result. Specifically, the device performs emotion analysis using a deep learning model to determine the user's emotional state as "tense" or "calm."

[0458] Step 8: Adjust and view alerts

[0459] The device adjusts the content of the alert based on the emotion analysis results and notifies the user by display and audio. The input is the emotion analysis results and traffic accident prediction information, and the output is the adjusted alert message. Specifically, the device generates an alert message according to the user's emotional state and notifies the user by visual and audio means.

[0460] Step 9: User Behavior

[0461] The user adjusts their behavior based on the traffic accident prediction information displayed on the device and the adjusted alert message. The input is the adjusted alert message, and the output is the user's behavior. Specific actions include choosing a detour route to avoid risk areas or reducing speed.

[0462] 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.

[0463] 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.

[0464] 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.

[0465] [Second embodiment]

[0466] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0467] 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.

[0468] 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).

[0469] 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.

[0470] 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.

[0471] 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).

[0472] 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.

[0473] 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.

[0474] 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.

[0475] 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.

[0476] 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.

[0477] 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."

[0478] The traffic accident forecasting system of the present invention is implemented as follows.

[0479] Server Processing

[0480] Data collection

[0481] The server collects traffic accident data from police, prosecutors, local governments, etc. via API or database connection. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0482] Data Preprocessing

[0483] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[0484] Analysis by generative AI models

[0485] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[0486] Predictive Map Generation

[0487] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[0488] Real-time streaming

[0489] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[0490] Terminal handling

[0491] Receiving information

[0492] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0493] Information display

[0494] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[0495] Alert delivery

[0496] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0497] User Action

[0498] Information confirmation

[0499] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[0500] behavior adjustment

[0501] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[0502] Alert Response

[0503] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0504] A traffic accident forecast system is realized by this series of processes.

[0505] The processing flow will be explained below.

[0506] Server Processing

[0507] Step 1:

[0508] The server obtains traffic accident data from police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the status of the victims.

[0509] Step 2:

[0510] The server retrieves weather and road condition data from the Japan Meteorological Agency and weather forecast sites, including date, time, temperature, precipitation, wind speed, etc.

[0511] Step 3:

[0512] The server stores the collected traffic accident data and weather data in a database and manages them centrally.

[0513] Step 4:

[0514] The server uses data cleaning techniques to fill in missing values, remove unnecessary information, and standardize formats, for example, filling in missing latitude and longitude information in accident data.

[0515] Step 5:

[0516] The server inputs the preprocessed data into the generative AI model, which then learns traffic accident occurrence patterns.The AI ​​model then analyzes past data and learns the risk of accidents occurring under specific conditions.

[0517] Step 6:

[0518] The server uses a trained generative AI model to input the day's weather data and predict the risk of traffic accidents.

[0519] Step 7:

[0520] The server plots the prediction results on a map using a GIS (geographic information system), which visually displays high-risk locations.

[0521] Step 8:

[0522] The server delivers the created predictive maps to various devices in real time via API or WebSocket.

[0523] Terminal handling

[0524] Step 1:

[0525] A terminal (e.g., a car navigation system) receives forecast information sent from the server via an API.

[0526] Step 2:

[0527] The terminal integrates the received forecast data with map data and stores it in its internal memory.

[0528] Step 3:

[0529] The terminal displays risk areas on the map in different colors (e.g., red), allowing the user to visually identify the risk areas.

[0530] Step 4:

[0531] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[0532] User Action

[0533] Step 1:

[0534] The user checks the traffic accident prediction information displayed on the device, which allows them to understand risk areas and high-risk time periods.

[0535] Step 2:

[0536] The user can adjust their driving route and speed based on the displayed forecast information, for example, by choosing a detour route to avoid risk areas.

[0537] Step 3:

[0538] Users will be able to drive more carefully when they receive alerts and warnings from their devices, for example by slowing down and remaining vigilant when an alert is issued.

[0539] Example 1

[0540] 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."

[0541] Conventional traffic accident prevention systems have had difficulty predicting traffic accidents in advance. In particular, they were unable to adequately consider external factors such as weather and road conditions, limiting the accuracy of their predictions. Furthermore, they were unable to provide real-time information or visually indicate high-risk areas, preventing them from providing useful information to users.

[0542] 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.

[0543] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model to learn traffic accident occurrence patterns and predict traffic accidents, a mapping means for plotting the prediction results on a map using a GIS, and a data distribution means for distributing the generated prediction map to multiple devices in real time via an API or WebSocket. This makes it possible to predict traffic accidents with high accuracy and provide useful information to users in real time.

[0544] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides the results to users.

[0545] A "data collection means" is a method or device for obtaining traffic accident data.

[0546] "Additional data acquisition means" is a method or device for collecting weather and road condition data.

[0547] A "data cleaning means" is a method or device for preprocessing collected data, such as filling in missing values ​​and removing outliers.

[0548] "Analysis means" refers to a method or device that uses a generative AI model to analyze preprocessed data, learn patterns of traffic accident occurrence, and make predictions.

[0549] A "generative AI model" is an artificial intelligence model that learns from collected data and analyzes and predicts traffic accident patterns.

[0550] "Mapping means" refers to a method or device for plotting the analysis results on a map using a geographic information system (GIS).

[0551] "Data distribution means" refers to a method or apparatus for distributing the generated predictive maps to multiple devices in real time via an API or WebSocket.

[0552] The traffic accident forecasting system of the present invention is constructed to accurately predict the occurrence of traffic accidents and provide the information to users in real time. This system includes a data collection means, an additional data acquisition means, a data cleaning means, an analysis means, a mapping means, and a data distribution means.

[0553] Server Processing

[0554] Data collection

[0555] The server collects traffic accident data from police and local governments via API. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast websites. As a specific example, the server obtains traffic accident data from a city's police station from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0556] Data Preprocessing

[0557] The server cleanses the collected data, fills in missing and outlier values, and converts it into a unified format. It also accurately maps date, time, and location information. For example, the server fills in missing latitude and longitude information in traffic accident data, integrates it with weather data, and unifies the date and time format.

[0558] Analysis by generative AI models

[0559] The server inputs the preprocessed data into a generative AI model and analyzes traffic accident occurrence patterns. The generative AI model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model and makes it learn the pattern that "accidents occur frequently during the daytime on rainy days." An example of a prompt is "Analyze the risk of traffic accidents occurring on rainy days based on past traffic accident data."

[0560] Predictive Map Generation

[0561] The server plots the prediction results on a map using a GIS (geographic information system) based on the prediction results from the generative AI model. High-risk locations are displayed in a different color (for example, red). As a specific example, the server displays areas in a city where traffic accidents are frequent in red on a map based on the output of the generative AI model.

[0562] Real-time streaming

[0563] The server distributes the generated predictive maps to various devices (e.g., automobile navigation systems) in real time. Distribution is typically performed via API or WebSocket. For example, the server transmits the generated predictive maps to the navigation system's API in real time.

[0564] Terminal handling

[0565] Receiving information

[0566] The terminal (for example, a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0567] Information display

[0568] The device displays the received forecast information on a map or in an interface. It provides a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data it receives.

[0569] Alert delivery

[0570] The device generates an alert under certain conditions to warn the user. Specifically, it provides audio guidance and visual warnings when approaching a high-risk area. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0571] User Action

[0572] Information confirmation

[0573] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[0574] behavior adjustment

[0575] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[0576] Alert Response

[0577] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0578] In this way, the traffic accident forecasting system can predict the risk of traffic accidents occurring in real time and provide users with immediately useful information.

[0579] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0580] Step 1: Data collection

[0581] Input: Requests to police and local government APIs, the Japan Meteorological Agency, and weather forecast site APIs

[0582] Output: Traffic accident data and weather data are stored on the server.

[0583] Specific operation: The server sends a request to an endpoint such as "https: / / example-police-api.com / data / accidents" or "https: / / example-weather-api.com / data / weather", receives the JSON-formatted data as a response, and then saves it to local storage as " / data / accidents / 2023-01-01_to_2023-07-31.json" or " / data / weather / 2023-01-01_to_2023-07-31.json".

[0584] Step 2: Data Preprocessing

[0585] Input: Traffic accident data and weather data collected in Step 1

[0586] Output: Preprocessed data in a unified format

[0587] What it does: The server cleanses the data and imputes missing and outlier values. For example, it imputes the mean or median for fields marked as "NA." Next, it converts all data to "YYYY-MM-DD HH:MM:SS" format and imputes missing latitude and longitude information using address information.

[0588] Step 3: Analysis by generative AI model

[0589] Input: Preprocessed data from step 2

[0590] Output: Traffic accident prediction results

[0591] Specific operation: The server generates a file called "ai_model_input.csv" and inputs it into the generative AI model. The AI ​​model runs the "train_model()" method and learns patterns from past accident data, such as "accidents tend to occur more frequently during the daytime on rainy days." The prediction results are saved as "predicted_accidents.csv."

[0592] Step 4: Generate a prediction map

[0593] Input: Prediction results obtained in step 3

[0594] Output: A map with risk areas plotted

[0595] Specific operation: The server reads "predicted_accidents.csv" and plots risk areas in red on a map using a GIS (geographic information system). It then executes the function "plot_map('predicted_accidents.csv')" to generate a map image file showing high-risk areas.

[0596] Step 5: Real-time delivery

[0597] Input: Prediction map generated in step 4

[0598] Output: Predictive maps sent to navigation systems via API or WebSocket

[0599] Specific operation: The server executes "send_data_to_api('https: / / example-navigation-api.com / predicted_map', 'predicted_map_data')" and sends the data to the navigation system.

[0600] Step 6: Receiving information (terminal)

[0601] Input: Predictive map data sent from the server

[0602] Output: Predictive map data stored in the device's internal memory

[0603] Specific operation: The device periodically sends a request to the server, executes "fetch(' / predicted_map')" to receive data, and stores it in its internal memory.

[0604] Step 7: Display Information (Terminal)

[0605] Input: Predicted map data received in step 6

[0606] Output: Risk areas displayed in a visually friendly interface

[0607] Specific operation: The device executes the function "display_map(data)" and displays the risk area in red on the map.

[0608] Step 8: Alert delivery (terminal)

[0609] Input: Your device's current location and the risk area information displayed in step 7

[0610] Output: Audio prompts and visual warning alerts

[0611] Specific operation: The device executes "if(current_location in high_risk_area): trigger_alert('There is a risk of an accident at the next intersection. Be careful.')" and issues an audio alert.

[0612] Step 9: User Confirmation

[0613] Input: Risk Area information displayed in step 7

[0614] Output: User who understands risk area and time period

[0615] Specific operation: The user checks the navigation screen and identifies the location of the risk area displayed in red.

[0616] Step 10: Behavioral Adjustment (User)

[0617] Input: Risk area information confirmed in step 9

[0618] Output: Adjusted driving route and speed

[0619] Specific operation: The user follows the re-route guidance provided by the navigation system and selects a safe detour.

[0620] Step 11: Alert response (user)

[0621] Input: The alert received from the terminal in step 8

[0622] Output: More careful driving

[0623] Specific actions: The user hears the alert sound, slows down, and drives safely.

[0624] Through each step in this way, the traffic accident forecasting system is able to predict the risk of traffic accidents occurring in real time and provide users with useful information immediately.

[0625] (Application example 1)

[0626] 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."

[0627] Traffic accidents are an unavoidable problem that seriously impacts human life and property. Even now, as autonomous vehicles become more widespread, the risk of traffic accidents still exists, posing a challenge to the reliability of autonomous driving systems. Conventional navigation systems only provide static map and traffic information, and are unable to predict and respond immediately to changing traffic accident risks in real time. Therefore, there is a need for a method to predict traffic accident risks in real time and efficiently incorporate them into autonomous driving systems.

[0628] 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.

[0629] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for displaying the prediction map information on the navigation system of the autonomous vehicle, an alert generation means for issuing an audio guide or a visual warning when approaching the risk area, and an interface means for allowing the user to check the risk area and adjust their behavior accordingly. This makes it possible to predict traffic accident risks in real time and efficiently reflect the results in the autonomous vehicle.

[0630] The "traffic accident forecast system" is a system that predicts the risk of traffic accidents and encourages appropriate responses.

[0631] "Data collection means" refers to a means for collecting traffic accident data.

[0632] The "additional data acquisition means" is a means for collecting weather and road surface condition data.

[0633] "Data cleaning means" refers to means for preprocessing collected traffic accident data and additional data to make them suitable for analysis.

[0634] "Analysis means" refers to a means for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[0635] "Mapping means" is a means for plotting the analysis results on a map.

[0636] The "data distribution means" is a means for distributing the generated predictive map to multiple devices in real time.

[0637] The "means for displaying on a navigation system" refers to a means for displaying the predicted map information on a navigation system of an autonomous driving vehicle.

[0638] The "alert generating means" is a means for issuing an audio guide or a visual warning when approaching the risk area.

[0639] The "interface means" is a means by which users can check risk areas and adjust their behavior accordingly.

[0640] As an embodiment of the present invention, a traffic accident prediction system includes the following steps: A server collects traffic accident data and acquires weather and road condition data as additional data; the data is preprocessed by a data cleaning means and analyzed using a generative AI model; the analyzed results are plotted on a map by a mapping means; and the predicted map is distributed to multiple devices in real time by a data distribution means.

[0641] The server uses the following specific hardware and software: A REST API is used to collect data, and data collection scripts are executed using the Python language. Weather and road condition data is obtained from external data sources via the API. For data preprocessing, the Python Pandas library is used to fill in missing values ​​and clean the data. For the generative AI model, machine learning libraries such as TensorFlow and PyTorch are used to train the model on past traffic accident data and weather data. For plotting on a map, the Folium library is used, and the prediction results are displayed on the map via the Google Maps API.

[0642] The terminal side acquires the predictive map information using a means to display it on the navigation system and provides the information visually to the user. This interface has an alert generation means that issues audio guidance and visual warnings when approaching a risk area. This alert generation uses WebSocket and API to receive data distribution in real time.

[0643] The user checks the displayed predictive map information to understand the relationship between their current location and the risk area. When approaching a risk area, they take appropriate action based on audio guidance and visual warnings. The user can select a detour or adjust their driving speed based on the information displayed on the navigation system.

[0644] As a concrete example, traffic accident data and weather data are collected on a server, and analyzed by a generative AI model, resulting in the learning result that "accidents occur frequently during the daytime on rainy days." This information is input to the generative AI model as the following prompt:

[0645] Example prompt sentence:

[0646] Analyze traffic accident patterns based on traffic accident data from January 1, 2023 to July 31, 2023, and weather data for the same period. Learn the tendency for accidents to occur more frequently during the daytime on rainy days, and build a prediction model.

[0647] In this way, predicted traffic accident risk information is delivered to user terminals in real time, which can contribute to reducing traffic accident risks in actual usage environments.

[0648] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0649] Step 1:

[0650] The server collects traffic accident data. This is done using APIs or database connections provided by the police or local government. The server retrieves traffic accident data from, for example, January 1, 2023 to July 31, 2023. The input is traffic accident data, and the output is the collected raw data. This data includes the date, time, location, and details of the accident.

[0651] Step 2:

[0652] The server collects additional weather and road condition data. For this, it uses the API of the Japan Meteorological Agency or weather forecast site. For example, it obtains weather data for the same period. The input is the weather and road condition data, and the output is the collected raw data. This data includes temperature, precipitation, road condition, etc.

[0653] Step 3:

[0654] The server preprocesses the collected traffic accident data and additional data. First, it completes missing values ​​and removes inappropriate data. Next, it unifies the data format and accurately maps date, time, and location information. The input is the collected raw data, and the output is the cleansed data. Specifically, it uses the Python Pandas library to cleanse the data.

[0655] Step 4:

[0656] The server inputs the preprocessed data into a generative AI model to predict the risk of traffic accidents. The generative AI model learns accident patterns from past traffic accident data and weather data and builds a predictive model. The input is the preprocessed data, and the output is risk prediction data. Specifically, the AI ​​model is run using TensorFlow or PyTorch.

[0657] Step 5:

[0658] The server plots the predicted risk data on a map. Specifically, it uses GIS technology to color-code risk areas. The input is risk prediction data, and the output is a prediction map. Specifically, it generates a map using the Folium library and displays it on the map using the Google Maps API.

[0659] Step 6:

[0660] The server delivers the generated predicted map to the device in real time. This is done using an API or WebSocket. The input is the predicted map, and the output is real-time delivery data. Specifically, data is delivered using a web framework such as Flask.

[0661] Step 7:

[0662] The device receives predictive map information from the server via API. The input is real-time data delivered from the server, and the output is predictive map information stored inside the device. Specifically, the navigation system stores the received data in memory.

[0663] Step 8:

[0664] The device displays the received predictive map information on a map or interface. The input is the predictive map information stored in the device, and the output is the visually displayed map information. Specifically, the navigation system displays risk areas in red on the map.

[0665] Step 9:

[0666] The device generates audio guidance and visual warnings when approaching a risk area. The input is the predicted map information and current location information stored in the device, and the output is the audio guidance and visual warning. Specifically, the navigation system issues an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0667] Step 10:

[0668] The user checks the forecast information displayed on the device and understands the risk areas near their current location. The input is the visually displayed map information, and the output is the user's perception. Specifically, the driver checks the navigation screen and confirms the location of the risk area displayed in red.

[0669] Step 11:

[0670] The user adjusts the driving route and speed based on the confirmed information. The input is the risk information recognized by the user, and the output is the adjusted driving behavior. Specifically, the driver follows the navigation system's instructions to select a detour to avoid the risk area.

[0671] Step 12:

[0672] Users receive alerts from their devices and drive more carefully. The input is audio guidance and visual warnings, and the output is safe driving behavior. Specifically, when a driver hears an alert and receives information that there is a risk of an accident at the next intersection, they slow down and continue driving vigilantly.

[0673] 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.

[0674] The traffic accident prediction system of the present invention provides more effective accident prevention information by combining the user's emotion engine. This system is implemented as follows.

[0675] Server Processing

[0676] Data collection

[0677] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0678] Data Preprocessing

[0679] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[0680] Analysis by generative AI models

[0681] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[0682] Predictive Map Generation

[0683] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[0684] Real-time streaming

[0685] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[0686] Terminal handling

[0687] Receiving information

[0688] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0689] Information display

[0690] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[0691] Alert delivery

[0692] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0693] Emotion engine processing

[0694] Emotional Data Collection

[0695] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera captures the user's face, and the microphone records their voice.

[0696] Emotion analysis

[0697] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. For example, the device may determine that the user is in a "tense state" based on changes in their facial expression.

[0698] Adjusting alert content

[0699] The displayed predictive information and alert content are adjusted based on the results of emotion analysis. For example, if the user is in a tense state, the alert wording will be softer. For example, a message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" will be displayed.

[0700] User Action

[0701] Information confirmation

[0702] Users can check the traffic accident prediction information and adjusted alert messages displayed on the device, which allows them to accurately identify risk areas and high-risk time periods.

[0703] behavior adjustment

[0704] The user can adjust their driving route and speed based on the displayed forecast information and alert messages. For example, to avoid risk areas, the user can select a detour route according to the navigation system's instructions.

[0705] Alert Response

[0706] Users can drive more carefully when they receive alerts and warnings from their devices. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0707] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[0708] The processing flow will be explained below.

[0709] Server Processing

[0710] Step 1:

[0711] The server receives traffic accident data from the police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the victim's condition.

[0712] Step 2:

[0713] The server obtains weather and road condition data from the Japan Meteorological Agency and weather forecast websites via API. Weather data includes date, time, temperature, precipitation, wind speed, etc.

[0714] Step 3:

[0715] The server stores the acquired traffic accident data and weather data in a database, allowing each data item to be managed in an integrated manner.

[0716] Step 4:

[0717] The server uses data cleaning means to fill in missing values ​​in the collected data, remove unnecessary information, and standardize the format.

[0718] Step 5:

[0719] The server inputs the preprocessed data into a generative AI model, which analyzes and learns patterns of traffic accident occurrence.

[0720] Step 6:

[0721] The server uses a trained generative AI model to predict the risk of traffic accidents based on the day's weather data.

[0722] Step 7:

[0723] The server plots the prediction results on a map using GIS (geographic information system), visually displaying high-risk locations.

[0724] Step 8:

[0725] The server delivers the generated predictive maps to various devices in real time via API or WebSocket.

[0726] Terminal handling

[0727] Step 1:

[0728] The terminal receives forecast information sent in real time from the server via an API.

[0729] Step 2:

[0730] The terminal integrates the received forecast information with map data and stores it in its internal memory.

[0731] Step 3:

[0732] Based on the received data, the terminal displays risk areas on a map, color-coding them (for example, red).

[0733] Step 4:

[0734] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[0735] Emotion engine processing

[0736] Step 1:

[0737] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors.

[0738] Step 2:

[0739] The device inputs the collected data into an emotion engine to analyze the user's emotional state in real time.

[0740] Step 3:

[0741] The device adjusts the content of the displayed forecast information and alerts based on the user's emotional state as determined by the emotion engine.

[0742] User Action

[0743] Step 1:

[0744] The user checks the traffic accident prediction information and the adjusted alert message displayed on the terminal.

[0745] Step 2:

[0746] The user can adjust their driving route and speed based on the displayed forecast information and alert messages.

[0747] Step 3:

[0748] Users will receive alerts and warnings from their devices and drive more carefully.

[0749] Example 2

[0750] 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."

[0751] Conventional traffic accident forecasting systems only identify risk areas based on past traffic accident data and weather data. As a result, they do not reflect the user's emotional state, and therefore may not provide accurate alerts. In particular, when emotional states such as tension and fatigue contribute to accident risk, not taking these data into account is a serious drawback.

[0752] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0753] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for checking traffic accident prediction information and alert messages displayed by the terminal, a driving adjustment means for adjusting the driving route and speed based on the prediction information and alert messages displayed by the terminal, an emotion analysis means for collecting and analyzing user emotion data by the terminal, and an alert adjustment means for adjusting the content of the alert based on the emotion analysis results. This enables the provision of more accurate traffic accident forecasts and alerts that take the user's emotional state into consideration.

[0754] A "traffic accident forecast system" is a system that predicts the risk of traffic accidents and provides users with that information.

[0755] "Data collection means" refers to devices and methods for collecting traffic accident data.

[0756] "Additional data acquisition means" refers to devices and methods for collecting weather and road condition data.

[0757] "Data cleaning means" refers to devices and methods for organizing collected data and removing unnecessary information.

[0758] "Analysis means" refers to devices or methods for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[0759] "Mapping means" refers to a device or method for plotting the analysis results on a map.

[0760] "Data distribution means" refers to an apparatus or method for distributing the generated predictive map to multiple devices in real time.

[0761] "Terminal" refers to a device that displays prediction information and collects user emotion data.

[0762] "Emotion analysis means" refers to a device or method for collecting and analyzing user emotion data.

[0763] "Alert adjustment means" refers to a device or method for adjusting the content of an alert based on the results of emotion analysis.

[0764] "Driving adjustment means" refers to a device or method for adjusting the driving route and speed based on the forecast information and alert messages displayed on the terminal.

[0765] The traffic accident forecasting system of this invention is a system that combines user emotional information to provide more effective accident prevention information. This system functions in cooperation with three parties: a server, a terminal, and a user.

[0766] Server Processing

[0767] Data collection

[0768] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, a detailed dataset will be created by collecting traffic accident and weather data in Tokyo from January 1, 2023 to July 31, 2023.

[0769] Data Preprocessing

[0770] Collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, missing latitude and longitude information in traffic accident data is filled, and weather data is integrated to unify the date and time format.

[0771] Analysis by generative AI models

[0772] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, past accident data is input into the generative AI model, and it learns the pattern that "accidents occur frequently during the daytime on rainy days."

[0773] Predictive Map Generation

[0774] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (e.g., red). For example, based on the output of the generative AI model, areas in Tokyo with a high incidence of traffic accidents are displayed in red on a map.

[0775] Real-time streaming

[0776] The generated predictive map is distributed in real time to various devices (e.g., automobile navigation systems). Distribution is performed via API or WebSocket. For example, the generated predictive map is sent in real time to the navigation system's API.

[0777] Terminal handling

[0778] Receiving information

[0779] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. The navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0780] Information display

[0781] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface. For example, based on the data received by the navigation system, risk areas are displayed in red on a map.

[0782] Alert delivery

[0783] When approaching a high-risk area, audio guidance and visual warnings are provided, such as generating an audio alert saying, "There is an accident risk at the next intersection. Be careful."

[0784] Emotion engine processing

[0785] Emotional Data Collection

[0786] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera scans the user's face and the microphone records their voice.

[0787] Emotion analysis

[0788] The collected user data is input into an emotion engine to analyze the user's emotional state in real time, thereby determining whether the user is tense or relaxed.

[0789] Adjusting alert content

[0790] The displayed forecast information and alerts are adjusted based on the results of emotion analysis. For example, if the user is feeling nervous, a gentler message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" is used.

[0791] User Action

[0792] Information confirmation

[0793] The user checks the traffic accident prediction information and coordinated alert messages displayed on the device, which allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action.

[0794] behavior adjustment

[0795] The user can adjust their driving route and speed based on the displayed forecast information and alert messages, and can select a detour route to avoid risk areas by following the navigation system's instructions.

[0796] Alert Response

[0797] Users can drive more carefully by receiving alerts and warnings from their devices. For example, a driver may hear an alert and receive information that there is a risk of an accident at the next intersection, so they can slow down and continue driving vigilantly.

[0798] Examples of specific examples and prompts

[0799] Specific examples

[0800] The server collects traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, and analyzes the data using a generative AI model. From the analysis results, it learns patterns of accidents occurring frequently during the daytime on rainy days, and identifies and plots risk areas within Tokyo. This information is then sent to the navigation system in real time, where the user can confirm it and take appropriate measures.

[0801] Prompt Sentence Examples

[0802] "Based on traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, the generative AI model identifies areas where accidents are expected to occur frequently during the daytime on rainy days, and displays the results on a map. If the user is in a tense state, the alert message will be softened."

[0803] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[0804] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0805] Step 1: Data collection

[0806] The server collects traffic accident data from police, prosecutors, local governments, etc. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. Specifically, it uses API or database connections to obtain traffic accident and weather data for Tokyo from January 1, 2023 to July 31, 2023. It receives traffic accident data and weather data as input and builds a detailed dataset as output.

[0807] Step 2: Data Preprocessing

[0808] The server preprocesses the collected data. Specifically, it removes unnecessary information and fills in missing values. For example, it fills in missing latitude and longitude information, integrates it with weather data, and standardizes the date and time format. It receives the dataset constructed in step 1 as input and generates cleansed and unified data as output.

[0809] Step 3: Analysis by generative AI model

[0810] The server inputs the preprocessed data into the generative AI model and analyzes traffic accident occurrence patterns. Specifically, it inputs past accident data and weather data into the AI ​​model, and has it learn patterns such as "accidents occur frequently during the daytime on rainy days." It receives the preprocessed data as input and generates a predicted result of the risk of traffic accidents as output.

[0811] Step 4: Generate a prediction map

[0812] The server uses a GIS (geographic information system) to plot the prediction results on a map based on the predictions from the generative AI model. Specifically, it colors the results by, for example, showing high-risk areas in red. It receives the prediction results from the generative AI model as input and generates a visualized prediction map as output.

[0813] Step 5: Real-time delivery

[0814] The server delivers the generated predictive map to the device in real time. Specifically, it sends the predictive map to a navigation system or other device using an API or WebSocket. It receives the predictive map as input and delivers it to the device in real time as output.

[0815] Step 6: Receiving information

[0816] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. Specifically, it receives the forecast data through the API and stores it in its internal memory. It receives the forecast information from the server as input and generates the forecast information stored in its internal memory as output.

[0817] Step 7: Display information

[0818] The device displays the received forecast information on a map or interface. Specifically, it displays risk areas in red on the map to allow the user to visually recognize them. It receives forecast information stored in its internal memory as input and displays visualized forecast information as output.

[0819] Step 8: Alert Delivery

[0820] The device provides audio guidance and visual warnings when approaching a high-risk area. Specifically, it generates an audio alert saying, "There is an accident risk at the next intersection. Please be careful." It receives visualized prediction information as input and generates an alert message as output.

[0821] Step 9: Emotional Data Collection

[0822] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the camera scans the user's face and the microphone records their voice. The device receives the user's emotional data as input and collects emotional data as output.

[0823] Step 10: Sentiment Analysis

[0824] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. Specifically, it determines whether the user is tense or relaxed. It receives emotional data as input and generates analysis results as output.

[0825] Step 11: Adjust the alert content

[0826] The device adjusts the content of the alert based on the emotion analysis results. Specifically, if the user is in a tense state, it displays a message such as, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly." It receives the emotion analysis results as input and generates a tailored alert message as output.

[0827] Step 12: Verify the information

[0828] The user checks the traffic accident prediction information and the adjusted alert message displayed on the device. This allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action. The user receives visualized prediction information and alert messages as input, and checks the information as output.

[0829] Step 13: Behavioral Adjustments

[0830] The user adjusts the driving route and speed based on the displayed forecast information and alert messages. Specifically, the user selects a detour route to avoid risk areas according to the navigation instructions. The user receives confirmed information as input and performs adjusted driving behavior as output.

[0831] Step 14: Respond to alerts

[0832] Users receive alerts and warnings from their devices and drive more carefully. Specifically, when a user hears an alert, they recognize that there is a risk of an accident at the next intersection, slow down, and continue driving vigilantly. The alert message is received as input, and the behavior of driving carefully is output.

[0833] (Application example 2)

[0834] 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."

[0835] There is a need to provide automated driving vehicles and drivers with appropriate and timely information on traffic accident risks. In particular, it is necessary to more effectively promote safe driving by providing alerts and guidance that take into account the driver's emotional state. However, existing systems are insufficient in providing such comprehensive information, and integrating risk information and alerts that respond to the driver's emotional state is a challenge.

[0836] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, an emotion data collection means for collecting user emotion data, an emotion analysis means for analyzing the collected emotion data, and an alert adjustment means for adjusting the display and alert content based on the emotion analysis results. This makes it possible to provide integrated traffic accident risk information and the driver's emotional state.

[0837] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides that information to users.

[0838] The "data collection means" is a device or program that has the function of acquiring traffic accident data.

[0839] The "additional data acquisition means" is a device or program that acquires weather and road condition data.

[0840] The "data cleaning means" is a device or program that has the function of preprocessing the acquired traffic accident data and additional data.

[0841] A "generative AI model" is an artificial intelligence model that performs predictive analysis of traffic accident occurrences based on collected data.

[0842] "Analysis means" refers to a device or program that analyzes preprocessed data using a generative AI model and predicts traffic accidents.

[0843] "Mapping means" refers to a device or program that has the function of plotting the analysis results on a map.

[0844] The "data distribution means" is a device or program that has the function of distributing the generated predictive map to multiple devices in real time.

[0845] "Emotion data collection means" refers to a device or program that has the function of acquiring user emotion data.

[0846] The "emotion analysis means" is a device or program that has the function of analyzing collected emotion data and determining the user's emotional state.

[0847] The "alert adjustment means" is a device or program that has the function of adjusting the display and alert content based on the emotion analysis results.

[0848] "Traffic Accident Data" refers to information relating to past and current traffic accidents.

[0849] "Weather Data" means information regarding current and forecast weather.

[0850] "Road Condition Data" refers to information regarding current and predicted road surface conditions.

[0851] "Plotting on a map" refers to visually displaying the analysis results using a geographic information system (GIS).

[0852] "Delivering in real time" refers to immediately transmitting the generated prediction map to the user device.

[0853] "Adjusting alert content" refers to changing the content and intensity of a warning message taking into account the user's emotional state.

[0854] The traffic accident forecasting system of this invention effectively utilizes various data to provide information useful for predicting and preventing traffic accidents. This system is mainly composed of three main components: a server, a terminal, and a user. Each component and its specific processing are explained below.

[0855] Server Processing

[0856] Data collection

[0857] The server collects traffic accident data and weather and road condition data. Traffic accident data is obtained from police and local government databases, while weather and road condition data is obtained from weather information websites and the Japan Meteorological Agency. These data are collected through APIs and database connections.

[0858] Data Preprocessing

[0859] The collected data is first preprocessed using data cleaning methods, such as imputing missing values, removing unnecessary information, and standardizing the format, to prepare the dataset for analysis.

[0860] Analysis by generative AI models

[0861] The preprocessed data is input into a generative AI model to predict traffic accidents. The generative AI model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents.

[0862] Predictive map generation and delivery

[0863] The prediction results of the generative AI model are plotted on a map using GIS, generating a predictive map. This predictive map is then distributed in real time to multiple devices, such as navigation systems and smartphone applications, via a data distribution method.

[0864] Terminal handling

[0865] Receiving information

[0866] The device receives the forecast information sent from the server, typically via an API.

[0867] Information display

[0868] The received forecast information is displayed on a map or in the UI, providing a visually easy-to-understand interface that makes it easy for users to identify risk areas.

[0869] Emotional data collection and analysis

[0870] The device is equipped with a camera, microphone, and sensors to collect the user's facial expressions, voice, and vital data. This data is analyzed in real time by emotion analysis. For example, if the user is in a tense state, this information can be obtained as an analysis result.

[0871] User Action

[0872] Information confirmation

[0873] The device displays traffic accident prediction information and alert messages tailored to the user's emotional state, allowing the user to instantly identify risk areas and dangerous time periods.

[0874] behavior adjustment

[0875] Users can adjust their driving route and speed based on the displayed forecast information and alert messages, for example by choosing a detour route to avoid risk areas.

[0876] Alert Response

[0877] Users receive alerts and warnings from their devices and strive to drive safely. For example, if they receive a message such as "There is a risk of an accident at the next intersection. Please be careful," they should slow down and be vigilant.

[0878] Specific examples

[0879] The server collects traffic accident data and weather data over a certain period of time and performs data cleaning. A generative AI model is used to learn a predictive pattern, such as "high accident rates during the daytime on rainy days." A risk map is then generated using GIS and sent to the device in real time. The device receives this information and displays risk areas in red on the map. In addition, the camera captures the user's face, and a microphone collects audio data for emotion analysis. If the device determines that the user is in a tense state, it will provide an audio message saying, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly."

[0880] Prompt Sentence Examples

[0881] "This driver is on edge. What kind of caution would work in the next high-accident area?"

[0882] "The current weather is rain, and the location is XX intersection. Please predict the accident risk under these conditions."

[0883] In this way, the traffic accident forecast system incorporates the user's emotional information to provide more effective accident prevention information.

[0884] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0885] Step 1: Collecting traffic accident and weather data

[0886] The server collects traffic accident data and weather data. It obtains traffic accident data from police and local government databases, and weather data from weather information websites and the Japan Meteorological Agency. It collects data through APIs and database connections and stores it. The input is traffic accident data and weather data, and the output is raw data.

[0887] Step 2: Preprocessing the data

[0888] The server preprocesses the collected data by completing missing values, removing unnecessary information, and standardizing the format. The input is raw data, and the output is preprocessed data. Specific operations include cleaning the data, completing missing values, and correcting improper formats.

[0889] Step 3: Analysis by generative AI model

[0890] The server inputs the preprocessed data into the generative AI model to predict traffic accidents. The AI ​​model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents. The input is the preprocessed data, and the output is the traffic accident risk prediction results. Specific operations include inputting data into the AI ​​model, running the model, and obtaining the results.

[0891] Step 4: Generate prediction maps

[0892] The server uses GIS to plot the prediction results of the generative AI model on a map and generate a prediction map. The input is the traffic accident risk prediction result, and the output is the prediction map. Specifically, the server uses GIS software to overlay geographical information and risk information.

[0893] Step 5: Serving the predicted map

[0894] The server distributes the generated predictive map to terminals in real time via a data distribution means. The input is the predictive map, and the output is the data distributed to each terminal. Specifically, the server uses an API or WebSocket to send the predictive map to multiple terminals.

[0895] Step 6: Collect emotion data

[0896] To collect the user's emotional data, the device uses a camera, microphone, and sensors to acquire the user's facial expressions, voice, and vital data. The input is the user's biometric information, and the output is emotional data. Specifically, the device performs processes to capture the user's face, record their voice, and record their vital data.

[0897] Step 7: Analyze the sentiment data

[0898] The device inputs the collected emotional data into an emotion analysis means to determine the user's emotional state. The input is emotional data, and the output is the emotion analysis result. Specifically, the device performs emotion analysis using a deep learning model to determine the user's emotional state as "tense" or "calm."

[0899] Step 8: Adjust and view alerts

[0900] The device adjusts the content of the alert based on the emotion analysis results and notifies the user by display and audio. The input is the emotion analysis results and traffic accident prediction information, and the output is the adjusted alert message. Specifically, the device generates an alert message according to the user's emotional state and notifies the user by visual and audio means.

[0901] Step 9: User Behavior

[0902] The user adjusts their behavior based on the traffic accident prediction information displayed on the device and the adjusted alert message. The input is the adjusted alert message, and the output is the user's behavior. Specific actions include choosing a detour route to avoid risk areas or reducing speed.

[0903] 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.

[0904] 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.

[0905] 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.

[0906] [Third embodiment]

[0907] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0908] 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.

[0909] 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).

[0910] 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.

[0911] 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.

[0912] 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).

[0913] 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.

[0914] 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.

[0915] 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.

[0916] 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.

[0917] 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.

[0918] 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."

[0919] The traffic accident forecasting system of the present invention is implemented as follows.

[0920] Server Processing

[0921] Data collection

[0922] The server collects traffic accident data from police, prosecutors, local governments, etc. via API or database connection. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0923] Data Preprocessing

[0924] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[0925] Analysis by generative AI models

[0926] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[0927] Predictive Map Generation

[0928] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[0929] Real-time streaming

[0930] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[0931] Terminal handling

[0932] Receiving information

[0933] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[0934] Information display

[0935] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[0936] Alert delivery

[0937] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[0938] User Action

[0939] Information confirmation

[0940] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[0941] behavior adjustment

[0942] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[0943] Alert Response

[0944] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[0945] A traffic accident forecast system is realized by this series of processes.

[0946] The processing flow will be explained below.

[0947] Server Processing

[0948] Step 1:

[0949] The server obtains traffic accident data from police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the status of the victims.

[0950] Step 2:

[0951] The server retrieves weather and road condition data from the Japan Meteorological Agency and weather forecast sites, including date, time, temperature, precipitation, wind speed, etc.

[0952] Step 3:

[0953] The server stores the collected traffic accident data and weather data in a database and manages them centrally.

[0954] Step 4:

[0955] The server uses data cleaning techniques to fill in missing values, remove unnecessary information, and standardize formats, for example, filling in missing latitude and longitude information in accident data.

[0956] Step 5:

[0957] The server inputs the preprocessed data into the generative AI model, which then learns traffic accident occurrence patterns.The AI ​​model then analyzes past data and learns the risk of accidents occurring under specific conditions.

[0958] Step 6:

[0959] The server uses a trained generative AI model to input the day's weather data and predict the risk of traffic accidents.

[0960] Step 7:

[0961] The server plots the prediction results on a map using a GIS (geographic information system), which visually displays high-risk locations.

[0962] Step 8:

[0963] The server delivers the created predictive maps to various devices in real time via API or WebSocket.

[0964] Terminal handling

[0965] Step 1:

[0966] A terminal (e.g., a car navigation system) receives forecast information sent from the server via an API.

[0967] Step 2:

[0968] The terminal integrates the received forecast data with map data and stores it in its internal memory.

[0969] Step 3:

[0970] The terminal displays risk areas on the map in different colors (e.g., red), allowing the user to visually identify the risk areas.

[0971] Step 4:

[0972] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[0973] User Action

[0974] Step 1:

[0975] The user checks the traffic accident prediction information displayed on the device, which allows them to understand risk areas and high-risk time periods.

[0976] Step 2:

[0977] The user can adjust their driving route and speed based on the displayed forecast information, for example, by choosing a detour route to avoid risk areas.

[0978] Step 3:

[0979] Users will be able to drive more carefully when they receive alerts and warnings from their devices, for example by slowing down and remaining vigilant when an alert is issued.

[0980] Example 1

[0981] 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."

[0982] Conventional traffic accident prevention systems have had difficulty predicting traffic accidents in advance. In particular, they were unable to adequately consider external factors such as weather and road conditions, limiting the accuracy of their predictions. Furthermore, they were unable to provide real-time information or visually indicate high-risk areas, preventing them from providing useful information to users.

[0983] 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.

[0984] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model to learn traffic accident occurrence patterns and predict traffic accidents, a mapping means for plotting the prediction results on a map using a GIS, and a data distribution means for distributing the generated prediction map to multiple devices in real time via an API or WebSocket. This makes it possible to predict traffic accidents with high accuracy and provide useful information to users in real time.

[0985] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides the results to users.

[0986] A "data collection means" is a method or device for obtaining traffic accident data.

[0987] "Additional data acquisition means" is a method or device for collecting weather and road condition data.

[0988] A "data cleaning means" is a method or device for preprocessing collected data, such as filling in missing values ​​and removing outliers.

[0989] "Analysis means" refers to a method or device that uses a generative AI model to analyze preprocessed data, learn patterns of traffic accident occurrence, and make predictions.

[0990] A "generative AI model" is an artificial intelligence model that learns from collected data and analyzes and predicts traffic accident patterns.

[0991] "Mapping means" refers to a method or device for plotting the analysis results on a map using a geographic information system (GIS).

[0992] "Data distribution means" refers to a method or apparatus for distributing the generated predictive maps to multiple devices in real time via an API or WebSocket.

[0993] The traffic accident forecasting system of the present invention is constructed to accurately predict the occurrence of traffic accidents and provide the information to users in real time. This system includes a data collection means, an additional data acquisition means, a data cleaning means, an analysis means, a mapping means, and a data distribution means.

[0994] Server Processing

[0995] Data collection

[0996] The server collects traffic accident data from police and local governments via API. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast websites. As a specific example, the server obtains traffic accident data from a city's police station from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[0997] Data Preprocessing

[0998] The server cleanses the collected data, fills in missing and outlier values, and converts it into a unified format. It also accurately maps date, time, and location information. For example, the server fills in missing latitude and longitude information in traffic accident data, integrates it with weather data, and unifies the date and time format.

[0999] Analysis by generative AI models

[1000] The server inputs the preprocessed data into a generative AI model and analyzes traffic accident occurrence patterns. The generative AI model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model and makes it learn the pattern that "accidents occur frequently during the daytime on rainy days." An example of a prompt is "Analyze the risk of traffic accidents occurring on rainy days based on past traffic accident data."

[1001] Predictive Map Generation

[1002] The server plots the prediction results on a map using a GIS (geographic information system) based on the prediction results from the generative AI model. High-risk locations are displayed in a different color (for example, red). As a specific example, the server displays areas in a city where traffic accidents are frequent in red on a map based on the output of the generative AI model.

[1003] Real-time streaming

[1004] The server distributes the generated predictive maps to various devices (e.g., automobile navigation systems) in real time. Distribution is typically performed via API or WebSocket. For example, the server transmits the generated predictive maps to the navigation system's API in real time.

[1005] Terminal handling

[1006] Receiving information

[1007] The terminal (for example, a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1008] Information display

[1009] The device displays the received forecast information on a map or in an interface. It provides a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data it receives.

[1010] Alert delivery

[1011] The device generates an alert under certain conditions to warn the user. Specifically, it provides audio guidance and visual warnings when approaching a high-risk area. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1012] User Action

[1013] Information confirmation

[1014] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[1015] behavior adjustment

[1016] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[1017] Alert Response

[1018] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[1019] In this way, the traffic accident forecasting system can predict the risk of traffic accidents occurring in real time and provide users with immediately useful information.

[1020] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1021] Step 1: Data collection

[1022] Input: Requests to police and local government APIs, the Japan Meteorological Agency, and weather forecast site APIs

[1023] Output: Traffic accident data and weather data are stored on the server.

[1024] Specific operation: The server sends a request to an endpoint such as "https: / / example-police-api.com / data / accidents" or "https: / / example-weather-api.com / data / weather", receives the JSON-formatted data as a response, and then saves it to local storage as " / data / accidents / 2023-01-01_to_2023-07-31.json" or " / data / weather / 2023-01-01_to_2023-07-31.json".

[1025] Step 2: Data Preprocessing

[1026] Input: Traffic accident data and weather data collected in Step 1

[1027] Output: Preprocessed data in a unified format

[1028] What it does: The server cleanses the data and imputes missing and outlier values. For example, it imputes the mean or median for fields marked as "NA." Next, it converts all data to "YYYY-MM-DD HH:MM:SS" format and imputes missing latitude and longitude information using address information.

[1029] Step 3: Analysis by generative AI model

[1030] Input: Preprocessed data from step 2

[1031] Output: Traffic accident prediction results

[1032] Specific operation: The server generates a file called "ai_model_input.csv" and inputs it into the generative AI model. The AI ​​model runs the "train_model()" method and learns patterns from past accident data, such as "accidents tend to occur more frequently during the daytime on rainy days." The prediction results are saved as "predicted_accidents.csv."

[1033] Step 4: Generate a prediction map

[1034] Input: Prediction results obtained in step 3

[1035] Output: A map with risk areas plotted

[1036] Specific operation: The server reads "predicted_accidents.csv" and plots risk areas in red on a map using a GIS (geographic information system). It then executes the function "plot_map('predicted_accidents.csv')" to generate a map image file showing high-risk areas.

[1037] Step 5: Real-time delivery

[1038] Input: Prediction map generated in step 4

[1039] Output: Predictive maps sent to navigation systems via API or WebSocket

[1040] Specific operation: The server executes "send_data_to_api('https: / / example-navigation-api.com / predicted_map', 'predicted_map_data')" and sends the data to the navigation system.

[1041] Step 6: Receiving information (terminal)

[1042] Input: Predictive map data sent from the server

[1043] Output: Predictive map data stored in the device's internal memory

[1044] Specific operation: The device periodically sends a request to the server, executes "fetch(' / predicted_map')" to receive data, and stores it in its internal memory.

[1045] Step 7: Display Information (Terminal)

[1046] Input: Predicted map data received in step 6

[1047] Output: Risk areas displayed in a visually friendly interface

[1048] Specific operation: The device executes the function "display_map(data)" and displays the risk area in red on the map.

[1049] Step 8: Alert delivery (terminal)

[1050] Input: Your device's current location and the risk area information displayed in step 7

[1051] Output: Audio prompts and visual warning alerts

[1052] Specific operation: The device executes "if(current_location in high_risk_area): trigger_alert('There is a risk of an accident at the next intersection. Be careful.')" and issues an audio alert.

[1053] Step 9: User Confirmation

[1054] Input: Risk Area information displayed in step 7

[1055] Output: User who understands risk area and time period

[1056] Specific operation: The user checks the navigation screen and identifies the location of the risk area displayed in red.

[1057] Step 10: Behavioral Adjustment (User)

[1058] Input: Risk area information confirmed in step 9

[1059] Output: Adjusted driving route and speed

[1060] Specific operation: The user follows the re-route guidance provided by the navigation system and selects a safe detour.

[1061] Step 11: Alert response (user)

[1062] Input: The alert received from the terminal in step 8

[1063] Output: More careful driving

[1064] Specific actions: The user hears the alert sound, slows down, and drives safely.

[1065] Through each step in this way, the traffic accident forecasting system is able to predict the risk of traffic accidents occurring in real time and provide users with useful information immediately.

[1066] (Application example 1)

[1067] 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."

[1068] Traffic accidents are an unavoidable problem that seriously impacts human life and property. Even now, as autonomous vehicles become more widespread, the risk of traffic accidents still exists, posing a challenge to the reliability of autonomous driving systems. Conventional navigation systems only provide static map and traffic information, and are unable to predict and respond immediately to changing traffic accident risks in real time. Therefore, there is a need for a method to predict traffic accident risks in real time and efficiently incorporate them into autonomous driving systems.

[1069] 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.

[1070] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for displaying the prediction map information on the navigation system of the autonomous vehicle, an alert generation means for issuing an audio guide or a visual warning when approaching the risk area, and an interface means for allowing the user to check the risk area and adjust their behavior accordingly. This makes it possible to predict traffic accident risks in real time and efficiently reflect the results in the autonomous vehicle.

[1071] The "traffic accident forecast system" is a system that predicts the risk of traffic accidents and encourages appropriate responses.

[1072] "Data collection means" refers to a means for collecting traffic accident data.

[1073] The "additional data acquisition means" is a means for collecting weather and road surface condition data.

[1074] "Data cleaning means" refers to means for preprocessing collected traffic accident data and additional data to make them suitable for analysis.

[1075] "Analysis means" refers to a means for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[1076] "Mapping means" is a means for plotting the analysis results on a map.

[1077] The "data distribution means" is a means for distributing the generated predictive map to multiple devices in real time.

[1078] The "means for displaying on a navigation system" refers to a means for displaying the predicted map information on a navigation system of an autonomous driving vehicle.

[1079] The "alert generating means" is a means for issuing an audio guide or a visual warning when approaching the risk area.

[1080] The "interface means" is a means by which users can check risk areas and adjust their behavior accordingly.

[1081] As an embodiment of the present invention, a traffic accident prediction system includes the following steps: A server collects traffic accident data and acquires weather and road condition data as additional data; the data is preprocessed by a data cleaning means and analyzed using a generative AI model; the analyzed results are plotted on a map by a mapping means; and the predicted map is distributed to multiple devices in real time by a data distribution means.

[1082] The server uses the following specific hardware and software: A REST API is used to collect data, and data collection scripts are executed using the Python language. Weather and road condition data is obtained from external data sources via the API. For data preprocessing, the Python Pandas library is used to fill in missing values ​​and clean the data. For the generative AI model, machine learning libraries such as TensorFlow and PyTorch are used to train the model on past traffic accident data and weather data. For plotting on a map, the Folium library is used, and the prediction results are displayed on the map via the Google Maps API.

[1083] The terminal side acquires the predictive map information using a means to display it on the navigation system and provides the information visually to the user. This interface has an alert generation means that issues audio guidance and visual warnings when approaching a risk area. This alert generation uses WebSocket and API to receive data distribution in real time.

[1084] The user checks the displayed predictive map information to understand the relationship between their current location and the risk area. When approaching a risk area, they take appropriate action based on audio guidance and visual warnings. The user can select a detour or adjust their driving speed based on the information displayed on the navigation system.

[1085] As a concrete example, traffic accident data and weather data are collected on a server, and analyzed by a generative AI model, resulting in the learning result that "accidents occur frequently during the daytime on rainy days." This information is input to the generative AI model as the following prompt:

[1086] Example prompt sentence:

[1087] Analyze traffic accident patterns based on traffic accident data from January 1, 2023 to July 31, 2023, and weather data for the same period. Learn the tendency for accidents to occur more frequently during the daytime on rainy days, and build a prediction model.

[1088] In this way, predicted traffic accident risk information is delivered to user terminals in real time, which can contribute to reducing traffic accident risks in actual usage environments.

[1089] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1090] Step 1:

[1091] The server collects traffic accident data. This is done using APIs or database connections provided by the police or local government. The server retrieves traffic accident data from, for example, January 1, 2023 to July 31, 2023. The input is traffic accident data, and the output is the collected raw data. This data includes the date, time, location, and details of the accident.

[1092] Step 2:

[1093] The server collects additional weather and road condition data. For this, it uses the API of the Japan Meteorological Agency or weather forecast site. For example, it obtains weather data for the same period. The input is the weather and road condition data, and the output is the collected raw data. This data includes temperature, precipitation, road condition, etc.

[1094] Step 3:

[1095] The server preprocesses the collected traffic accident data and additional data. First, it completes missing values ​​and removes inappropriate data. Next, it unifies the data format and accurately maps date, time, and location information. The input is the collected raw data, and the output is the cleansed data. Specifically, it uses the Python Pandas library to cleanse the data.

[1096] Step 4:

[1097] The server inputs the preprocessed data into a generative AI model to predict the risk of traffic accidents. The generative AI model learns accident patterns from past traffic accident data and weather data and builds a predictive model. The input is the preprocessed data, and the output is risk prediction data. Specifically, the AI ​​model is run using TensorFlow or PyTorch.

[1098] Step 5:

[1099] The server plots the predicted risk data on a map. Specifically, it uses GIS technology to color-code risk areas. The input is risk prediction data, and the output is a prediction map. Specifically, it generates a map using the Folium library and displays it on the map using the Google Maps API.

[1100] Step 6:

[1101] The server delivers the generated predicted map to the device in real time. This is done using an API or WebSocket. The input is the predicted map, and the output is real-time delivery data. Specifically, data is delivered using a web framework such as Flask.

[1102] Step 7:

[1103] The device receives predictive map information from the server via API. The input is real-time data delivered from the server, and the output is predictive map information stored inside the device. Specifically, the navigation system stores the received data in memory.

[1104] Step 8:

[1105] The device displays the received predictive map information on a map or interface. The input is the predictive map information stored in the device, and the output is the visually displayed map information. Specifically, the navigation system displays risk areas in red on the map.

[1106] Step 9:

[1107] The device generates audio guidance and visual warnings when approaching a risk area. The input is the predicted map information and current location information stored in the device, and the output is the audio guidance and visual warning. Specifically, the navigation system issues an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1108] Step 10:

[1109] The user checks the forecast information displayed on the device and understands the risk areas near their current location. The input is the visually displayed map information, and the output is the user's perception. Specifically, the driver checks the navigation screen and confirms the location of the risk area displayed in red.

[1110] Step 11:

[1111] The user adjusts the driving route and speed based on the confirmed information. The input is the risk information recognized by the user, and the output is the adjusted driving behavior. Specifically, the driver follows the navigation system's instructions to select a detour to avoid the risk area.

[1112] Step 12:

[1113] Users receive alerts from their devices and drive more carefully. The input is audio guidance and visual warnings, and the output is safe driving behavior. Specifically, when a driver hears an alert and receives information that there is a risk of an accident at the next intersection, they slow down and continue driving vigilantly.

[1114] 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.

[1115] The traffic accident prediction system of the present invention provides more effective accident prevention information by combining the user's emotion engine. This system is implemented as follows.

[1116] Server Processing

[1117] Data collection

[1118] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[1119] Data Preprocessing

[1120] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[1121] Analysis by generative AI models

[1122] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[1123] Predictive Map Generation

[1124] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[1125] Real-time streaming

[1126] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[1127] Terminal handling

[1128] Receiving information

[1129] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1130] Information display

[1131] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[1132] Alert delivery

[1133] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1134] Emotion engine processing

[1135] Emotional Data Collection

[1136] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera captures the user's face, and the microphone records their voice.

[1137] Emotion analysis

[1138] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. For example, the device may determine that the user is in a "tense state" based on changes in their facial expression.

[1139] Adjusting alert content

[1140] The displayed predictive information and alert content are adjusted based on the results of emotion analysis. For example, if the user is in a tense state, the alert wording will be softer. For example, a message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" will be displayed.

[1141] User Action

[1142] Information confirmation

[1143] Users can check the traffic accident prediction information and adjusted alert messages displayed on the device, which allows them to accurately identify risk areas and high-risk time periods.

[1144] behavior adjustment

[1145] The user can adjust their driving route and speed based on the displayed forecast information and alert messages. For example, to avoid risk areas, the user can select a detour route according to the navigation system's instructions.

[1146] Alert Response

[1147] Users can drive more carefully when they receive alerts and warnings from their devices. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[1148] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[1149] The processing flow will be explained below.

[1150] Server Processing

[1151] Step 1:

[1152] The server receives traffic accident data from the police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the victim's condition.

[1153] Step 2:

[1154] The server obtains weather and road condition data from the Japan Meteorological Agency and weather forecast websites via API. Weather data includes date, time, temperature, precipitation, wind speed, etc.

[1155] Step 3:

[1156] The server stores the acquired traffic accident data and weather data in a database, allowing each data item to be managed in an integrated manner.

[1157] Step 4:

[1158] The server uses data cleaning means to fill in missing values ​​in the collected data, remove unnecessary information, and standardize the format.

[1159] Step 5:

[1160] The server inputs the preprocessed data into a generative AI model, which analyzes and learns patterns of traffic accident occurrence.

[1161] Step 6:

[1162] The server uses a trained generative AI model to predict the risk of traffic accidents based on the day's weather data.

[1163] Step 7:

[1164] The server plots the prediction results on a map using GIS (geographic information system), visually displaying high-risk locations.

[1165] Step 8:

[1166] The server delivers the generated predictive maps to various devices in real time via API or WebSocket.

[1167] Terminal handling

[1168] Step 1:

[1169] The terminal receives forecast information sent in real time from the server via an API.

[1170] Step 2:

[1171] The terminal integrates the received forecast information with map data and stores it in its internal memory.

[1172] Step 3:

[1173] Based on the received data, the terminal displays risk areas on a map, color-coding them (for example, red).

[1174] Step 4:

[1175] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[1176] Emotion engine processing

[1177] Step 1:

[1178] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors.

[1179] Step 2:

[1180] The device inputs the collected data into an emotion engine to analyze the user's emotional state in real time.

[1181] Step 3:

[1182] The device adjusts the content of the displayed forecast information and alerts based on the user's emotional state as determined by the emotion engine.

[1183] User Action

[1184] Step 1:

[1185] The user checks the traffic accident prediction information and the adjusted alert message displayed on the terminal.

[1186] Step 2:

[1187] The user can adjust their driving route and speed based on the displayed forecast information and alert messages.

[1188] Step 3:

[1189] Users will receive alerts and warnings from their devices and drive more carefully.

[1190] Example 2

[1191] 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."

[1192] Conventional traffic accident forecasting systems only identify risk areas based on past traffic accident data and weather data. As a result, they do not reflect the user's emotional state, and therefore may not provide accurate alerts. In particular, when emotional states such as tension and fatigue contribute to accident risk, not taking these data into account is a serious drawback.

[1193] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1194] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for checking traffic accident prediction information and alert messages displayed by the terminal, a driving adjustment means for adjusting the driving route and speed based on the prediction information and alert messages displayed by the terminal, an emotion analysis means for collecting and analyzing user emotion data by the terminal, and an alert adjustment means for adjusting the content of the alert based on the emotion analysis results. This enables the provision of more accurate traffic accident forecasts and alerts that take the user's emotional state into consideration.

[1195] A "traffic accident forecast system" is a system that predicts the risk of traffic accidents and provides users with that information.

[1196] "Data collection means" refers to devices and methods for collecting traffic accident data.

[1197] "Additional data acquisition means" refers to devices and methods for collecting weather and road condition data.

[1198] "Data cleaning means" refers to devices and methods for organizing collected data and removing unnecessary information.

[1199] "Analysis means" refers to devices or methods for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[1200] "Mapping means" refers to a device or method for plotting the analysis results on a map.

[1201] "Data distribution means" refers to an apparatus or method for distributing the generated predictive map to multiple devices in real time.

[1202] "Terminal" refers to a device that displays prediction information and collects user emotion data.

[1203] "Emotion analysis means" refers to a device or method for collecting and analyzing user emotion data.

[1204] "Alert adjustment means" refers to a device or method for adjusting the content of an alert based on the results of emotion analysis.

[1205] "Driving adjustment means" refers to a device or method for adjusting the driving route and speed based on the forecast information and alert messages displayed on the terminal.

[1206] The traffic accident forecasting system of this invention is a system that combines user emotional information to provide more effective accident prevention information. This system functions in cooperation with three parties: a server, a terminal, and a user.

[1207] Server Processing

[1208] Data collection

[1209] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, a detailed dataset will be created by collecting traffic accident and weather data in Tokyo from January 1, 2023 to July 31, 2023.

[1210] Data Preprocessing

[1211] Collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, missing latitude and longitude information in traffic accident data is filled, and weather data is integrated to unify the date and time format.

[1212] Analysis by generative AI models

[1213] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, past accident data is input into the generative AI model, and it learns the pattern that "accidents occur frequently during the daytime on rainy days."

[1214] Predictive Map Generation

[1215] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (e.g., red). For example, based on the output of the generative AI model, areas in Tokyo with a high incidence of traffic accidents are displayed in red on a map.

[1216] Real-time streaming

[1217] The generated predictive map is distributed in real time to various devices (e.g., automobile navigation systems). Distribution is performed via API or WebSocket. For example, the generated predictive map is sent in real time to the navigation system's API.

[1218] Terminal handling

[1219] Receiving information

[1220] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. The navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1221] Information display

[1222] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface. For example, based on the data received by the navigation system, risk areas are displayed in red on a map.

[1223] Alert delivery

[1224] When approaching a high-risk area, audio guidance and visual warnings are provided, such as generating an audio alert saying, "There is an accident risk at the next intersection. Be careful."

[1225] Emotion engine processing

[1226] Emotional Data Collection

[1227] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera scans the user's face and the microphone records their voice.

[1228] Emotion analysis

[1229] The collected user data is input into an emotion engine to analyze the user's emotional state in real time, thereby determining whether the user is tense or relaxed.

[1230] Adjusting alert content

[1231] The displayed forecast information and alerts are adjusted based on the results of emotion analysis. For example, if the user is feeling nervous, a gentler message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" is used.

[1232] User Action

[1233] Information confirmation

[1234] The user checks the traffic accident prediction information and coordinated alert messages displayed on the device, which allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action.

[1235] behavior adjustment

[1236] The user can adjust their driving route and speed based on the displayed forecast information and alert messages, and can select a detour route to avoid risk areas by following the navigation system's instructions.

[1237] Alert Response

[1238] Users can drive more carefully by receiving alerts and warnings from their devices. For example, a driver may hear an alert and receive information that there is a risk of an accident at the next intersection, so they can slow down and continue driving vigilantly.

[1239] Examples of specific examples and prompts

[1240] Specific examples

[1241] The server collects traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, and analyzes the data using a generative AI model. From the analysis results, it learns patterns of accidents occurring frequently during the daytime on rainy days, and identifies and plots risk areas within Tokyo. This information is then sent to the navigation system in real time, where the user can confirm it and take appropriate measures.

[1242] Prompt Sentence Examples

[1243] "Based on traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, the generative AI model identifies areas where accidents are expected to occur frequently during the daytime on rainy days, and displays the results on a map. If the user is in a tense state, the alert message will be softened."

[1244] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[1245] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1246] Step 1: Data collection

[1247] The server collects traffic accident data from police, prosecutors, local governments, etc. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. Specifically, it uses API or database connections to obtain traffic accident and weather data for Tokyo from January 1, 2023 to July 31, 2023. It receives traffic accident data and weather data as input and builds a detailed dataset as output.

[1248] Step 2: Data Preprocessing

[1249] The server preprocesses the collected data. Specifically, it removes unnecessary information and fills in missing values. For example, it fills in missing latitude and longitude information, integrates it with weather data, and standardizes the date and time format. It receives the dataset constructed in step 1 as input and generates cleansed and unified data as output.

[1250] Step 3: Analysis by generative AI model

[1251] The server inputs the preprocessed data into the generative AI model and analyzes traffic accident occurrence patterns. Specifically, it inputs past accident data and weather data into the AI ​​model, and has it learn patterns such as "accidents occur frequently during the daytime on rainy days." It receives the preprocessed data as input and generates a predicted result of the risk of traffic accidents as output.

[1252] Step 4: Generate a prediction map

[1253] The server uses a GIS (geographic information system) to plot the prediction results on a map based on the predictions from the generative AI model. Specifically, it colors the results by, for example, showing high-risk areas in red. It receives the prediction results from the generative AI model as input and generates a visualized prediction map as output.

[1254] Step 5: Real-time delivery

[1255] The server delivers the generated predictive map to the device in real time. Specifically, it sends the predictive map to a navigation system or other device using an API or WebSocket. It receives the predictive map as input and delivers it to the device in real time as output.

[1256] Step 6: Receiving information

[1257] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. Specifically, it receives the forecast data through the API and stores it in its internal memory. It receives the forecast information from the server as input and generates the forecast information stored in its internal memory as output.

[1258] Step 7: Display information

[1259] The device displays the received forecast information on a map or interface. Specifically, it displays risk areas in red on the map to allow the user to visually recognize them. It receives forecast information stored in its internal memory as input and displays visualized forecast information as output.

[1260] Step 8: Alert Delivery

[1261] The device provides audio guidance and visual warnings when approaching a high-risk area. Specifically, it generates an audio alert saying, "There is an accident risk at the next intersection. Please be careful." It receives visualized prediction information as input and generates an alert message as output.

[1262] Step 9: Emotional Data Collection

[1263] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the camera scans the user's face and the microphone records their voice. The device receives the user's emotional data as input and collects emotional data as output.

[1264] Step 10: Sentiment Analysis

[1265] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. Specifically, it determines whether the user is tense or relaxed. It receives emotional data as input and generates analysis results as output.

[1266] Step 11: Adjust the alert content

[1267] The device adjusts the content of the alert based on the emotion analysis results. Specifically, if the user is in a tense state, it displays a message such as, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly." It receives the emotion analysis results as input and generates a tailored alert message as output.

[1268] Step 12: Verify the information

[1269] The user checks the traffic accident prediction information and the adjusted alert message displayed on the device. This allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action. The user receives visualized prediction information and alert messages as input, and checks the information as output.

[1270] Step 13: Behavioral Adjustments

[1271] The user adjusts the driving route and speed based on the displayed forecast information and alert messages. Specifically, the user selects a detour route to avoid risk areas according to the navigation instructions. The user receives confirmed information as input and performs adjusted driving behavior as output.

[1272] Step 14: Respond to alerts

[1273] Users receive alerts and warnings from their devices and drive more carefully. Specifically, when a user hears an alert, they recognize that there is a risk of an accident at the next intersection, slow down, and continue driving vigilantly. The alert message is received as input, and the behavior of driving carefully is output.

[1274] (Application example 2)

[1275] 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."

[1276] There is a need to provide automated driving vehicles and drivers with appropriate and timely information on traffic accident risks. In particular, it is necessary to more effectively promote safe driving by providing alerts and guidance that take into account the driver's emotional state. However, existing systems are insufficient in providing such comprehensive information, and integrating risk information and alerts that respond to the driver's emotional state is a challenge.

[1277] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, an emotion data collection means for collecting user emotion data, an emotion analysis means for analyzing the collected emotion data, and an alert adjustment means for adjusting the display and alert content based on the emotion analysis results. This makes it possible to provide integrated traffic accident risk information and the driver's emotional state.

[1278] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides that information to users.

[1279] The "data collection means" is a device or program that has the function of acquiring traffic accident data.

[1280] The "additional data acquisition means" is a device or program that acquires weather and road condition data.

[1281] The "data cleaning means" is a device or program that has the function of preprocessing the acquired traffic accident data and additional data.

[1282] A "generative AI model" is an artificial intelligence model that performs predictive analysis of traffic accident occurrences based on collected data.

[1283] "Analysis means" refers to a device or program that analyzes preprocessed data using a generative AI model and predicts traffic accidents.

[1284] "Mapping means" refers to a device or program that has the function of plotting the analysis results on a map.

[1285] The "data distribution means" is a device or program that has the function of distributing the generated predictive map to multiple devices in real time.

[1286] "Emotion data collection means" refers to a device or program that has the function of acquiring user emotion data.

[1287] The "emotion analysis means" is a device or program that has the function of analyzing collected emotion data and determining the user's emotional state.

[1288] The "alert adjustment means" is a device or program that has the function of adjusting the display and alert content based on the emotion analysis results.

[1289] "Traffic Accident Data" refers to information relating to past and current traffic accidents.

[1290] "Weather Data" means information regarding current and forecast weather.

[1291] "Road Condition Data" refers to information regarding current and predicted road surface conditions.

[1292] "Plotting on a map" refers to visually displaying the analysis results using a geographic information system (GIS).

[1293] "Delivering in real time" refers to immediately transmitting the generated prediction map to the user device.

[1294] "Adjusting alert content" refers to changing the content and intensity of a warning message taking into account the user's emotional state.

[1295] The traffic accident forecasting system of this invention effectively utilizes various data to provide information useful for predicting and preventing traffic accidents. This system is mainly composed of three main components: a server, a terminal, and a user. Each component and its specific processing are explained below.

[1296] Server Processing

[1297] Data collection

[1298] The server collects traffic accident data and weather and road condition data. Traffic accident data is obtained from police and local government databases, while weather and road condition data is obtained from weather information websites and the Japan Meteorological Agency. These data are collected through APIs and database connections.

[1299] Data Preprocessing

[1300] The collected data is first preprocessed using data cleaning methods, such as imputing missing values, removing unnecessary information, and standardizing the format, to prepare the dataset for analysis.

[1301] Analysis by generative AI models

[1302] The preprocessed data is input into a generative AI model to predict traffic accidents. The generative AI model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents.

[1303] Predictive map generation and delivery

[1304] The prediction results of the generative AI model are plotted on a map using GIS, generating a predictive map. This predictive map is then distributed in real time to multiple devices, such as navigation systems and smartphone applications, via a data distribution method.

[1305] Terminal handling

[1306] Receiving information

[1307] The device receives the forecast information sent from the server, typically via an API.

[1308] Information display

[1309] The received forecast information is displayed on a map or in the UI, providing a visually easy-to-understand interface that makes it easy for users to identify risk areas.

[1310] Emotional data collection and analysis

[1311] The device is equipped with a camera, microphone, and sensors to collect the user's facial expressions, voice, and vital data. This data is analyzed in real time by emotion analysis. For example, if the user is in a tense state, this information can be obtained as an analysis result.

[1312] User Action

[1313] Information confirmation

[1314] The device displays traffic accident prediction information and alert messages tailored to the user's emotional state, allowing the user to instantly identify risk areas and dangerous time periods.

[1315] behavior adjustment

[1316] Users can adjust their driving route and speed based on the displayed forecast information and alert messages, for example by choosing a detour route to avoid risk areas.

[1317] Alert Response

[1318] Users receive alerts and warnings from their devices and strive to drive safely. For example, if they receive a message such as "There is a risk of an accident at the next intersection. Please be careful," they should slow down and be vigilant.

[1319] Specific examples

[1320] The server collects traffic accident data and weather data over a certain period of time and performs data cleaning. A generative AI model is used to learn a predictive pattern, such as "high accident rates during the daytime on rainy days." A risk map is then generated using GIS and sent to the device in real time. The device receives this information and displays risk areas in red on the map. In addition, the camera captures the user's face, and a microphone collects audio data for emotion analysis. If the device determines that the user is in a tense state, it will provide an audio message saying, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly."

[1321] Prompt Sentence Examples

[1322] "This driver is on edge. What kind of caution would work in the next high-accident area?"

[1323] "The current weather is rain, and the location is XX intersection. Please predict the accident risk under these conditions."

[1324] In this way, the traffic accident forecast system incorporates the user's emotional information to provide more effective accident prevention information.

[1325] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1326] Step 1: Collecting traffic accident and weather data

[1327] The server collects traffic accident data and weather data. It obtains traffic accident data from police and local government databases, and weather data from weather information websites and the Japan Meteorological Agency. It collects data through APIs and database connections and stores it. The input is traffic accident data and weather data, and the output is raw data.

[1328] Step 2: Preprocessing the data

[1329] The server preprocesses the collected data by completing missing values, removing unnecessary information, and standardizing the format. The input is raw data, and the output is preprocessed data. Specific operations include cleaning the data, completing missing values, and correcting improper formats.

[1330] Step 3: Analysis by generative AI model

[1331] The server inputs the preprocessed data into the generative AI model to predict traffic accidents. The AI ​​model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents. The input is the preprocessed data, and the output is the traffic accident risk prediction results. Specific operations include inputting data into the AI ​​model, running the model, and obtaining the results.

[1332] Step 4: Generate prediction maps

[1333] The server uses GIS to plot the prediction results of the generative AI model on a map and generate a prediction map. The input is the traffic accident risk prediction result, and the output is the prediction map. Specifically, the server uses GIS software to overlay geographical information and risk information.

[1334] Step 5: Serving the predicted map

[1335] The server distributes the generated predictive map to terminals in real time via a data distribution means. The input is the predictive map, and the output is the data distributed to each terminal. Specifically, the server uses an API or WebSocket to send the predictive map to multiple terminals.

[1336] Step 6: Collect emotion data

[1337] To collect the user's emotional data, the device uses a camera, microphone, and sensors to acquire the user's facial expressions, voice, and vital data. The input is the user's biometric information, and the output is emotional data. Specifically, the device performs processes to capture the user's face, record their voice, and record their vital data.

[1338] Step 7: Analyze the sentiment data

[1339] The device inputs the collected emotional data into an emotion analysis means to determine the user's emotional state. The input is emotional data, and the output is the emotion analysis result. Specifically, the device performs emotion analysis using a deep learning model to determine the user's emotional state as "tense" or "calm."

[1340] Step 8: Adjust and view alerts

[1341] The device adjusts the content of the alert based on the emotion analysis results and notifies the user by display and audio. The input is the emotion analysis results and traffic accident prediction information, and the output is the adjusted alert message. Specifically, the device generates an alert message according to the user's emotional state and notifies the user by visual and audio means.

[1342] Step 9: User Behavior

[1343] The user adjusts their behavior based on the traffic accident prediction information displayed on the device and the adjusted alert message. The input is the adjusted alert message, and the output is the user's behavior. Specific actions include choosing a detour route to avoid risk areas or reducing speed.

[1344] 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.

[1345] 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.

[1346] 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.

[1347] [Fourth embodiment]

[1348] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1349] 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.

[1350] 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).

[1351] 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.

[1352] 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.

[1353] 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).

[1354] 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.

[1355] 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.

[1356] 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.

[1357] 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.

[1358] 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.

[1359] 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.

[1360] 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."

[1361] The traffic accident forecasting system of the present invention is implemented as follows.

[1362] Server Processing

[1363] Data collection

[1364] The server collects traffic accident data from police, prosecutors, local governments, etc. via API or database connection. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[1365] Data Preprocessing

[1366] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[1367] Analysis by generative AI models

[1368] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[1369] Predictive Map Generation

[1370] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[1371] Real-time streaming

[1372] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[1373] Terminal handling

[1374] Receiving information

[1375] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1376] Information display

[1377] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[1378] Alert delivery

[1379] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1380] User Action

[1381] Information confirmation

[1382] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[1383] behavior adjustment

[1384] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[1385] Alert Response

[1386] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[1387] A traffic accident forecast system is realized by this series of processes.

[1388] The processing flow will be explained below.

[1389] Server Processing

[1390] Step 1:

[1391] The server obtains traffic accident data from police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the status of the victims.

[1392] Step 2:

[1393] The server retrieves weather and road condition data from the Japan Meteorological Agency and weather forecast sites, including date, time, temperature, precipitation, wind speed, etc.

[1394] Step 3:

[1395] The server stores the collected traffic accident data and weather data in a database and manages them centrally.

[1396] Step 4:

[1397] The server uses data cleaning techniques to fill in missing values, remove unnecessary information, and standardize formats, for example, filling in missing latitude and longitude information in accident data.

[1398] Step 5:

[1399] The server inputs the preprocessed data into the generative AI model, which then learns traffic accident occurrence patterns.The AI ​​model then analyzes past data and learns the risk of accidents occurring under specific conditions.

[1400] Step 6:

[1401] The server uses a trained generative AI model to input the day's weather data and predict the risk of traffic accidents.

[1402] Step 7:

[1403] The server plots the prediction results on a map using a GIS (geographic information system), which visually displays high-risk locations.

[1404] Step 8:

[1405] The server delivers the created predictive maps to various devices in real time via API or WebSocket.

[1406] Terminal handling

[1407] Step 1:

[1408] A terminal (e.g., a car navigation system) receives forecast information sent from the server via an API.

[1409] Step 2:

[1410] The terminal integrates the received forecast data with map data and stores it in its internal memory.

[1411] Step 3:

[1412] The terminal displays risk areas on the map in different colors (e.g., red), allowing the user to visually identify the risk areas.

[1413] Step 4:

[1414] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[1415] User Action

[1416] Step 1:

[1417] The user checks the traffic accident prediction information displayed on the device, which allows them to understand risk areas and high-risk time periods.

[1418] Step 2:

[1419] The user can adjust their driving route and speed based on the displayed forecast information, for example, by choosing a detour route to avoid risk areas.

[1420] Step 3:

[1421] Users will be able to drive more carefully when they receive alerts and warnings from their devices, for example by slowing down and remaining vigilant when an alert is issued.

[1422] Example 1

[1423] 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."

[1424] Conventional traffic accident prevention systems have had difficulty predicting traffic accidents in advance. In particular, they were unable to adequately consider external factors such as weather and road conditions, limiting the accuracy of their predictions. Furthermore, they were unable to provide real-time information or visually indicate high-risk areas, preventing them from providing useful information to users.

[1425] 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.

[1426] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model to learn traffic accident occurrence patterns and predict traffic accidents, a mapping means for plotting the prediction results on a map using a GIS, and a data distribution means for distributing the generated prediction map to multiple devices in real time via an API or WebSocket. This makes it possible to predict traffic accidents with high accuracy and provide useful information to users in real time.

[1427] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides the results to users.

[1428] A "data collection means" is a method or device for obtaining traffic accident data.

[1429] "Additional data acquisition means" is a method or device for collecting weather and road condition data.

[1430] A "data cleaning means" is a method or device for preprocessing collected data, such as filling in missing values ​​and removing outliers.

[1431] "Analysis means" refers to a method or device that uses a generative AI model to analyze preprocessed data, learn patterns of traffic accident occurrence, and make predictions.

[1432] A "generative AI model" is an artificial intelligence model that learns from collected data and analyzes and predicts traffic accident patterns.

[1433] "Mapping means" refers to a method or device for plotting the analysis results on a map using a geographic information system (GIS).

[1434] "Data distribution means" refers to a method or apparatus for distributing the generated predictive maps to multiple devices in real time via an API or WebSocket.

[1435] The traffic accident forecasting system of the present invention is constructed to accurately predict the occurrence of traffic accidents and provide the information to users in real time. This system includes a data collection means, an additional data acquisition means, a data cleaning means, an analysis means, a mapping means, and a data distribution means.

[1436] Server Processing

[1437] Data collection

[1438] The server collects traffic accident data from police and local governments via API. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast websites. As a specific example, the server obtains traffic accident data from a city's police station from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[1439] Data Preprocessing

[1440] The server cleanses the collected data, fills in missing and outlier values, and converts it into a unified format. It also accurately maps date, time, and location information. For example, the server fills in missing latitude and longitude information in traffic accident data, integrates it with weather data, and unifies the date and time format.

[1441] Analysis by generative AI models

[1442] The server inputs the preprocessed data into a generative AI model and analyzes traffic accident occurrence patterns. The generative AI model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model and makes it learn the pattern that "accidents occur frequently during the daytime on rainy days." An example of a prompt is "Analyze the risk of traffic accidents occurring on rainy days based on past traffic accident data."

[1443] Predictive Map Generation

[1444] The server plots the prediction results on a map using a GIS (geographic information system) based on the prediction results from the generative AI model. High-risk locations are displayed in a different color (for example, red). As a specific example, the server displays areas in a city where traffic accidents are frequent in red on a map based on the output of the generative AI model.

[1445] Real-time streaming

[1446] The server distributes the generated predictive maps to various devices (e.g., automobile navigation systems) in real time. Distribution is typically performed via API or WebSocket. For example, the server transmits the generated predictive maps to the navigation system's API in real time.

[1447] Terminal handling

[1448] Receiving information

[1449] The terminal (for example, a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1450] Information display

[1451] The device displays the received forecast information on a map or in an interface. It provides a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data it receives.

[1452] Alert delivery

[1453] The device generates an alert under certain conditions to warn the user. Specifically, it provides audio guidance and visual warnings when approaching a high-risk area. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1454] User Action

[1455] Information confirmation

[1456] The user checks the forecast information displayed on the device, which allows them to understand risk areas and time periods. For example, a driver checks the navigation screen and realizes that a risk area displayed in red is near their current location.

[1457] behavior adjustment

[1458] The user can adjust their driving route and speed based on the information they have confirmed, allowing them to take actions to avoid risks. For example, a driver may follow the navigation system's instructions to select a detour to avoid a risk area.

[1459] Alert Response

[1460] The user receives an alert from the device and takes action such as driving more carefully. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[1461] In this way, the traffic accident forecasting system can predict the risk of traffic accidents occurring in real time and provide users with immediately useful information.

[1462] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1463] Step 1: Data collection

[1464] Input: Requests to police and local government APIs, the Japan Meteorological Agency, and weather forecast site APIs

[1465] Output: Traffic accident data and weather data are stored on the server.

[1466] Specific operation: The server sends a request to an endpoint such as "https: / / example-police-api.com / data / accidents" or "https: / / example-weather-api.com / data / weather", receives the JSON-formatted data as a response, and then saves it to local storage as " / data / accidents / 2023-01-01_to_2023-07-31.json" or " / data / weather / 2023-01-01_to_2023-07-31.json".

[1467] Step 2: Data Preprocessing

[1468] Input: Traffic accident data and weather data collected in Step 1

[1469] Output: Preprocessed data in a unified format

[1470] What it does: The server cleanses the data and imputes missing and outlier values. For example, it imputes the mean or median for fields marked as "NA." Next, it converts all data to "YYYY-MM-DD HH:MM:SS" format and imputes missing latitude and longitude information using address information.

[1471] Step 3: Analysis by generative AI model

[1472] Input: Preprocessed data from step 2

[1473] Output: Traffic accident prediction results

[1474] Specific operation: The server generates a file called "ai_model_input.csv" and inputs it into the generative AI model. The AI ​​model runs the "train_model()" method and learns patterns from past accident data, such as "accidents tend to occur more frequently during the daytime on rainy days." The prediction results are saved as "predicted_accidents.csv."

[1475] Step 4: Generate a prediction map

[1476] Input: Prediction results obtained in step 3

[1477] Output: A map with risk areas plotted

[1478] Specific operation: The server reads "predicted_accidents.csv" and plots risk areas in red on a map using a GIS (geographic information system). It then executes the function "plot_map('predicted_accidents.csv')" to generate a map image file showing high-risk areas.

[1479] Step 5: Real-time delivery

[1480] Input: Prediction map generated in step 4

[1481] Output: Predictive maps sent to navigation systems via API or WebSocket

[1482] Specific operation: The server executes "send_data_to_api('https: / / example-navigation-api.com / predicted_map', 'predicted_map_data')" and sends the data to the navigation system.

[1483] Step 6: Receiving information (terminal)

[1484] Input: Predictive map data sent from the server

[1485] Output: Predictive map data stored in the device's internal memory

[1486] Specific operation: The device periodically sends a request to the server, executes "fetch(' / predicted_map')" to receive data, and stores it in its internal memory.

[1487] Step 7: Display Information (Terminal)

[1488] Input: Predicted map data received in step 6

[1489] Output: Risk areas displayed in a visually friendly interface

[1490] Specific operation: The device executes the function "display_map(data)" and displays the risk area in red on the map.

[1491] Step 8: Alert delivery (terminal)

[1492] Input: Your device's current location and the risk area information displayed in step 7

[1493] Output: Audio prompts and visual warning alerts

[1494] Specific operation: The device executes "if(current_location in high_risk_area): trigger_alert('There is a risk of an accident at the next intersection. Be careful.')" and issues an audio alert.

[1495] Step 9: User Confirmation

[1496] Input: Risk Area information displayed in step 7

[1497] Output: User who understands risk area and time period

[1498] Specific operation: The user checks the navigation screen and identifies the location of the risk area displayed in red.

[1499] Step 10: Behavioral Adjustment (User)

[1500] Input: Risk area information confirmed in step 9

[1501] Output: Adjusted driving route and speed

[1502] Specific operation: The user follows the re-route guidance provided by the navigation system and selects a safe detour.

[1503] Step 11: Alert response (user)

[1504] Input: The alert received from the terminal in step 8

[1505] Output: More careful driving

[1506] Specific actions: The user hears the alert sound, slows down, and drives safely.

[1507] Through each step in this way, the traffic accident forecasting system is able to predict the risk of traffic accidents occurring in real time and provide users with useful information immediately.

[1508] (Application example 1)

[1509] 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."

[1510] Traffic accidents are an unavoidable problem that seriously impacts human life and property. Even now, as autonomous vehicles become more widespread, the risk of traffic accidents still exists, posing a challenge to the reliability of autonomous driving systems. Conventional navigation systems only provide static map and traffic information, and are unable to predict and respond immediately to changing traffic accident risks in real time. Therefore, there is a need for a method to predict traffic accident risks in real time and efficiently incorporate them into autonomous driving systems.

[1511] 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.

[1512] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for displaying the prediction map information on the navigation system of the autonomous vehicle, an alert generation means for issuing an audio guide or a visual warning when approaching the risk area, and an interface means for allowing the user to check the risk area and adjust their behavior accordingly. This makes it possible to predict traffic accident risks in real time and efficiently reflect the results in the autonomous vehicle.

[1513] The "traffic accident forecast system" is a system that predicts the risk of traffic accidents and encourages appropriate responses.

[1514] "Data collection means" refers to a means for collecting traffic accident data.

[1515] The "additional data acquisition means" is a means for collecting weather and road surface condition data.

[1516] "Data cleaning means" refers to means for preprocessing collected traffic accident data and additional data to make them suitable for analysis.

[1517] "Analysis means" refers to a means for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[1518] "Mapping means" is a means for plotting the analysis results on a map.

[1519] The "data distribution means" is a means for distributing the generated predictive map to multiple devices in real time.

[1520] The "means for displaying on a navigation system" refers to a means for displaying the predicted map information on a navigation system of an autonomous driving vehicle.

[1521] The "alert generating means" is a means for issuing an audio guide or a visual warning when approaching the risk area.

[1522] The "interface means" is a means by which users can check risk areas and adjust their behavior accordingly.

[1523] As an embodiment of the present invention, a traffic accident prediction system includes the following steps: A server collects traffic accident data and acquires weather and road condition data as additional data; the data is preprocessed by a data cleaning means and analyzed using a generative AI model; the analyzed results are plotted on a map by a mapping means; and the predicted map is distributed to multiple devices in real time by a data distribution means.

[1524] The server uses the following specific hardware and software: A REST API is used to collect data, and data collection scripts are executed using the Python language. Weather and road condition data is obtained from external data sources via the API. For data preprocessing, the Python Pandas library is used to fill in missing values ​​and clean the data. For the generative AI model, machine learning libraries such as TensorFlow and PyTorch are used to train the model on past traffic accident data and weather data. For plotting on a map, the Folium library is used, and the prediction results are displayed on the map via the Google Maps API.

[1525] The terminal side acquires the predictive map information using a means to display it on the navigation system and provides the information visually to the user. This interface has an alert generation means that issues audio guidance and visual warnings when approaching a risk area. This alert generation uses WebSocket and API to receive data distribution in real time.

[1526] The user checks the displayed predictive map information to understand the relationship between their current location and the risk area. When approaching a risk area, they take appropriate action based on audio guidance and visual warnings. The user can select a detour or adjust their driving speed based on the information displayed on the navigation system.

[1527] As a concrete example, traffic accident data and weather data are collected on a server, and analyzed by a generative AI model, resulting in the learning result that "accidents occur frequently during the daytime on rainy days." This information is input to the generative AI model as the following prompt:

[1528] Example prompt sentence:

[1529] Analyze traffic accident patterns based on traffic accident data from January 1, 2023 to July 31, 2023, and weather data for the same period. Learn the tendency for accidents to occur more frequently during the daytime on rainy days, and build a prediction model.

[1530] In this way, predicted traffic accident risk information is delivered to user terminals in real time, which can contribute to reducing traffic accident risks in actual usage environments.

[1531] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1532] Step 1:

[1533] The server collects traffic accident data. This is done using APIs or database connections provided by the police or local government. The server retrieves traffic accident data from, for example, January 1, 2023 to July 31, 2023. The input is traffic accident data, and the output is the collected raw data. This data includes the date, time, location, and details of the accident.

[1534] Step 2:

[1535] The server collects additional weather and road condition data. For this, it uses the API of the Japan Meteorological Agency or weather forecast site. For example, it obtains weather data for the same period. The input is the weather and road condition data, and the output is the collected raw data. This data includes temperature, precipitation, road condition, etc.

[1536] Step 3:

[1537] The server preprocesses the collected traffic accident data and additional data. First, it completes missing values ​​and removes inappropriate data. Next, it unifies the data format and accurately maps date, time, and location information. The input is the collected raw data, and the output is the cleansed data. Specifically, it uses the Python Pandas library to cleanse the data.

[1538] Step 4:

[1539] The server inputs the preprocessed data into a generative AI model to predict the risk of traffic accidents. The generative AI model learns accident patterns from past traffic accident data and weather data and builds a predictive model. The input is the preprocessed data, and the output is risk prediction data. Specifically, the AI ​​model is run using TensorFlow or PyTorch.

[1540] Step 5:

[1541] The server plots the predicted risk data on a map. Specifically, it uses GIS technology to color-code risk areas. The input is risk prediction data, and the output is a prediction map. Specifically, it generates a map using the Folium library and displays it on the map using the Google Maps API.

[1542] Step 6:

[1543] The server delivers the generated predicted map to the device in real time. This is done using an API or WebSocket. The input is the predicted map, and the output is real-time delivery data. Specifically, data is delivered using a web framework such as Flask.

[1544] Step 7:

[1545] The device receives predictive map information from the server via API. The input is real-time data delivered from the server, and the output is predictive map information stored inside the device. Specifically, the navigation system stores the received data in memory.

[1546] Step 8:

[1547] The device displays the received predictive map information on a map or interface. The input is the predictive map information stored in the device, and the output is the visually displayed map information. Specifically, the navigation system displays risk areas in red on the map.

[1548] Step 9:

[1549] The device generates audio guidance and visual warnings when approaching a risk area. The input is the predicted map information and current location information stored in the device, and the output is the audio guidance and visual warning. Specifically, the navigation system issues an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1550] Step 10:

[1551] The user checks the forecast information displayed on the device and understands the risk areas near their current location. The input is the visually displayed map information, and the output is the user's perception. Specifically, the driver checks the navigation screen and confirms the location of the risk area displayed in red.

[1552] Step 11:

[1553] The user adjusts the driving route and speed based on the confirmed information. The input is the risk information recognized by the user, and the output is the adjusted driving behavior. Specifically, the driver follows the navigation system's instructions to select a detour to avoid the risk area.

[1554] Step 12:

[1555] Users receive alerts from their devices and drive more carefully. The input is audio guidance and visual warnings, and the output is safe driving behavior. Specifically, when a driver hears an alert and receives information that there is a risk of an accident at the next intersection, they slow down and continue driving vigilantly.

[1556] 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.

[1557] The traffic accident prediction system of the present invention provides more effective accident prevention information by combining the user's emotion engine. This system is implemented as follows.

[1558] Server Processing

[1559] Data collection

[1560] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, the server obtains traffic accident data from police stations in Tokyo from January 1, 2023 to July 31, 2023, as well as weather data for the same period. This creates a detailed dataset.

[1561] Data Preprocessing

[1562] The collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, the server fills in missing latitude and longitude information in traffic accident data, and integrates it with weather data to unify the date and time format.

[1563] Analysis by generative AI models

[1564] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, the server inputs past accident data into the generative AI model, and has it learn the pattern that "accidents occur frequently during the daytime on rainy days."

[1565] Predictive Map Generation

[1566] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (for example, red). For example, based on the output of the generative AI model, the server displays areas in Tokyo with a high incidence of traffic accidents in red on a map.

[1567] Real-time streaming

[1568] The generated predictive map is distributed to various devices (e.g., automobile navigation systems) in real time. Distribution is generally performed via API or WebSocket. For example, the server sends the generated predictive map to the navigation system's API in real time.

[1569] Terminal handling

[1570] Receiving information

[1571] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. For example, the navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1572] Information display

[1573] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface so that users can check the map. For example, a navigation system displays risk areas in red on a map based on the data received.

[1574] Alert delivery

[1575] Under certain conditions, an alert is generated to warn the user. Specifically, when approaching a high-risk area, audio guidance and visual warnings are given. For example, the navigation system may issue an audio alert saying, "There is a risk of an accident at the next intersection. Please be careful."

[1576] Emotion engine processing

[1577] Emotional Data Collection

[1578] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera captures the user's face, and the microphone records their voice.

[1579] Emotion analysis

[1580] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. For example, the device may determine that the user is in a "tense state" based on changes in their facial expression.

[1581] Adjusting alert content

[1582] The displayed predictive information and alert content are adjusted based on the results of emotion analysis. For example, if the user is in a tense state, the alert wording will be softer. For example, a message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" will be displayed.

[1583] User Action

[1584] Information confirmation

[1585] Users can check the traffic accident prediction information and adjusted alert messages displayed on the device, which allows them to accurately identify risk areas and high-risk time periods.

[1586] behavior adjustment

[1587] The user can adjust their driving route and speed based on the displayed forecast information and alert messages. For example, to avoid risk areas, the user can select a detour route according to the navigation system's instructions.

[1588] Alert Response

[1589] Users can drive more carefully when they receive alerts and warnings from their devices. For example, a driver hears an alert and receives information that there is a risk of an accident at the next intersection, so they slow down and continue driving vigilantly.

[1590] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[1591] The processing flow will be explained below.

[1592] Server Processing

[1593] Step 1:

[1594] The server receives traffic accident data from the police and prosecutors via API, including the date and time of the accident, location, cause, participating vehicles, and the victim's condition.

[1595] Step 2:

[1596] The server obtains weather and road condition data from the Japan Meteorological Agency and weather forecast websites via API. Weather data includes date, time, temperature, precipitation, wind speed, etc.

[1597] Step 3:

[1598] The server stores the acquired traffic accident data and weather data in a database, allowing each data item to be managed in an integrated manner.

[1599] Step 4:

[1600] The server uses data cleaning means to fill in missing values ​​in the collected data, remove unnecessary information, and standardize the format.

[1601] Step 5:

[1602] The server inputs the preprocessed data into a generative AI model, which analyzes and learns patterns of traffic accident occurrence.

[1603] Step 6:

[1604] The server uses a trained generative AI model to predict the risk of traffic accidents based on the day's weather data.

[1605] Step 7:

[1606] The server plots the prediction results on a map using GIS (geographic information system), visually displaying high-risk locations.

[1607] Step 8:

[1608] The server delivers the generated predictive maps to various devices in real time via API or WebSocket.

[1609] Terminal handling

[1610] Step 1:

[1611] The terminal receives forecast information sent in real time from the server via an API.

[1612] Step 2:

[1613] The terminal integrates the received forecast information with map data and stores it in its internal memory.

[1614] Step 3:

[1615] Based on the received data, the terminal displays risk areas on a map, color-coding them (for example, red).

[1616] Step 4:

[1617] The device generates audio and visual warnings to alert users when they approach a high-risk area.

[1618] Emotion engine processing

[1619] Step 1:

[1620] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors.

[1621] Step 2:

[1622] The device inputs the collected data into an emotion engine to analyze the user's emotional state in real time.

[1623] Step 3:

[1624] The device adjusts the content of the displayed forecast information and alerts based on the user's emotional state as determined by the emotion engine.

[1625] User Action

[1626] Step 1:

[1627] The user checks the traffic accident prediction information and the adjusted alert message displayed on the terminal.

[1628] Step 2:

[1629] The user can adjust their driving route and speed based on the displayed forecast information and alert messages.

[1630] Step 3:

[1631] Users will receive alerts and warnings from their devices and drive more carefully.

[1632] Example 2

[1633] 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."

[1634] Conventional traffic accident forecasting systems only identify risk areas based on past traffic accident data and weather data. As a result, they do not reflect the user's emotional state, and therefore may not provide accurate alerts. In particular, when emotional states such as tension and fatigue contribute to accident risk, not taking these data into account is a serious drawback.

[1635] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1636] In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road surface condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, a means for checking traffic accident prediction information and alert messages displayed by the terminal, a driving adjustment means for adjusting the driving route and speed based on the prediction information and alert messages displayed by the terminal, an emotion analysis means for collecting and analyzing user emotion data by the terminal, and an alert adjustment means for adjusting the content of the alert based on the emotion analysis results. This enables the provision of more accurate traffic accident forecasts and alerts that take the user's emotional state into consideration.

[1637] A "traffic accident forecast system" is a system that predicts the risk of traffic accidents and provides users with that information.

[1638] "Data collection means" refers to devices and methods for collecting traffic accident data.

[1639] "Additional data acquisition means" refers to devices and methods for collecting weather and road condition data.

[1640] "Data cleaning means" refers to devices and methods for organizing collected data and removing unnecessary information.

[1641] "Analysis means" refers to devices or methods for analyzing preprocessed data using a generative AI model and predicting traffic accidents.

[1642] "Mapping means" refers to a device or method for plotting the analysis results on a map.

[1643] "Data distribution means" refers to an apparatus or method for distributing the generated predictive map to multiple devices in real time.

[1644] "Terminal" refers to a device that displays prediction information and collects user emotion data.

[1645] "Emotion analysis means" refers to a device or method for collecting and analyzing user emotion data.

[1646] "Alert adjustment means" refers to a device or method for adjusting the content of an alert based on the results of emotion analysis.

[1647] "Driving adjustment means" refers to a device or method for adjusting the driving route and speed based on the forecast information and alert messages displayed on the terminal.

[1648] The traffic accident forecasting system of this invention is a system that combines user emotional information to provide more effective accident prevention information. This system functions in cooperation with three parties: a server, a terminal, and a user.

[1649] Server Processing

[1650] Data collection

[1651] The server collects traffic accident data from police, prosecutors, and local governments via API or database connections. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. As a specific example, a detailed dataset will be created by collecting traffic accident and weather data in Tokyo from January 1, 2023 to July 31, 2023.

[1652] Data Preprocessing

[1653] Collected data is cleansed, missing values ​​are filled, and unnecessary information is removed. It is then converted into a unified format, and date, time, and location information are accurately mapped. For example, missing latitude and longitude information in traffic accident data is filled, and weather data is integrated to unify the date and time format.

[1654] Analysis by generative AI models

[1655] The preprocessed data is input into a generative AI model to analyze traffic accident occurrence patterns. The AI ​​model learns the frequency and conditions of traffic accidents from past data. As a specific example, past accident data is input into the generative AI model, and it learns the pattern that "accidents occur frequently during the daytime on rainy days."

[1656] Predictive Map Generation

[1657] Based on the prediction results from the generative AI model, the prediction results are plotted on a map using a GIS (geographic information system). High-risk locations are displayed in a different color (e.g., red). For example, based on the output of the generative AI model, areas in Tokyo with a high incidence of traffic accidents are displayed in red on a map.

[1658] Real-time streaming

[1659] The generated predictive map is distributed in real time to various devices (e.g., automobile navigation systems). Distribution is performed via API or WebSocket. For example, the generated predictive map is sent in real time to the navigation system's API.

[1660] Terminal handling

[1661] Receiving information

[1662] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. The navigation system receives the forecast data sent from the server via the API and stores the data in its internal memory.

[1663] Information display

[1664] The received forecast information is displayed on a map or in an interface. It is provided in a visually easy-to-understand interface. For example, based on the data received by the navigation system, risk areas are displayed in red on a map.

[1665] Alert delivery

[1666] When approaching a high-risk area, audio guidance and visual warnings are provided, such as generating an audio alert saying, "There is an accident risk at the next intersection. Be careful."

[1667] Emotion engine processing

[1668] Emotional Data Collection

[1669] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the navigation system's camera scans the user's face and the microphone records their voice.

[1670] Emotion analysis

[1671] The collected user data is input into an emotion engine to analyze the user's emotional state in real time, thereby determining whether the user is tense or relaxed.

[1672] Adjusting alert content

[1673] The displayed forecast information and alerts are adjusted based on the results of emotion analysis. For example, if the user is feeling nervous, a gentler message such as "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly" is used.

[1674] User Action

[1675] Information confirmation

[1676] The user checks the traffic accident prediction information and coordinated alert messages displayed on the device, which allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action.

[1677] behavior adjustment

[1678] The user can adjust their driving route and speed based on the displayed forecast information and alert messages, and can select a detour route to avoid risk areas by following the navigation system's instructions.

[1679] Alert Response

[1680] Users can drive more carefully by receiving alerts and warnings from their devices. For example, a driver may hear an alert and receive information that there is a risk of an accident at the next intersection, so they can slow down and continue driving vigilantly.

[1681] Examples of specific examples and prompts

[1682] Specific examples

[1683] The server collects traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, and analyzes the data using a generative AI model. From the analysis results, it learns patterns of accidents occurring frequently during the daytime on rainy days, and identifies and plots risk areas within Tokyo. This information is then sent to the navigation system in real time, where the user can confirm it and take appropriate measures.

[1684] Prompt Sentence Examples

[1685] "Based on traffic accident data and weather data for Tokyo from January 1, 2023 to July 31, 2023, the generative AI model identifies areas where accidents are expected to occur frequently during the daytime on rainy days, and displays the results on a map. If the user is in a tense state, the alert message will be softened."

[1686] In this way, the traffic accident forecasting system can incorporate the user's emotional information to provide more effective accident prevention information.

[1687] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1688] Step 1: Data collection

[1689] The server collects traffic accident data from police, prosecutors, local governments, etc. It also collects weather and road condition data from external data sources such as the Japan Meteorological Agency and weather forecast sites. Specifically, it uses API or database connections to obtain traffic accident and weather data for Tokyo from January 1, 2023 to July 31, 2023. It receives traffic accident data and weather data as input and builds a detailed dataset as output.

[1690] Step 2: Data Preprocessing

[1691] The server preprocesses the collected data. Specifically, it removes unnecessary information and fills in missing values. For example, it fills in missing latitude and longitude information, integrates it with weather data, and standardizes the date and time format. It receives the dataset constructed in step 1 as input and generates cleansed and unified data as output.

[1692] Step 3: Analysis by generative AI model

[1693] The server inputs the preprocessed data into the generative AI model and analyzes traffic accident occurrence patterns. Specifically, it inputs past accident data and weather data into the AI ​​model, and has it learn patterns such as "accidents occur frequently during the daytime on rainy days." It receives the preprocessed data as input and generates a predicted result of the risk of traffic accidents as output.

[1694] Step 4: Generate a prediction map

[1695] The server uses a GIS (geographic information system) to plot the prediction results on a map based on the predictions from the generative AI model. Specifically, it colors the results by, for example, showing high-risk areas in red. It receives the prediction results from the generative AI model as input and generates a visualized prediction map as output.

[1696] Step 5: Real-time delivery

[1697] The server delivers the generated predictive map to the device in real time. Specifically, it sends the predictive map to a navigation system or other device using an API or WebSocket. It receives the predictive map as input and delivers it to the device in real time as output.

[1698] Step 6: Receiving information

[1699] The terminal (e.g., a car navigation system) receives the forecast information sent from the server via the API. Specifically, it receives the forecast data through the API and stores it in its internal memory. It receives the forecast information from the server as input and generates the forecast information stored in its internal memory as output.

[1700] Step 7: Display information

[1701] The device displays the received forecast information on a map or interface. Specifically, it displays risk areas in red on the map to allow the user to visually recognize them. It receives forecast information stored in its internal memory as input and displays visualized forecast information as output.

[1702] Step 8: Alert Delivery

[1703] The device provides audio guidance and visual warnings when approaching a high-risk area. Specifically, it generates an audio alert saying, "There is an accident risk at the next intersection. Please be careful." It receives visualized prediction information as input and generates an alert message as output.

[1704] Step 9: Emotional Data Collection

[1705] The device collects the user's facial expressions, voice, and vital data using a camera, microphone, and sensors. For example, the camera scans the user's face and the microphone records their voice. The device receives the user's emotional data as input and collects emotional data as output.

[1706] Step 10: Sentiment Analysis

[1707] The device inputs the collected user data into an emotion engine and analyzes the user's emotional state in real time. Specifically, it determines whether the user is tense or relaxed. It receives emotional data as input and generates analysis results as output.

[1708] Step 11: Adjust the alert content

[1709] The device adjusts the content of the alert based on the emotion analysis results. Specifically, if the user is in a tense state, it displays a message such as, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly." It receives the emotion analysis results as input and generates a tailored alert message as output.

[1710] Step 12: Verify the information

[1711] The user checks the traffic accident prediction information and the adjusted alert message displayed on the device. This allows the user to accurately identify risk areas and high-risk time periods, and serves as a guide for taking appropriate action. The user receives visualized prediction information and alert messages as input, and checks the information as output.

[1712] Step 13: Behavioral Adjustments

[1713] The user adjusts the driving route and speed based on the displayed forecast information and alert messages. Specifically, the user selects a detour route to avoid risk areas according to the navigation instructions. The user receives confirmed information as input and performs adjusted driving behavior as output.

[1714] Step 14: Respond to alerts

[1715] Users receive alerts and warnings from their devices and drive more carefully. Specifically, when a user hears an alert, they recognize that there is a risk of an accident at the next intersection, slow down, and continue driving vigilantly. The alert message is received as input, and the behavior of driving carefully is output.

[1716] (Application example 2)

[1717] 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."

[1718] There is a need to provide automated driving vehicles and drivers with appropriate and timely information on traffic accident risks. In particular, it is necessary to more effectively promote safe driving by providing alerts and guidance that take into account the driver's emotional state. However, existing systems are insufficient in providing such comprehensive information, and integrating risk information and alerts that respond to the driver's emotional state is a challenge.

[1719] The identification processing by the identification processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means. In this invention, the server includes a data collection means for collecting traffic accident data, an additional data acquisition means for collecting weather and road condition data, a data cleaning means for preprocessing the traffic accident data and the additional data, an analysis means for analyzing the preprocessed data using a generative AI model and predicting traffic accidents, a mapping means for plotting the analysis results on a map, a data distribution means for distributing the generated prediction map to multiple devices in real time, an emotion data collection means for collecting user emotion data, an emotion analysis means for analyzing the collected emotion data, and an alert adjustment means for adjusting the display and alert content based on the emotion analysis results. This makes it possible to provide integrated traffic accident risk information and the driver's emotional state.

[1720] A "traffic accident forecast system" is a system that predicts the occurrence of traffic accidents and provides that information to users.

[1721] The "data collection means" is a device or program that has the function of acquiring traffic accident data.

[1722] The "additional data acquisition means" is a device or program that acquires weather and road condition data.

[1723] The "data cleaning means" is a device or program that has the function of preprocessing the acquired traffic accident data and additional data.

[1724] A "generative AI model" is an artificial intelligence model that performs predictive analysis of traffic accident occurrences based on collected data.

[1725] "Analysis means" refers to a device or program that analyzes preprocessed data using a generative AI model and predicts traffic accidents.

[1726] "Mapping means" refers to a device or program that has the function of plotting the analysis results on a map.

[1727] The "data distribution means" is a device or program that has the function of distributing the generated predictive map to multiple devices in real time.

[1728] "Emotion data collection means" refers to a device or program that has the function of acquiring user emotion data.

[1729] The "emotion analysis means" is a device or program that has the function of analyzing collected emotion data and determining the user's emotional state.

[1730] The "alert adjustment means" is a device or program that has the function of adjusting the display and alert content based on the emotion analysis results.

[1731] "Traffic Accident Data" refers to information relating to past and current traffic accidents.

[1732] "Weather Data" means information regarding current and forecast weather.

[1733] "Road Condition Data" refers to information regarding current and predicted road surface conditions.

[1734] "Plotting on a map" refers to visually displaying the analysis results using a geographic information system (GIS).

[1735] "Delivering in real time" refers to immediately transmitting the generated prediction map to the user device.

[1736] "Adjusting alert content" refers to changing the content and intensity of a warning message taking into account the user's emotional state.

[1737] The traffic accident forecasting system of this invention effectively utilizes various data to provide information useful for predicting and preventing traffic accidents. This system is mainly composed of three main components: a server, a terminal, and a user. Each component and its specific processing are explained below.

[1738] Server Processing

[1739] Data collection

[1740] The server collects traffic accident data and weather and road condition data. Traffic accident data is obtained from police and local government databases, while weather and road condition data is obtained from weather information websites and the Japan Meteorological Agency. These data are collected through APIs and database connections.

[1741] Data Preprocessing

[1742] The collected data is first preprocessed using data cleaning methods, such as imputing missing values, removing unnecessary information, and standardizing the format, to prepare the dataset for analysis.

[1743] Analysis by generative AI models

[1744] The preprocessed data is input into a generative AI model to predict traffic accidents. The generative AI model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents.

[1745] Predictive map generation and delivery

[1746] The prediction results of the generative AI model are plotted on a map using GIS, generating a predictive map. This predictive map is then distributed in real time to multiple devices, such as navigation systems and smartphone applications, via a data distribution method.

[1747] Terminal handling

[1748] Receiving information

[1749] The device receives the forecast information sent from the server, typically via an API.

[1750] Information display

[1751] The received forecast information is displayed on a map or in the UI, providing a visually easy-to-understand interface that makes it easy for users to identify risk areas.

[1752] Emotional data collection and analysis

[1753] The device is equipped with a camera, microphone, and sensors to collect the user's facial expressions, voice, and vital data. This data is analyzed in real time by emotion analysis. For example, if the user is in a tense state, this information can be obtained as an analysis result.

[1754] User Action

[1755] Information confirmation

[1756] The device displays traffic accident prediction information and alert messages tailored to the user's emotional state, allowing the user to instantly identify risk areas and dangerous time periods.

[1757] behavior adjustment

[1758] Users can adjust their driving route and speed based on the displayed forecast information and alert messages, for example by choosing a detour route to avoid risk areas.

[1759] Alert Response

[1760] Users receive alerts and warnings from their devices and strive to drive safely. For example, if they receive a message such as "There is a risk of an accident at the next intersection. Please be careful," they should slow down and be vigilant.

[1761] Specific examples

[1762] The server collects traffic accident data and weather data over a certain period of time and performs data cleaning. A generative AI model is used to learn a predictive pattern, such as "high accident rates during the daytime on rainy days." A risk map is then generated using GIS and sent to the device in real time. The device receives this information and displays risk areas in red on the map. In addition, the camera captures the user's face, and a microphone collects audio data for emotion analysis. If the device determines that the user is in a tense state, it will provide an audio message saying, "There is a risk of an accident at the next intersection. Please take a deep breath and drive calmly."

[1763] Prompt Sentence Examples

[1764] "This driver is on edge. What kind of caution would work in the next high-accident area?"

[1765] "The current weather is rain, and the location is XX intersection. Please predict the accident risk under these conditions."

[1766] In this way, the traffic accident forecast system incorporates the user's emotional information to provide more effective accident prevention information.

[1767] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1768] Step 1: Collecting traffic accident and weather data

[1769] The server collects traffic accident data and weather data. It obtains traffic accident data from police and local government databases, and weather data from weather information websites and the Japan Meteorological Agency. It collects data through APIs and database connections and stores it. The input is traffic accident data and weather data, and the output is raw data.

[1770] Step 2: Preprocessing the data

[1771] The server preprocesses the collected data by completing missing values, removing unnecessary information, and standardizing the format. The input is raw data, and the output is preprocessed data. Specific operations include cleaning the data, completing missing values, and correcting improper formats.

[1772] Step 3: Analysis by generative AI model

[1773] The server inputs the preprocessed data into the generative AI model to predict traffic accidents. The AI ​​model learns traffic accident occurrence patterns from past data and calculates the risk of traffic accidents. The input is the preprocessed data, and the output is the traffic accident risk prediction results. Specific operations include inputting data into the AI ​​model, running the model, and obtaining the results.

[1774] Step 4: Generate prediction maps

[1775] The server uses GIS to plot the prediction results of the generative AI model on a map and generate a prediction map. The input is the traffic accident risk prediction result, and the output is the prediction map. Specifically, the server uses GIS software to overlay geographical information and risk information.

[1776] Step 5: Serving the predicted map

[1777] The server distributes the generated predictive map to terminals in real time via a data distribution means. The input is the predictive map, and the output is the data distributed to each terminal. Specifically, the server uses an API or WebSocket to send the predictive map to multiple terminals.

[1778] Step 6: Collect emotion data

[1779] To collect the user's emotional data, the device uses a camera, microphone, and sensors to acquire the user's facial expressions, voice, and vital data. The input is the user's biometric information, and the output is emotional data. Specifically, the device performs processes to capture the user's face, record their voice, and record their vital data.

[1780] Step 7: Analyze the sentiment data

[1781] The device inputs the collected emotional data into an emotion analysis means to determine the user's emotional state. The input is emotional data, and the output is the emotion analysis result. Specifically, the device performs emotion analysis using a deep learning model to determine the user's emotional state as "tense" or "calm."

[1782] Step 8: Adjust and view alerts

[1783] The device adjusts the content of the alert based on the emotion analysis results and notifies the user by display and audio. The input is the emotion analysis results and traffic accident prediction information, and the output is the adjusted alert message. Specifically, the device generates an alert message according to the user's emotional state and notifies the user by visual and audio means.

[1784] Step 9: User Behavior

[1785] The user adjusts their behavior based on the traffic accident prediction information displayed on the device and the adjusted alert message. The input is the adjusted alert message, and the output is the user's behavior. Specific actions include choosing a detour route to avoid risk areas or reducing speed.

[1786] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the control target 443 to output the result of the specific processing. The microphone 238 acquires voice indicating a user input regarding the result of the specific processing. The control unit 46A transm...

Claims

1. A traffic accident forecasting system, a data collection means for collecting traffic accident data; additional data acquisition means for collecting weather and road condition data; a data cleaning means for preprocessing the traffic accident data and additional data; An analytical means for analyzing the preprocessed data using the generative AI model and predicting traffic accidents; a mapping means for plotting the analysis results on a map; a data distribution means for distributing the generated predicted map to a plurality of devices in real time; A system including:

2. 2. The system according to claim 1, wherein the data collection means acquires traffic accident information from the police and prosecutors.

3. 2. The system according to claim 1, wherein the analysis means predicts the risk of a traffic accident occurring by learning past traffic accident data and weather data.

4. The system according to claim 1 , wherein the data distribution means distributes the predicted map through an API or WebSocket.

5. 2. The system of claim 1, wherein said mapping means visually displays high-risk locations using a GIS (geographic information system).

6. The system of claim 1 , wherein the device comprises an automobile navigation system.

Images