Traffic accident risk evaluation system

Abstract

To provide a traffic accident risk evaluation system capable of evaluating an accident risk at an arbitrary point or in an arbitrary situation.SOLUTION: A traffic accident risk evaluation system includes: a situation selection unit which accepts selection of a road of an evaluation object; a condition setting unit which sets analysis conditions including occupant (driver and passenger) information, weather, own vehicle and counterpart information; a vehicle damage level estimation unit which estimates a vehicle damage level when an accident occurs in at least one position on the road or in a situation from the selected road environment and the analysis conditions; and an occupant injury estimation unit which estimates the severity of injury to the vehicle occupant at occurrence of the accident at the point or the situation from the vehicle damage level and the occupant information.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a traffic accident risk assessment system. [Background technology]

[0002] Conventionally, accidents have been predicted and warnings issued using road traffic conditions and past accident data, etc. For example, a method has been proposed in which an accident occurrence rate table is created by combining the past traffic conditions and the number of past accidents at a certain evaluation point, and the accident occurrence rate table is combined with the current traffic conditions to calculate the current accident occurrence rate (likelihood of an accident occurring) (see, for example, Patent Document 1).

[0003] However, this method cannot calculate the accident occurrence rate for points where traffic data is not available, so there are some areas where it cannot be used.In addition, even for points where data is available, while it can quantify the accident occurrence rate, it is not possible to calculate the damage (risk) such as the degree of vehicle damage or the extent of injuries in the event of an accident. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-189597 Summary of the Invention [Problem to be solved by the invention]

[0005] An object of the present invention is to provide a traffic accident risk assessment system capable of assessing the risk of an accident at any location or in any situation. [Means for solving the problem]

[0006] The traffic accident risk assessment system of the present invention comprises a situation selection unit that accepts the selection of a road to be evaluated, a condition setting unit that sets analysis conditions including occupant (driver and passenger) information, weather, the vehicle and the other party information, a vehicle damage degree estimation unit that estimates the degree of vehicle damage in the event of an accident at at least one point or situation on the road from the road environment of the selected road and the analysis conditions, and an occupant injury estimation unit that estimates the extent of trauma to vehicle occupants in the event of an accident at the point or situation from the vehicle damage degree and the occupant information.

[0007] In one aspect of the present invention, the vehicle safety system further includes an accident risk map creator that synthesizes the severity of injury to the vehicle occupant estimated for each of the multiple points into a map to create an accident risk map.

[0008] In one aspect of the present invention, the vehicle damage degree estimation unit estimates the vehicle damage degree using a vehicle damage degree estimation model constructed by machine learning with a combination of road environment, occupant information, weather, and vehicle as explanatory variables and the vehicle damage degree as a target variable.

[0009] In one aspect of the present invention, the occupant injury estimation unit estimates the extent of trauma to the vehicle occupant using an occupant injury estimation model constructed by machine learning with a combination of occupant information and the degree of vehicle damage as explanatory variables and the extent of the occupant injury as a target variable.

[0010] In one aspect of the present invention, the vehicle further includes an accident risk assessment unit that estimates an overall accident risk assessment based on the degree of vehicle damage estimated by the vehicle damage degree estimation unit and the degree of injury to the vehicle occupant estimated by the occupant injury estimation unit. Effect of the Invention

[0011] According to the present invention, it is possible to evaluate the risk of an accident at any location or in any situation, including locations where no accidents have occurred. [Brief description of the drawings]

[0012] [Figure 1] 1 is a schematic configuration diagram of a traffic accident risk assessment system according to an embodiment. [Diagram 2] FIG. 2 is a functional block diagram of a computing device. [Diagram 3] FIG. 13 is a diagram showing an example of an accident risk map. [Figure 4] FIG. 13 is a diagram showing an example of an accident risk map. [Diagram 5] FIG. 13 is a diagram illustrating an example of a collision mode. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0014] As shown in FIG. 1, the traffic accident risk assessment system according to the present embodiment includes a calculation device 1, an input unit 2, and an output unit 3. When factors related to traffic accidents are input, the calculation device 1 estimates the accident risk (such as the degree of vehicle damage and the severity of injuries) at any point and / or in any situation on the road to be assessed, and obtains an accident risk assessment value and creates an accident risk map. The input unit 2 is an input interface for information obtained from a mouse, a keyboard, or other device, and is used to input analysis conditions including personal information of vehicle occupants (such as height, weight, sex, age, underlying disease, and seat position). The output unit 3 displays the accident risk assessment value estimated by the calculation device 1 and the created accident risk map, and may be, for example, a liquid crystal display. The input unit 2 and the output unit 3 may be touch panels. The output unit 3 may output the accident risk assessment value and the accident risk map to other devices.

[0015] The arithmetic device 1 is configured by, for example, a computer or an external server equipped with hardware such as a CPU, a communication module, and a storage unit. The storage unit includes, for example, a RAM, a ROM, a hard disk, or an external server. The arithmetic device 1 may be a computer system including multiple computers. The arithmetic device 1 may also be configured by cloud computing.

[0016] As shown in Fig. 2, the calculation device 1 includes a model generation system 10 and an accident risk estimation system 20. The model generation system 10 includes a learning data acquisition unit 11, a vehicle damage degree estimation model generation unit 12, and an occupant injury estimation model generation unit 13. The accident risk estimation system 20 includes a situation selection unit 21, a condition setting unit 22, a vehicle damage degree estimation unit 23, an occupant injury estimation unit 24, an accident risk assessment unit 25, and an accident risk map creation unit 26. The functions of each unit of the calculation device 1 are realized by the CPU executing a program.

[0017] The vehicle damage estimation model generation unit 12 of the model generation system 10 generates a vehicle damage estimation model, which is a trained model. Also, the occupant injury estimation model generation unit 13 generates an occupant injury estimation model, which is a trained model.

[0018] The vehicle damage degree estimation model is a model that inputs road environment, weather, the subject vehicle, other party information, driver information, etc., and outputs the vehicle damage degree when an accident occurs. Examples of other party information include automobiles, motorcycles, bicycles, etc. in the case of a vehicle-to-vehicle accident, utility poles, guardrails, etc. in the case of a single-vehicle accident, and pedestrians, etc. in the case of a person-to-vehicle accident. The vehicle damage degree estimation model may output not only the vehicle damage degree of the subject vehicle, but also the vehicle damage degree of other vehicles.

[0019] The occupant injury estimation model is a model that inputs the vehicle damage degree, the collision angle, personal information of the occupant including medical history such as underlying diseases, safety devices installed in the vehicle, etc., and outputs the extent of the injury to the occupant in the event of an accident. The vehicle damage degree used as input is the output from the vehicle damage estimation model. The occupant injury estimation model may output the extent of injury not only to the vehicle occupants (driver and passengers) but also to others (occupants of other vehicles, pedestrians, etc.).

[0020] The vehicle damage estimation model and the occupant injury estimation model are expected to be used as program modules that are part of artificial intelligence software. The vehicle damage estimation model and the occupant injury estimation model are used, for example, in a computer equipped with a CPU and memory, and the CPU of the computer operates according to instructions from the model stored in the memory. For example, the CPU of the computer operates according to the instructions to input information to the model, perform calculations according to the model, and output results from the model.

[0021] The vehicle damage degree estimation model generation unit 12 generates a vehicle damage degree estimation model by machine learning from the learning data acquired by the learning data acquisition unit 11. The learning data is data related to the risk of traffic accidents including road environment (road width, sidewalk, median, curve, traffic light, etc.), traffic volume, weather (season, time, weather, road surface condition, etc.), vehicles (pedestrians, two-wheeled vehicles, four-wheeled vehicles (normal passenger cars, light freight, large vehicles), etc.), driver information (age, height, weight, sex, underlying diseases, etc.), other party information, and vehicle damage degree, and for example, open data of traffic accident statistical information from the Metropolitan Police Department can be used.

[0022] The vehicle damage estimation model generation unit 12 uses the learning data, and constructs a vehicle damage estimation model using machine learning and statistical methods with road environment, traffic volume, weather, vehicle, driver information, and other party information as explanatory variables and the vehicle damage degree as a target variable. Here, the vehicle damage degree is classified and quantified into four levels, such as 0 for no damage, 1 for minor damage (e.g., damage to the extent that only slight scratches have been made on a part of the vehicle body), 2 for moderate damage (e.g., damage to the extent that extensive sheet metal repair is required), and 3 for major damage (e.g., damage to the extent that it is determined that the vehicle cannot be restored). Note that the classification of the vehicle damage degree is not limited to this.

[0023] To build a model, regression analysis, random forests (e.g., distributed random forests), support vector machines, deep learning using neural networks, etc. can be used.

[0024] The occupant injury estimation model generation unit 13 generates an occupant injury estimation model by machine learning from the learning data acquired by the learning data acquisition unit 11. The learning data is data including the collision angle, the degree of vehicle damage, occupant information (age, height, weight, sex, underlying diseases, seat position, etc.), safety devices installed in the vehicle (presence or absence of curtain airbags, etc.), and injuries sustained by occupants, and for example, US_CISS data (North American accident investigation database) can be used.

[0025] The occupant injury estimation model generation unit 13 uses the learning data, and estimates the severity of injury (MAIS (Maximum Abbreviated Injury Score), ISS (Injury Score)) by using the collision angle, the degree of vehicle damage, vehicle occupant information, and information on safety devices installed in the vehicle as explanatory variables. Using machine learning and statistical methods, we develop an injury estimation model for occupants using the vehicle injury severity score (HAS) or the AIS for each body part as the objective variable. Here, we quantify the injury severity of occupants in the same way as when we generated the vehicle damage estimation model.

[0026] To build a model, regression analysis (e.g., stepwise regression), random forests, support vector machines, deep learning using neural networks, etc. can be used.

[0027] The constructed vehicle damage estimation model and occupant injury estimation model are weakly coupled.

[0028] The situation selection unit 21 of the accident risk estimation system 20 accepts the selection of a road to be evaluated for accident risk. Here, the selection of a road may mean the selection of the entire road from the beginning (starting point) to the end (ending point) of the road (for example, the entire National Route 307), or the selection of a range on the road specified by the user, or the selection of a pinpoint intersection or the like. The situation selection unit 21 acquires road environment data of the selected road. The road environment data can be acquired from data published by the Ministry of Land, Infrastructure, Transport and Tourism, the Geospatial Information Authority of Japan, etc.

[0029] The condition setting unit 22 sets analysis conditions for accident risk evaluation. The analysis conditions are, for example, occupant information (age, height, weight, sex, underlying disease, seat position, etc.), weather, vehicle, moving direction, location to be evaluated, collision direction, other party information, etc. Evaluation target conditions such as the location and situation to be evaluated can be specified by the user. Vehicle information set as analysis conditions may be not only classified as two-wheeled, four-wheeled, etc., but also the vehicle model can be specified.

[0030] The vehicle damage degree estimation unit 23 uses the vehicle damage degree estimation model generated by the vehicle damage degree estimation model generation unit 12 to estimate the vehicle damage degree for each evaluation target condition from the road environment data of the evaluation target condition and the set analysis conditions.

[0031] The occupant injury estimation unit 24 uses the occupant injury estimation model generated by the occupant injury estimation model generation unit 13 to estimate the severity of injuries to occupants for each evaluation condition based on the driver information and collision direction within the analysis conditions and the vehicle damage degree estimated by the vehicle damage degree estimation unit 23 (output from the vehicle damage degree estimation model).

[0032] The accident risk assessment unit 25 estimates an overall assessment of the accident risk (accident risk assessment value) using the vehicle damage degree estimated by the vehicle damage degree estimation unit 23 and the extent of the injury to the occupant (occupant injury degree) estimated by the occupant injury estimation unit 24. For example, there are multiple collision modes in a traffic accident as shown in Fig. 5, and each collision mode, the vehicle itself, the other vehicle, and occupant information (age, gender, etc.) are weighted to calculate an overall assessment of the accident risk (total score).

[0033] The accident risk map creation unit 26 synthesizes various calculated indices such as the degree of vehicle damage, the degree of injury to the occupants, and the accident risk evaluation value into a map to create an accident risk map. For example, it marks evaluation points on the map with markers of colors according to the size of the external damage. The accident risk map is displayed on the output unit 3.

[0034] FIG. 3 shows an example of an accident risk map created by the accident risk map creation unit 26 when the entire section of National Route 307 in Shiga Prefecture is selected as the road and range to be evaluated for accident risk and the following analysis conditions are set: Driver: 40-year-old male / normal build / no underlying medical conditions Weather: September / morning / sunny Vehicle: Passenger car / 4-wheel mutual or single Evaluation points: 100m intervals Travel direction: North to South

[0035] Figure 4 shows an example of an accident risk map when the same analysis conditions as in Figure 3 were set, except that the driver's age was set to 75 years old.

[0036] According to this embodiment, it is possible to evaluate the accident risk (degree of vehicle damage and extent of external damage) at any point, including not only points where accidents have occurred in the past, but also points where no accidents have occurred.

[0037] The traffic accident risk assessment system according to this embodiment can be linked to, for example, a car navigation system (hereinafter, referred to as "car navigation"). The driver inputs his / her personal information (height, weight, sex, age, underlying diseases, etc.) into the car navigation system in advance.

[0038] When a driver sets a destination in the car navigation system, the car navigation system searches for and sets a route to the destination. The traffic accident risk assessment system obtains current weather information from the Japan Meteorological Agency database and uses pre-entered driver information to estimate the extent of injuries to occupants at multiple points (e.g., multiple points at 50m intervals) on the set route. The car navigation system synthesizes and displays the estimated extent of injuries on a map. The car navigation system may output an alert when approaching a point where the estimated injuries are severe.

[0039] The traffic accident risk assessment system according to this embodiment may be output as a hazard map by time, condition, or individual by displaying it on paper or a display.

[0040] One of the factors that has a large impact on the accident risk estimated by the traffic accident risk assessment system is whether or not the driver has underlying diseases. Therefore, accident risk maps can be created based on the presence or absence of underlying diseases and by type of underlying disease, which can be used to plan accident prevention measures.

[0041] The accident risk assessment part value may be used for setting goals as an assessment criterion for traffic accidents.

[0042] The present invention is not limited to the above-described embodiment, and various modifications are possible without departing from the spirit of the present invention. [Explanation of symbols]

[0043] 1 Computing device 2 Input section 3. Output section 10 Model Generation System 11 Learning data acquisition section 12 Vehicle damage estimation model generation unit 13. Occupant injury estimation model generation unit 20 Accident Risk Estimation System 21 Situation Selection Section 22 Condition setting section 23 Vehicle Damage Estimation Department 24. Crew Injury Estimation Section 25 Accident Risk Assessment Department 26 Accident Risk Map Creation Department

Claims

1. a situation selection unit that accepts a selection of a road to be evaluated; a condition setting unit for setting analysis conditions including occupant information, weather, the vehicle itself, and other party information; a vehicle damage degree estimation unit that estimates a vehicle damage degree at the time of an accident occurring at at least one point or situation on the road based on a road environment of the selected road and the analysis conditions; an occupant injury estimation unit that estimates the extent of injury to a vehicle occupant at the time of an accident occurring at the location or situation based on the vehicle damage degree and the occupant information; A traffic accident risk assessment system.

2. 2. The traffic accident risk assessment system according to claim 1, further comprising an accident risk map creation unit that creates an accident risk map by synthesizing the magnitude of injury to the vehicle occupant estimated for each of the plurality of points into a map.

3. The traffic accident risk assessment system of claim 1, wherein the vehicle damage degree estimation unit estimates the vehicle damage degree using a vehicle damage degree estimation model constructed by machine learning with a combination of road environment, occupant information, weather and vehicle as explanatory variables and the vehicle damage degree as a target variable.

4. The traffic accident risk assessment system of claim 1, wherein the occupant injury estimation unit estimates the extent of trauma to the vehicle occupant using an occupant injury estimation model constructed by machine learning with a combination of occupant information and vehicle damage degree as explanatory variables and the extent of trauma to the occupant as a target variable.

5. The traffic accident risk assessment system of claim 1, further comprising an accident risk assessment unit that estimates an overall accident risk assessment based on the degree of vehicle damage estimated by the vehicle damage degree estimation unit and the degree of injury to the vehicle occupant estimated by the occupant injury estimation unit.

Images