WO2023177277A1

WO2023177277A1 - Artificial intelligence-based traffic safety management method

Info

Publication number: WO2023177277A1
Application number: PCT/KR2023/003679
Authority: WO
Inventors: 박선영; 박경범; 황경승; 김민석; 이재원; 윤진수; 여화수
Original assignee: 한국교통안전공단
Priority date: 2022-03-18
Filing date: 2023-03-20
Publication date: 2023-09-21
Also published as: KR102448174B1

Abstract

The present invention relates to an artificial intelligence-based traffic safety management method, and comprises the steps of: partitioning a region to be managed into multiple unit areas, and processing pre-collected region management information for each of the multiple unit areas to extract at least one feature information; predicting the possibility of traffic accidents happening of each of the multiple unit areas by using the feature information, and grading same; analyzing the correlation between the feature information and the possibility of traffic accidents happening for each of the multiple unit areas, to detect a risk factor; and analyzing the risk factor to recommend a countermeasure.

Description

Artificial intelligence-based traffic safety management method

The present invention relates to an artificial intelligence-based traffic safety management method, and more specifically, to a traffic safety management method that can predict traffic accident risk areas, analyze causes, and suggest countermeasures based on artificial intelligence.

Cars have many positive aspects as a means of transportation, but they also have negative aspects that can cause traffic accidents and cause damage to life and property. In modern society, traffic accidents have emerged as an important social problem.

Traffic accidents can occur due to complex factors between people, roads, and vehicles. To prevent traffic accidents, various methods are used, such as installing sensors on the road or installing cameras or CCTV at specific points on the road, but traffic accidents still occur frequently.

Accordingly, technologies to prevent and predict traffic accidents are being developed, but are only applied as simple statistical models to areas with frequent traffic accidents and road facility risk areas. This has limitations in preventing traffic accidents because it only serves to alert drivers to dangerous spots and encourage them to drive carefully, and does not improve the fundamental problems on the roads where traffic accidents occur.

In addition, technologies using artificial intelligence are being developed, but most of them are limited to traffic flow or self-driving cars. Therefore, technology is needed to prevent and manage traffic accidents in advance.

An embodiment of the present invention seeks to provide a traffic safety management method that can predict traffic accident risk areas, analyze causes, and suggest countermeasures based on artificial intelligence.

Among embodiments, an artificial intelligence-based traffic safety management method includes dividing a management target area into a plurality of unit areas, processing area management information collected in advance for each of the plurality of unit areas, and extracting at least one feature information; predicting and grading the likelihood of a traffic accident occurring in each of the plurality of unit areas using the characteristic information; detecting risk factors by analyzing a correlation between characteristic information of each of the plurality of unit areas and the possibility of a traffic accident occurring; and analyzing the risk factors and recommending countermeasures.

The disclosed technology can have the following effects. However, since it does not mean that a specific embodiment must include all of the following effects or only the following effects, the scope of rights of the disclosed technology should not be understood as being limited thereby.

The traffic safety management method according to an embodiment of the present invention can enable traffic safety management centered on prevention by predicting traffic accident risk areas, analyzing causes, and suggesting countermeasures based on artificial intelligence.

1 is a diagram illustrating a traffic safety management device according to an embodiment of the present invention.

FIG. 2 is a flowchart illustrating a traffic safety management method performed by the traffic safety management device shown in FIG. 1.

FIG. 3 is an example diagram illustrating a unit area partitioned by the data preprocessor shown in FIG. 1.

FIG. 4 is an example diagram illustrating a screen displaying grades classified by the accident grade prediction unit shown in FIG. 1 .

FIG. 5 is an example diagram illustrating a report prepared by the alternative recommendation unit shown in FIG. 1.

Since the description of the present invention is only an example for structural or functional explanation, the scope of the present invention should not be construed as limited by the examples described in the text. In other words, since the embodiments can be modified in various ways and can have various forms, the scope of rights of the present invention should be understood to include equivalents that can realize the technical idea. In addition, the purpose or effect presented in the present invention does not mean that a specific embodiment must include all or only such effects, so the scope of the present invention should not be understood as limited thereby.

Meanwhile, the meaning of the terms described in this application should be understood as follows.

Terms such as “first” and “second” are used to distinguish one component from another component, and the scope of rights should not be limited by these terms. For example, a first component may be named a second component, and similarly, the second component may also be named a first component.

When a component is referred to as being “connected” to another component, it should be understood that it may be directly connected to the other component, but that other components may exist in between. On the other hand, when a component is referred to as being “directly connected” to another component, it should be understood that there are no other components in between. Meanwhile, other expressions that describe the relationship between components, such as "between" and "immediately between" or "neighboring" and "directly neighboring" should be interpreted similarly.

Singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as “comprise” or “have” refer to implemented features, numbers, steps, operations, components, parts, or them. It is intended to specify the existence of a combination, and should be understood as not excluding in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts, or combinations thereof.

For each step, identification codes (e.g., a, b, c, etc.) are used for convenience of explanation. The identification codes do not explain the order of each step, and each step clearly follows a specific order in context. Unless specified, events may occur differently from the specified order. That is, each step may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the opposite order.

The present invention can be implemented as computer-readable code on a computer-readable recording medium, and the computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system. . Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices. Additionally, the computer-readable recording medium can be distributed across computer systems connected to a network, so that computer-readable code can be stored and executed in a distributed manner.

All terms used herein, unless otherwise defined, have the same meaning as commonly understood by a person of ordinary skill in the field to which the present invention pertains. Terms defined in commonly used dictionaries should be interpreted as consistent with the meaning they have in the context of the related technology, and cannot be interpreted as having an ideal or excessively formal meaning unless clearly defined in the present application.

1 is a diagram illustrating an artificial intelligence-based traffic safety management system according to an embodiment of the present invention.

Referring to FIG. 1, the artificial intelligence-based traffic safety management system 100 according to an embodiment of the present invention includes a data preprocessing unit 110, an accident risk area prediction unit 120, an accident grade prediction unit 130, and a factor It may include an analysis unit 140, an alternative recommendation unit 150, and a control unit 160. The data preprocessing unit 110 receives and preprocesses regional management information for managing traffic safety in the management target area. Here, local management information is information related to accident factors that may cause traffic accidents, and may be collected in advance and stored in a separate database (not shown).

Local management information may include human information, vehicle information, traffic accident information, environmental information, social structure information, and transportation system information. Personal information may include information related to people, such as driving behavior characteristics such as drowsiness, inattention, and dangerous driving behavior, and pedestrian characteristics such as the number and distribution of pedestrians. Here, risky driving behavior refers to at least one of 11 behaviors, such as rapid acceleration, rapid deceleration, sudden start, sudden stop, sudden right turn, sudden left turn, sudden U-turn, sudden overtaking, sudden lane change, speeding, and prolonged speeding. It can include either one. Speeding can be the act of driving at a speed that exceeds 20km/h over the speed limit, and long-term speeding can be the act of maintaining speeding for more than 3 minutes.

Vehicle information may include information related to the vehicle, such as vehicle type, traffic volume, average speed, advanced device installation, and illegal tuning. Traffic accident information may include information related to traffic accidents, such as the number and severity of preceding traffic accidents. Here, the severity can be classified as death, serious injury, minor injury, etc.

Environmental information may include information related to the environment, such as the type and distribution of transportation facilities and weather factors. Here, traffic facilities may include signal facilities, road facilities, sign facilities, etc. For example, signaling facilities may include pedestrian activated signals, mounts, traffic lights, acoustic signals, remaining time indicators, controllers, and poles. Road facilities may include road obstacles, text signs, speed signs, safety zones, no stopping zones, crosswalks and other road markings, directional signs, lanes, etc. Sign facilities can include 48 types of facilities, including safety signs, child protection area signs, and elderly protection area signs. Additionally, traffic facilities may include other facilities such as CCTV, detectors, intersections, and gaze guides.

Social structure information may include information related to social structure, such as population structure, population density, lifestyle, road area, and commercial district area. Traffic system information may include information related to the traffic system, such as traffic law violations or crackdowns.

The data pre-processing unit 110 divides the management target area into a plurality of unit areas, processes area management information for each of the plurality of unit areas, and extracts at least one feature information for each of the plurality of unit areas. The data pre-processing unit 110 may classify regional management information into a plurality of unit areas and extract characteristics of the regional management information for each of the plurality of unit areas as feature information.

The accident risk area prediction unit 120 predicts whether or not an accident will occur in each of the plurality of unit areas. The accident risk area prediction unit 120 can integrate characteristic information of each of the plurality of unit areas and predict whether or not an accident will occur according to the integration result. Here, the accident risk area prediction unit 120 may predict whether or not an accident will occur using an artificial intelligence model, for example, a convolutional neural network.

The accident risk area prediction unit 120 can predict whether or not a traffic accident will occur by grouping a plurality of unit areas and comparing the integration results of characteristic information for each group. In general, because urban areas have more accident factors than suburban areas, if no distinction is made between the urban area and suburban areas, all areas located in the urban area can be predicted to be areas where traffic accidents occur.

Accordingly, the accident risk area prediction unit 120 according to an embodiment of the present invention classifies the plurality of unit areas into areas located in the downtown area and areas located in outlying areas other than the city center, and divides the areas located in the downtown area into the downtown area group. After grouping the areas located in the outskirts into outlying area groups, prediction accuracy can be improved by predicting the presence or absence of traffic accidents in each of the plurality of unit areas in the city center group and outskirt group units. One embodiment of the present invention is not limited to this, and the accident risk area prediction unit 120 may select an arbitrary number of unit areas in order and group them into one group.

The accident risk area prediction unit 120 may perform data augmentation on the number of traffic accidents among the extracted feature information and then predict whether or not a traffic accident will occur. In general, since traffic accidents are relatively small compared to non-traffic accidents, traffic accident data is absorbed into non-traffic accident data, so that a plurality of unit areas can be predicted to have no traffic accidents. Accordingly, the accident risk area prediction unit 120 according to an embodiment of the present invention can generate data similar to actual traffic accident data, amplify the number of traffic accidents, and then predict whether or not a traffic accident will occur.

The accident grade prediction unit 130 predicts and rates the likelihood of a traffic accident occurring in each of the plurality of unit areas using feature information extracted from each of the plurality of unit areas. The accident grade prediction unit 130 may determine the probability of a traffic accident occurring based on characteristic information of each of the plurality of unit areas and grade the probability of a traffic accident occurring according to the degree of probability of a traffic accident occurring.

Here, the accident grade prediction unit 130 sets the grade by classifying the probability of traffic accident occurrence into a certain range, and grades the probability of traffic accident occurrence into a grade corresponding to the probability of traffic accident occurrence in each of the plurality of unit areas. there is. The possibility of a traffic accident occurring according to an embodiment of the present invention can be classified into four levels. For example, Grade 1 has a traffic accident probability of 5% to less than 25%, Grade 2 has a traffic accident probability of 25% to less than 50%, and Grade 3 has a traffic accident probability of 50% to 75%. It is less than %, and level 4 corresponds to a traffic accident probability of more than 75%. In other words, the higher the probability of a traffic accident occurring, the higher the level can be classified.

The accident grade prediction unit 130 may use an artificial intelligence model to determine the probability of a traffic accident occurring and grade the probability of a traffic accident occurring. For example, the accident grade prediction unit 130 may use various non-linear relationship learning methods including multiple hidden layers such as deep neural networks. In addition, an embodiment of the present invention is not limited to this, and the accident grade prediction unit 130 may predict the possibility of a traffic accident occurring by targeting only a unit area predicted to cause a traffic accident among a plurality of unit areas.

The factor analysis unit 140 may detect risk factors by analyzing the correlation between characteristic information of each of the plurality of unit areas and the possibility of a traffic accident occurring. Here, the factor analysis unit 140 enables the function of suggesting the cause of traffic accident risk and alternatives, and can analyze the correlation between the characteristic information of each of the plurality of unit areas and the possibility of traffic accident occurrence based on a decision tree. However, an embodiment of the present invention is not limited thereto.

The factor analysis unit 140 may detect an area where risk factors are concentrated within a unit area where the possibility of a traffic accident occurring is higher than a preset risk level among a plurality of unit areas and select the area as a risk factor concentration area (hot-spot). Here, the risk level may be level 3. That is, the factor analysis unit 140 may select a risk factor concentration area targeting unit areas where the probability of traffic accidents occurring is level 3 or 4 among a plurality of unit areas.

The factor analysis unit 140 may select the area where traffic accidents occur and where traffic facilities are concentrated within a specific unit area as the risk factor concentration area. For example, the factor analysis unit 140 may select an area where a traffic accident occurred as a risk factor concentration area even though it is an area where major traffic facilities such as traffic lights, crosswalks, child protection zones, safety signs, and elderly protection zones are concentrated. there is.

The alternative recommendation unit 150 analyzes risk factors detected in areas where risk factors are concentrated and recommends countermeasures to prevent the risk factors. The alternative recommendation unit 150 can classify risk factors according to pre-selected accident factor types and analyze the characteristics of risk factors in risk factor concentration areas in connection with traffic accident use cases.

Here, accident factor types can be classified into human factors, vehicle factors, traffic accident factors, environmental factors, social structure factors, and institutional factors. Human factors may be accident factors related to human information, vehicle factors may be factors related to vehicle information, and environmental factors may be factors related to environmental information. Additionally, social structure factors may be factors related to social structure information, and institutional factors may be factors related to transportation system information.

The alternative recommendation unit 150 may recommend a response plan according to the characteristics of the risk factors for the area where the risk factors are concentrated, and may write and issue the recommended response plan as a report. The alternative recommendation unit 150 can write the risk factors in the area where risk factors are concentrated in a chart format and record the analysis results of the risk factors and countermeasures in graphs or text. Additionally, countermeasures can be created by superimposing them on a map image of the area where risk factors are concentrated.

The control unit 160 controls the overall operation of the traffic safety management system 100 and includes a data preprocessing unit 110, an accident risk area prediction unit 120, an accident grade prediction unit 130, a factor analysis unit 140, and Control flow or data flow between the alternative recommendation units 150 can be managed.

FIG. 2 is a flowchart illustrating a traffic safety management method performed in the traffic safety management system shown in FIG. 1, and FIG. 3 is a flowchart illustrating a unit area partitioned by the data preprocessor shown in FIG. 1. This is just one example. Figure 4 is an example diagram shown to explain a screen displaying grades classified by the accident grade prediction unit shown in Figure 1, and Figure 5 is shown to explain a report prepared by the alternative recommendation unit shown in Figure 1. This is an example diagram.

Referring to FIG. 2, the data pre-processing unit 110 first divides the management target area into a plurality of unit areas. At this time, the data pre-processing unit 110 may partition the management target area in a grid format. For example, as shown in FIG. 3, the data preprocessor 110 may divide the management area (TA) into unit areas (UA) of a certain size.

The data pre-processing unit 110 processes regional management information for each unit area and extracts characteristic information for each of a plurality of unit areas. Next, the accident risk area prediction unit 120 predicts whether or not a traffic accident will occur in each of the plurality of unit areas using the extracted feature information.

At this time, the accident risk area prediction unit 120 can predict whether a traffic accident will occur by integrating the characteristic information of each of the plurality of unit areas. For example, if dangerous driving behavior exists within a specific unit area among a plurality of unit areas, the number of pedestrians is large, and the average speed of vehicles is high, the unit area can be predicted as an area where a traffic accident will occur.

Next, the accident grade prediction unit 130 predicts and rates the likelihood of a traffic accident occurring in each of the plurality of unit areas using feature information extracted from each of the plurality of unit areas (S110).

For example, the accident grade prediction unit 130 classifies unit areas with a traffic accident probability of less than 25% among a plurality of unit areas as grade 1, and unit areas with a traffic accident probability of more than 25% and less than 50% as grade 1. It can be classified into 2 grades.

At this time, as shown in FIG. 4, the accident grade prediction unit 130 may display the predicted traffic accident probability grades in different colors and provide them to the user. For example, the accident grade prediction unit 130 may change the color from yellow to red as the grade increases.

Meanwhile, the accident grade prediction unit 130 can predict and rate the possibility of a traffic accident occurring for a unit area predicted as an area where a traffic accident will occur through the accident risk area prediction unit 120 among a plurality of unit areas. .

Next, the factor analysis unit 140 detects risk factors by analyzing the correlation between the characteristic information of each of the plurality of unit areas and the possibility of a traffic accident occurring (S120). For example, the factor analysis unit 140 predicts that the probability of a traffic accident occurring in a specific unit area is level 4, that there are 5 or less direction sign facilities in the unit area, that there are 1 or less safety sign facilities, and that the temporary stop line If there is no left turn sign, these traffic facilities can be analyzed as a factor correlated with the possibility of a traffic accident occurring, and information about the traffic facility can be detected as a risk factor.

In addition, the factor analysis unit 140 selects a risk factor concentration area (Hot-spot) where risk factors are concentrated within a unit area where the probability of traffic accidents occurring is at a risk level or higher among a plurality of unit areas. Then, the alternative recommendation unit 150 analyzes the characteristics of the risk factors detected in the risk factor concentration area and recommends a countermeasure to prevent the risk factors according to the analyzed characteristics (S130).

For example, the alternative recommendation unit 150 determines that the risk factors detected within a specific unit area are human factors and environmental factors, that there is a risk of speeding and rapid acceleration/deceleration within the unit area, and that there is a risk of speeding and rapid acceleration/deceleration within the unit area, and If the analysis results show that there is a high risk of conflict, installation of speeding/red light violation cameras, lowering the speed limit, and installing a median barrier to prevent illegal U-turns can be recommended as countermeasures.

The alternative recommendation unit 150 can write a report on recommended response measures in areas where risk factors are concentrated. For example, as shown in FIG. 5, the alternative recommendation unit 150 can display risk factors in areas where risk factors are concentrated in a chart format and record the analysis results of risk factors and countermeasures in graphs or text. . Additionally, a report can be created by superimposing countermeasures on a map image of the risk factor concentration area.

As described above, the artificial intelligence-based traffic safety management method according to an embodiment of the present invention predicts the traffic accident risk area in the management target area, detects risk factors for the predicted traffic accident risk area, and determines the risk factors. A response plan can be recommended. Therefore, traffic safety management can be proactively carried out with a focus on prevention.

Claims

Dividing a management target area into a plurality of unit areas, processing area management information collected in advance for each of the plurality of unit areas, and extracting at least one feature information;

predicting and grading the likelihood of a traffic accident occurring in each of the plurality of unit areas using the characteristic information;

detecting risk factors by analyzing a correlation between characteristic information of each of the plurality of unit areas and the possibility of a traffic accident occurring; and

An artificial intelligence-based traffic safety management method that includes analyzing the risk factors and recommending countermeasures.
The method of claim 1, wherein the regional management information is

An artificial intelligence-based traffic safety management method that includes at least one of human information, vehicle information, traffic accident information, environmental information, social structure information, and transportation system information.
The method of claim 1, wherein the step of predicting the possibility of an accident occurring is

An artificial intelligence-based traffic safety management method comprising determining the probability of a traffic accident occurring using the characteristic information of each of the plurality of unit areas.
The method of claim 3, wherein the step of grading the likelihood of an accident occurring is

An artificial intelligence-based traffic safety management method comprising the step of classifying the probability of occurrence of the accident according to the probability of occurrence of the traffic accident.
According to paragraph 1,

An artificial intelligence-based traffic safety management method further comprising the step of detecting an area where the risk factors are concentrated among the plurality of unit areas and selecting a unit area where the probability of an accident occurring is higher than a preset risk level and selecting it as a risk factor concentration area.
The method of claim 1, wherein the step of recommending the countermeasure is

An artificial intelligence-based traffic safety management method including the step of analyzing the characteristics of the risk factors based on pre-selected accident factor types and traffic accident use cases.
According to paragraph 1,

An artificial intelligence-based traffic safety management method further comprising the step of writing and issuing the above countermeasures in a report.
According to paragraph 1,

An artificial intelligence-based traffic safety management method further comprising predicting whether or not a traffic accident will occur in each of the plurality of unit areas using the characteristic information.