JP7792770B2 - Traffic congestion determination system, traffic congestion determination method, and traffic congestion determination program - Google Patents

Description

The present invention relates to a congestion determination system, a congestion determination method, and a congestion determination program for determining congestion conditions.

When traveling on roads in vehicles such as automobiles and motorcycles, users have traditionally used driving assistance systems installed in in-vehicle navigation devices and various information terminals such as smartphones to select the optimal route to their destination (e.g., a route with less congestion). When a user selects a route with less congestion, the driving assistance system acquires traffic information such as congestion from VICS (registered trademark: Vehicle Information and Communication System), determines the congestion status of the road based on the acquired traffic information, searches for a route with less congestion, and provides the searched route to the user. Here, VICS acquires traffic information for vehicles from congestion sensors installed on expressways and major public roads, analyzes traffic conditions such as congestion based on the acquired traffic information, and provides the analyzed traffic information to the driving assistance system.

In recent years, there has been an increasing trend toward road traffic information systems that utilize probe information. These systems not only collect traffic information from congestion sensors installed on expressways and major public roads, but also collect location information, vehicle speed information, and other information from moving objects on the road (probe information) and analyze traffic conditions based on large amounts of probe information. Road traffic information systems that utilize probe information can also grasp traffic conditions on roads where VICS is not installed, raising hopes that they will enable the analysis of congestion conditions on a wider range of roads.

Amid this, as traffic congestion analysis methods become more sophisticated, it is becoming possible to analyze congestion not only on an individual road basis, but also on an individual lane basis. By understanding the congestion situation on an individual lane basis, driving assistance systems can search for routes that avoid unnecessary congestion and detours, thereby providing users with more optimal routes.

An example of how understanding the congestion situation for each lane can be effective in optimal route search is an analysis of a road with two lanes on each side that has a fork in the road (Road A1 and Road A2) in the direction of travel, where only the lanes on Road A1 are congested just before the fork. If the user heads toward the other, less congested road, Road A2, a driving assistance system that understands the congestion situation for each lane can quickly guide the user to the lane on the fork A2 side. As a result, the user can pass through the fork using the optimal lane route.

When analyzing congestion conditions for each lane as described above, attention is being paid to analyzing congestion conditions for each mobile object. Analysis of congestion conditions for each mobile object is highly compatible with road traffic information systems that utilize probe information. In fact, there is technology (for example, Patent Document 1) that uses road traffic information systems that utilize probe information to analyze congestion conditions for each mobile object based on speed information of mobile objects traveling on the road, and grasp congestion conditions for each lane.

In addition, methods for analyzing congestion conditions for each lane without analyzing congestion conditions for each moving object include an analysis method based on information acquired from VICS (e.g., Patent Document 2) and an analysis method based on in-vehicle camera footage (e.g., Patent Document 3).

JP 2019-109708 A Re-table No. 2018-151005 JP 2020-4235 A

However, Patent Document 1 uses vehicle speed information for the analysis to determine that a moving object traveling at a speed below a certain level is a moving object in a traffic jam. As a result, there is a risk that a slow-moving object that is not necessarily the cause of a traffic jam, such as a taxi waiting for a customer, may be mistakenly determined to be a moving object in a traffic jam. Therefore, Patent Document 1 cannot accurately provide users with the traffic congestion status of slow-moving objects that are not the cause of a traffic jam.

Furthermore, Patent Document 2 analyzes the congestion situation for each lane using traffic condition information collected by VICS. As a result, Patent Document 2 cannot analyze the congestion situation for each lane on roads where VICS is not installed. Therefore, in the current situation where roads where VICS is installed are limited to expressways and major trunk roads, Patent Document 2 cannot provide users with information on the congestion situation for each lane on the many roads where VICS is not installed.

Furthermore, Patent Document 3 uses in-vehicle camera footage captured by an in-vehicle camera to analyze the movement status of surrounding moving objects captured in the footage and analyze the congestion status for each lane. Therefore, Patent Document 3 requires that a certain number of moving objects equipped with in-vehicle cameras be present on the road being analyzed. Therefore, Patent Document 3 cannot provide users with accurate information about the congestion status on roads where the penetration rate of in-vehicle cameras is low.

The present invention therefore aims to provide a congestion determination system, congestion determination method, and congestion determination program that can accurately determine the congestion situation for each mobile object on a greater number of roads.

A traffic congestion determination system according to one aspect of the present invention includes a determiner that determines the traffic congestion status of each of a plurality of mobile bodies, the determiner being constructed by machine learning based on learning information including learning travel information indicating the travel information of each of a plurality of mobile bodies and learning traffic congestion status information indicating the traffic congestion status of the corresponding mobile bodies; an input travel information acquisition unit that acquires input travel information indicating the travel information of each of at least one mobile body; and a mobile body determination unit that inputs the input travel information to the determiner and determines the traffic congestion status of each of the at least one mobile body.

In addition, a congestion determination method according to one aspect of the present invention involves a computer equipped with a determiner for determining the congestion status of each of a plurality of mobile bodies, the determiner being constructed by machine learning based on learning information including learning movement information indicating the movement information of each of a plurality of mobile bodies and learning congestion status information indicating the congestion status of the corresponding mobile body, acquiring input movement information indicating the movement information of at least one of the mobile bodies, inputting the input movement information to the determiner, and determining the congestion status of each of the at least one of the mobile bodies.

In addition, a congestion determination program according to one embodiment of the present invention causes a computer to implement a determiner that determines the congestion status of each mobile object, the determiner being constructed by machine learning based on learning information including learning movement information indicating the movement information of each of a plurality of mobile objects and learning congestion status information indicating the congestion status of the corresponding mobile object; an input movement information acquisition unit that acquires input movement information indicating the movement information of at least one mobile object; and a mobile object determination unit that inputs the input movement information to the determiner and determines the congestion status of each of the at least one mobile object.

In this invention, the term "unit" does not simply mean a physical means, but also includes cases where the functions of that "unit" are realized by software. Furthermore, the functions of a single "unit" or device may be realized by two or more physical means or devices, and the functions of two or more "units" or devices may be realized by a single physical means or device.

The present invention makes it possible to provide a congestion determination system, congestion determination method, and congestion determination program that can accurately determine the congestion status of each mobile object on more roads.

1 is a diagram showing the configuration of a congestion determination system 100 according to an embodiment of the present invention. FIG. 2 is a diagram showing an example of information stored in a learning information storage unit 131. 10 is a diagram showing an example of information stored in an input movement information storage unit 141. FIG. 10 is a diagram showing an example of determination by a moving body determination unit 142 using a determiner 133. FIG. 10 is a diagram showing an example of information stored in a moving body determination information storage unit 144. FIG. FIG. 10 is a diagram showing an example of information stored in a tally information storage unit 151. 10 is a diagram showing an example of information stored in a link determination information storage unit 154. FIG. 1 is a flowchart illustrating an example of machine learning processing in the traffic congestion determination system 100. 10 is a flowchart showing an example of a process for determining congestion in the congestion determination system 100.

A preferred embodiment of the present invention will be described with reference to the accompanying drawings. Figure 1 is a diagram showing the configuration of a congestion determination system 100, which is one embodiment of the present invention.

The traffic congestion determination system 100 is a system that is communicatively connected to the information processing system 110 via a network such as the Internet. The information processing system 110 is further communicatively connected to a terminal device 120 via a network such as the Internet. The traffic congestion determination system 100 may also be communicatively connected to the terminal device 120 via a network such as the Internet. Furthermore, while Figure 1 shows three examples of terminal devices 120, namely, terminal devices 120a, 120b, and 120c, the number of terminal devices 120 is not limited to this.

The congestion determination system 100 acquires learning information from the information processing system 110, including learning movement information indicating the movement information of each of a plurality of moving bodies and learning congestion situation information indicating the congestion situation of the corresponding moving bodies, and generates a determiner that determines the congestion situation of each moving body through machine learning based on the learning information. The congestion determination system 100 then inputs input movement information indicating the movement information of at least one moving body acquired from the information processing system 110 or the terminal device 120 into the determiner, and determines the congestion situation of the moving body.

Furthermore, the congestion determination system 100 determines the congestion status of at least one link based on the congestion determination result of at least one mobile object. Then, the congestion determination system 100 provides the congestion determination result of at least one mobile object and the congestion determination result of at least one link to the information processing system 110. Here, a mobile object is an object, such as an automobile or motorcycle, traveling on a road used for the movement of people or goods. Details of the congestion determination system 100 will be described later.

The information processing system 110 is a system that collects and analyzes traffic condition information, such as movement information for each mobile object (time information, speed information, acceleration information, location information, mileage information, etc.), congestion information indicating road congestion conditions, and information on road accidents and lane restrictions. The information processing system 110 works in conjunction with a system that manages traffic condition information and mobile objects, etc. to collect traffic condition information. The information processing system 110 may be, for example, a system that analyzes traffic condition information based on the movement information and congestion information of mobile objects and provides the analyzed traffic information to a terminal device or base station, or a driving assistance system that supports the driving of mobile objects based on the analyzed traffic information. The information processing system 110 may also be managed by a public institution or independently by a mobile object manufacturer.

The terminal device 120 is an information communication terminal mounted on a mobile object or an information communication terminal carried by a user who travels using the mobile object. The terminal device 120 collects movement information (time information, speed information, acceleration information, location information, mileage information, etc.) of each mobile object from various parts of the mobile object and provides the collected movement information to the information processing system 110. The terminal device 120 may also calculate movement information of each mobile object based on GPS coordinate information measured by a GPS system using a GPS device mounted on the mobile object or the terminal device 120, and provide the calculated movement information to the information processing system 110. The terminal device 120 may also provide the movement information to the congestion determination system 100.

Next, the details of the traffic congestion determination system 100 will be described. As shown in FIG. 1, the traffic congestion determination system 100 includes a learning information acquisition unit 130, a learning information storage unit 131, a machine learning execution unit 132, a determiner 133, an input movement information acquisition unit 140, an input movement information storage unit 141, a moving object determination unit 142, a moving object determination information generation unit 143, a moving object determination information storage unit 144, an aggregate information generation unit 150, an aggregate information storage unit 151, a link determination unit 152, a link determination information generation unit 153, a link determination information storage unit 154, and a determination information provision unit 160. The computer that constitutes the traffic congestion determination system 100 includes a processor and a memory area. Each unit shown in FIG. 1 can be realized, for example, by using the memory area or by the processor executing a program stored in the memory area.

The learning information acquisition unit 130 accesses the information processing system 110, acquires from the information processing system 110 learning information that serves as the basis for training data when performing machine learning, and stores the acquired learning information in the learning information storage unit 131. The learning information includes, for example, learning movement information indicating the movement information of each of multiple moving objects, and learning congestion status information indicating the congestion status of the corresponding moving objects, which have been accumulated in the information processing system 110 over a certain period of time in the past.

The learning movement information is movement information for each of multiple moving objects, and includes, for example, time information indicating the time of data acquisition, speed information indicating the speed of the moving object, and GPS coordinate information indicating GPS coordinates measured by a GPS system using a GPS device installed in the moving object or terminal device 120.

The learning traffic congestion information indicates the traffic congestion situation of the corresponding mobile object, and is indicated by indicators such as "traffic jam," "slightly congested," or "empty." The traffic congestion information may be a qualitative indicator such as "A," "B," or "C," or a quantitative indicator such as a percentage. The learning traffic congestion information may be collected by having the mobile object used for information collection travel along the road and having the user of the mobile object (driver or passenger) make a judgment, or it may be calculated based on the traffic congestion judgment results determined by a system different from the traffic congestion judgment system 100 (for example, a system that analyzes traffic congestion situations based on road images).

The timing at which the learning information acquisition unit 130 acquires learning information from the information processing system 110 can be set arbitrarily. For example, the learning information acquisition unit 130 may acquire the learning information from the information processing system 110 before machine learning by the machine learning execution unit 132 (described below), or may acquire the learning information from the information processing system 110 periodically, such as once a day. Furthermore, the learning information acquisition unit 130 may acquire the learning information from the information processing system 110 in response to an instruction from the information processing system 110.

Figure 2 is a diagram showing an example of information stored in the learning information storage unit 131. The information stored in the learning information storage unit 131 includes, for example, a mobile object ID, learning movement information, and learning traffic congestion information. The mobile object ID is mobile object identification information that identifies the mobile object that provides information to the information processing system 110.

Note that the learning information stored in the learning information storage unit 131 does not have to be the learning information itself obtained from the information processing system 110. Furthermore, the data included in the learning information may be partially missing, or the items may be different.

The machine learning execution unit 132 generates a determiner 133 that determines the congestion status of each mobile object by performing machine learning based on the learning information stored in the learning information storage unit 131. Machine learning by the machine learning execution unit 132 is performed, for example, in response to instructions from a system administrator. Note that machine learning may also be performed automatically at predetermined times, such as during system maintenance. There are no particular limitations on the machine learning algorithm, but examples that can be used include decision tree learning, deep learning, random forest, LightGBM (Light Gradient Boosting Machine), and SVM (Support Vector Machine).

In machine learning by the machine learning execution unit 132, information calculated based on speed information included in the input movement information stored in the input movement information storage unit 141 can be used as a feature. Information calculated based on speed information includes, for example, the maximum speed, minimum speed, average speed, acceleration, and the variance, standard deviation, skewness, and kurtosis of the frequency distribution within a certain period of time.

Machine learning by the machine learning execution unit 132 enables highly accurate congestion detection by using multiple pieces of information calculated based on speed information as feature quantities.

For example, by generating a determiner 133 based on machine learning that combines maximum speed information and acceleration information, a more accurate congestion determination system 100 can be constructed compared to simply determining that a mobile object whose maximum speed is below a certain value is in a "traffic jam." This is because the information characterizing the traffic jam situation of a mobile object is contained not only in the maximum speed information of the mobile object, but also in its acceleration information. A specific example is described below.

Both a moving object that is simply driving slowly regardless of the traffic jam and a moving object that is driving slowly while repeatedly starting and stopping due to the traffic jam are traveling at a low speed. Meanwhile, when it comes to acceleration, the acceleration of a moving object that is simply driving slowly regardless of the traffic jam is small and constant, whereas the acceleration of a moving object that is driving slowly while repeatedly starting and stopping due to the traffic jam fluctuates. Therefore, if a moving object whose maximum speed is below a certain value is simply determined to be in a "traffic jam," it is not possible to distinguish between a moving object that is simply driving slowly regardless of the traffic jam and a moving object that is driving slowly while repeatedly starting and stopping due to the traffic jam. However, by utilizing machine learning that combines maximum speed information and acceleration information, it is possible to distinguish between a moving object that is simply driving slowly regardless of the traffic jam and a moving object that is driving slowly while repeatedly starting and stopping due to the traffic jam.

In this way, machine learning that combines maximum speed information and acceleration information can generate a classifier 133 that, for example, makes it easier to determine that a moving object with a low maximum speed and high acceleration is in a "traffic jam," enabling more accurate traffic jam determination compared to simply determining that a moving object with a maximum speed below a certain value is in a "traffic jam."

In addition, in machine learning by the machine learning execution unit 132, in addition to information calculated based on speed information included in the input movement information stored in the input movement information storage unit 141, GPS coordinate information indicating GPS coordinates measured by a GPS system using a GPS device installed in the mobile body or terminal device 120 can also be used as a feature.

When information calculated based on speed information and GPS coordinate information are used as feature quantities, for example, a mobile object determined to be located on a link where congestion occurs frequently based on GPS coordinate information can be more easily determined to be in "congestion," while a mobile object determined to be located on a link where congestion is rare can be more easily determined to be in "non-congestion" even if its speed is low. Congestion determination can also be made based on movement information (travel distance, average speed, etc.) calculated from GPS coordinate information.

In addition, in machine learning by the machine learning execution unit 132, GPS movement distance information indicating the movement distance on GPS coordinates based on the input movement information and GPS coordinate information, and reference movement distance information indicating the movement distance of the mobile object calculated based on learning movement information excluding GPS coordinate information, can be used as feature quantities. In machine learning by the machine learning execution unit 132, for example, the deviation between the GPS movement distance information and the reference movement distance information can be used as feature quantities.

Here, GPS travel distance information is information indicating the travel distance of a mobile object calculated based on GPS coordinate information. GPS travel distance information includes processing errors by the GPS system. One factor that can cause processing errors by the GPS system is error based on the frequency with which GPS coordinate information is acquired. Specifically, since GPS coordinate information is merely information about a single point, the travel distance indicated by the GPS travel distance information indicates the length of the trajectory connecting each point. And since this length is the length of the trajectory (broken line) connecting each point, this length is not the length of the actual trajectory (smooth curve) of the mobile object, but merely an approximation of the length of the actual trajectory of the mobile object. In this case, the greater the speed or acceleration of the mobile object, the greater the difference between the trajectory connecting each point included in the GPS coordinate information and the actual trajectory of the mobile object.

On the other hand, reference travel distance information is, for example, information indicating a travel distance calculated from time information and speed information included in the input travel information, or information indicating a travel distance included in the input travel information. Travel distance information is, for example, information calculated based on the number of rotations of the wheels of a moving object.

While both GPS travel distance information and reference travel distance information indicate the distance traveled by a mobile object over a certain period of time, there are cases in which the GPS travel distance information and the reference travel distance information do not match due to processing errors by the GPS system. As mentioned above, for example, when a mobile object is moving at high speed or high acceleration, a discrepancy occurs between the trajectory connecting the points included in the GPS coordinate information and the actual trajectory of the mobile object, resulting in a discrepancy between the GPS travel distance information and the reference travel distance. Therefore, the greater the discrepancy between the GPS travel distance information and the reference travel distance information, the more likely the mobile object is to be evaluated as moving at high speed or high acceleration. Furthermore, mobile objects with a large discrepancy between the GPS travel distance information and the reference travel distance information can be more easily determined to be in a "non-traffic jam." In this way, in machine learning by the machine learning execution unit 132, the GPS travel distance information and the reference travel distance information can be used as feature quantities for traffic jam determination.

In machine learning by the machine learning execution unit 132, only information based on the movement information of the mobile object that is the target of congestion determination may be used as a feature, or information based on the movement information of other mobile objects, such as mobile objects present in the vicinity, may also be used as a feature.

The input movement information acquisition unit 140 accesses the information processing system 110, acquires movement information of at least one moving object from the information processing system 110, and stores the information in the input movement information storage unit 141. Note that the input movement information acquisition unit may also access the terminal device 120 and acquire the movement information of the moving object directly from the terminal device 120.

The timing at which the input movement information acquisition unit 140 acquires the input movement information from the information processing system 110 or the terminal device 120 can be set arbitrarily. For example, the input movement information acquisition unit 140 may acquire the input movement information from the terminal device 120 immediately when the information processing system 110 or the terminal device 120 collects the movement information, or the input movement information acquisition unit 140 may acquire the input movement information from the information processing system 110 or the terminal device 120 at a fixed timing, such as every second. Furthermore, the input movement information acquisition unit 140 may acquire the input movement information from the information processing system 110 or the terminal device 120 in response to an instruction from the information processing system 110 or the terminal device 120.

The input movement information may be movement information immediately before the determination by the moving body determination unit 142, past movement information, or virtual movement information calculated by simulation. The input movement information may also include not only the movement information of the moving body that is the subject of determination by the moving body determination unit 142, but also the movement information of other moving bodies, such as moving bodies that exist around the moving body.

Figure 3 is a diagram showing an example of information stored in the input movement information storage unit 141. Information stored in the input movement information storage unit 141 includes, for example, a moving object ID, time information, speed information, and GPS coordinate information.

The moving object determination unit 142 inputs the input movement information stored in the input movement information storage unit 141 to the determiner 133, determines the traffic congestion status of each moving object, and stores the information in the moving object determination information storage unit 144.

Figure 4 is a diagram showing an example of a determination by the moving object determination unit 142 using the determiner 133. The determiner 133 classifies the moving object, for example, for each branch, based on the input movement information stored in the input movement information storage unit 141, and ultimately generates a congestion determination result for the moving object. The congestion determination result includes, for example, "congested," "slightly congested," "empty," etc. Note that the congestion status information may be a qualitative indicator such as "A," "B," or "C," or a quantitative indicator such as a percentage. Furthermore, the indicators for the congestion determination result for the moving object and the learning congestion status information may be the same or different. Note that while Figure 4 shows a determination method that utilizes a decision tree learning method, the machine learning algorithm is not limited to this.

Based on the congestion determination result by the mobile object determination unit 142 using the determiner 133, the mobile object determination information generation unit 143 generates mobile object determination information indicating information associating the congestion determination result by the mobile object determination unit 142 using the determiner 133 with mobile object identification information that identifies the mobile object, and stores this information in the mobile object determination information storage unit 144.

Figure 5 is a diagram showing an example of information stored in the mobile object determination information storage unit 144. The information stored in the mobile object determination information storage unit 144 includes, for example, a mobile object ID, a link ID, time information, and congestion determination result information related to the congestion determination results for each mobile object by the determiner 133. The link ID is link identification information that identifies the link on which the mobile object is located. The link ID may be generated, for example, on the congestion determination system 100, or on the information processing system 110, or may be included in the input movement information stored in the input movement information storage unit 141. The link ID may also be information determined based on GPS coordinate information, or may be information based on information obtained by the mobile object or terminal device 120 from equipment installed on the road.

The aggregated information generation unit 150 aggregates the congestion determination results for each mobile object by the determiner 133 for each link based on the mobile object determination information stored in the mobile object determination information storage unit 144, and stores the results in the aggregated information storage unit 151. The aggregated information generation unit 150 may, for example, aggregate the number of mobile objects corresponding to each congestion determination result, or aggregate the proportion of the number of mobile objects for each congestion determination result to the total number of mobile objects. The aggregated information generation unit 150 may also classify the congestion determination results into multiple categories and then aggregate them.

Figure 6 is a diagram showing an example of information stored in the tally information storage unit 151. The information stored in the tally information storage unit 151 includes, for example, a link ID, time information, and judgment result tally information. In Figure 6, the judgment result tally information indicates the number of mobile objects that exist within the same link and have been given each congestion judgment result, but the tallying method is not limited to this.

The link determination unit 152 determines the congestion status of each link based on the aggregated information stored in the aggregated information storage unit 151, and stores the determined status in the link determination information storage unit 154. The link determination unit 152 determines the congestion status of each link based on, for example, the aggregated information of the determination results for each link stored in the aggregated information storage unit 151.

The conditions for determining the congestion status of each link can be, for example, (a) "complete congestion" when the number of "congested" or "slightly congested" mobile units is 1 or more and the number of "free" mobile units is 0; (b) "partial congestion" when the number of "congested" or "slightly congested" mobile units is 1 or more and the number of "free" mobile units is also 1 or more; and (c) "free" when the number of "congested" and "slightly congested" mobile units is 0 and the number of "free" mobile units is 1 or more. Here, "complete congestion" refers to a state in which all lanes are congested, "partial congestion" refers to a state in which only a portion of the link is congested, such as a state in which congested and non-congested lanes are mixed or where only a specific section of the link is congested, and "free" refers to a state in which no lanes or sections are congested; however, the indicators for the congestion determination results for each link are not limited to these. Furthermore, the indicator for the congestion determination result for each link may be a qualitative indicator such as "A," "B," or "C," or it may be a quantitative indicator such as a percentage or number of moving objects.

The conditions for determining the congestion status of each link may be multiple or more mobile objects instead of one or more, or multiple conditions may be combined. The conditions for determining the congestion status of each link may also be based on the number of mobile objects on links other than the link in question, such as links adjacent to the link in question. The conditions for determining the congestion status of each link may also be based on the proportion of the number of mobile objects for each congestion determination result to the total number of mobile objects.

Based on the congestion determination result by the link determination unit 152, the link determination information generation unit 153 generates link determination information indicating information associating the congestion determination result by the link determination unit 152 with link identification information that identifies the link, and stores this information in the link determination information storage unit 154.

Figure 7 shows an example of information stored in the link determination information storage unit 154. The information stored in the link determination information storage unit 154 includes, for example, a link ID, time information, and congestion determination result information related to the congestion determination result by the link determination unit 152.

The determination information providing unit 160 provides at least one of the moving body determination information stored in the moving body determination information storage unit 144 and the link determination information stored in the link determination information storage unit 154 to the information processing system 110 as reference information when analyzing traffic conditions.

Figure 8 is a flowchart showing an example of machine learning processing in the congestion judgment system 100. First, the learning information acquisition unit 130 acquires learning information from the information processing system 110 and stores it in the learning information storage unit 131 (S801). Then, the machine learning execution unit 132 executes machine learning based on the learning information stored in the learning information storage unit 131, and generates a determinator 133 (S802).

Figure 9 is a flowchart showing an example of the congestion determination process in the congestion determination system 100. First, the input movement information acquisition unit 140 acquires movement information of moving objects from the information processing system 110 or the terminal device 120 and stores it in the input movement information storage unit 141 (S901). Next, the moving object determination unit 142 inputs the input movement information stored in the input movement information storage unit 141 to the determiner 133, determines the congestion status of each moving object, and stores the information in the moving object determination information storage unit 144 (S902).

The aggregated information generation unit 150 aggregates the mobile object determination information stored in the mobile object determination information storage unit 144 for each link and stores it in the aggregated information storage unit 151 (S903). The link determination unit 152 determines the congestion status of each link based on the aggregated information stored in the aggregated information storage unit 151 (S904). Then, at least one of the mobile object determination information and the link determination information is provided to the information processing system 110 as reference information for analyzing traffic conditions (S905).

The above describes one embodiment of the present invention. The congestion determination system 100 can determine the congestion status of at least one mobile object based on a determiner that determines the congestion status of each mobile object, constructed by machine learning based on learning information, and input movement information that indicates the movement information of at least one mobile object. This makes it possible to determine the congestion status of each mobile object on more roads with high accuracy.

In addition, the congestion determination system 100 can determine the congestion status of each link and the congestion status of each portion of a link based on the congestion status of each mobile object. This makes it possible to grasp the congestion status of a link, a situation in which congested and uncongested lanes are mixed within a link, and a situation in which only specific sections within a link are congested.

In addition, the congestion determination system 100 can provide the congestion determination results for each mobile object and each link to the information processing system 110. This allows the information processing system 110 to use the determination results from the congestion determination system 100 to analyze traffic conditions or support the driving of mobile objects.

Furthermore, the results of the congestion status determination for each mobile object by the congestion determination system 100 are not limited to being used to determine the congestion status of each link. For example, they can also be used to identify mobile objects to which a service can be provided when providing a unique service to mobile objects in a traffic jam, or to quantify the likelihood of congestion on a road by analyzing the relationship between the movement of congested vehicles and the shape of the road, etc.

Note that the present embodiment is intended to facilitate understanding of the present invention and is not intended to limit the scope of the present invention. The present invention may be modified or improved without departing from its spirit, and equivalents thereof are also included within the scope of the present invention.

100 Traffic congestion determination system, 110 Information processing system, 120 Terminal device, 130 Learning information acquisition unit, 131 Learning information storage unit, 132 Machine learning execution unit, 133 Determinator, 140 Input movement information acquisition unit, 141 Input movement information storage unit, 142 Moving object determination unit, 143 Moving object determination information generation unit, 144 Moving object determination information storage unit, 150 Aggregated information generation unit, 151 Aggregated information storage unit, 152 Link determination unit, 153 Link determination information generation unit, 154 Link determination information storage unit, 160 Determination information provision unit

Claims (11)

a determiner for determining the congestion status of each of the mobile bodies, the determiner being constructed by machine learning based on learning information including learning movement information indicating movement information including speed information indicating the speed of each of the multiple mobile bodies and acceleration information indicating the acceleration, and learning congestion status information indicating the congestion status of the corresponding mobile body;
an input movement information acquisition unit that acquires input movement information indicating the movement information of each of at least one moving object;
a moving body determination unit that inputs the input movement information to the determiner and determines a traffic congestion state of each of the at least one moving body;
a summary information generating unit that generates summary information by correlating the congestion determination result of the at least one mobile object determined by the mobile object determining unit with link identification information that identifies a link on which the at least one mobile object is located;
a link determination unit that determines a congestion state of each of at least one of the links based on the aggregated information,
the link determination unit determines a congestion state of each of the at least one links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregate information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Traffic congestion detection system. a determiner for determining the congestion status of each of the mobile bodies, the determiner being constructed by machine learning based on learning movement information indicating movement information including speed information indicating the speed of each of the multiple mobile bodies and GPS coordinate information indicating the GPS coordinates measured by a GPS system, learning congestion status information indicating the congestion status of the corresponding mobile body, and a correspondence relationship associating congestion occurrence frequency with the GPS coordinates;
an input movement information acquisition unit that acquires input movement information indicating the movement information of each of at least one moving object;
a moving body determination unit that inputs the input movement information to the determiner and determines a traffic congestion state of each of the at least one moving body;
a summary information generating unit that generates summary information by correlating the congestion determination result of the at least one mobile object determined by the mobile object determining unit with link identification information that identifies a link on which the at least one mobile object is located;
a link determination unit that determines a congestion state of each of at least one of the links based on the aggregated information,
the link determination unit determines a congestion state of each of the at least one links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregation information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Traffic congestion detection system. a determiner for determining the congestion status of each of the mobile bodies, the determiner being constructed by machine learning based on learning movement information indicating movement information including speed information indicating the speed of each of the multiple mobile bodies and GPS coordinate information indicating the GPS coordinates measured by a GPS system, learning congestion status information indicating the congestion status of the corresponding mobile body, and a correspondence relationship associating congestion occurrence frequency with the GPS coordinates;
an input movement information acquisition unit that acquires input movement information indicating the movement information of each of at least one moving object;
a moving body determination unit that inputs the input movement information to the determiner and determines a congestion state of each of the at least one moving body,
the learning movement information includes at least the speed information and the GPS coordinate information of each of the plurality of moving bodies;
the determiner is constructed by machine learning using, as features, at least the speed information, the GPS coordinate information, and information calculated based on a correspondence relationship between a congestion occurrence frequency and a GPS coordinate;
the input movement information further includes GPS coordinate information indicating a GPS coordinate measured by a GPS system;
the determiner is constructed by machine learning using, as at least a feature, information calculated based on a discrepancy between GPS movement distance information indicating a movement distance of the moving body on GPS coordinates, which is calculated based on the GPS coordinate information of each of the plurality of moving bodies, and reference movement distance information indicating a movement distance of the moving body, which is calculated based on the learning movement information excluding the GPS coordinate information of each of the plurality of moving bodies;
Traffic congestion detection system. The congestion determination system according to claim 1 , wherein the link determination unit determines congestion conditions of a plurality of specific portions included in each of at least one of the links based on the aggregate information. the learning movement information includes at least the speed information and the acceleration information of each of the plurality of moving bodies;
the determiner is constructed by machine learning using information calculated based on the speed information and the acceleration information as a feature,
the input movement information includes at least the velocity information and the acceleration information of each of the at least one moving object;
The congestion determination system according to claim 1 . The congestion determination system of claim 1 or 2 further comprises a determination information providing unit that provides the congestion status of at least one mobile object determined by the mobile object determination unit to an information processing system as reference information when analyzing traffic conditions. The congestion determination system of claim 1 or 2 further comprises a determination information providing unit that provides at least one of the congestion status of the at least one mobile body determined by the mobile body determination unit and the congestion status of the at least one link determined by the link determination unit to an information processing system as reference information when analyzing traffic conditions. a computer including a determiner for determining a congestion status of each of a plurality of moving bodies, the determiner being constructed by machine learning based on learning information including learning movement information indicating movement information including speed information indicating the speed of each of a plurality of moving bodies and acceleration information indicating the acceleration, and learning congestion status information indicating the congestion status of the corresponding moving body;
acquiring input movement information indicating the movement information of each of at least one moving object;
inputting the input movement information to the determiner to determine a congestion state of each of the at least one moving body;
generating aggregate information by correlating the congestion determination result of the at least one mobile object with link identification information that identifies the link on which the at least one mobile object is located;
determining a congestion state of each of the at least one links based on the aggregated information;
determining a congestion state of each of at least one of the links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregate information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Method for determining congestion. On the computer,
a determiner for determining the congestion status of each of the mobile bodies, the determiner being constructed by machine learning based on learning information including learning movement information indicating movement information including speed information indicating the speed of each of the multiple mobile bodies and acceleration information indicating the acceleration, and learning congestion status information indicating the congestion status of the corresponding mobile body;
an input movement information acquisition unit that acquires input movement information indicating the movement information of each of at least one moving object;
a moving body determination unit that inputs the input movement information to the determiner and determines a traffic congestion state of each of the at least one moving body;
a summary information generating unit that generates summary information by correlating the congestion determination result of the at least one mobile object determined by the mobile object determining unit with link identification information that identifies a link on which the at least one mobile object is located;
a link determination unit that determines a congestion state of each of at least one of the links based on the aggregated information,
the link determination unit determines a congestion state of each of the at least one links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregation information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Congestion detection program. a computer including a determiner for determining the congestion status of each of a plurality of mobile bodies, the determiner being constructed by machine learning based on learning movement information indicating movement information including speed information indicating the speed of each of a plurality of mobile bodies and GPS coordinate information indicating GPS coordinates measured by a GPS system, learning congestion status information indicating the congestion status of the corresponding mobile bodies, and a correspondence relationship associating congestion occurrence frequency with the GPS coordinates;
acquiring input movement information indicating the movement information of each of at least one moving object;
inputting the input movement information to the determiner to determine a congestion state of each of the at least one moving body;
aggregating the congestion determination results of the at least one mobile object in association with link identification information identifying the link on which the at least one mobile object is located, thereby generating aggregate information;
determining a congestion state of each of the at least one links based on the aggregated information;
determining a congestion state of each of at least one of the links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregate information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Method for determining congestion. On the computer,
a determiner for determining the congestion status of each of the mobile bodies constructed by machine learning based on: learning movement information indicating movement information including speed information indicating the speed of each of the multiple mobile bodies and GPS coordinate information indicating GPS coordinates measured by a GPS system; learning information including congestion status information indicating the congestion status of the corresponding mobile body; and a correspondence relationship correlating congestion occurrence frequency with the GPS coordinates;
an input movement information acquisition unit that acquires input movement information indicating the movement information of each of at least one moving object;
a moving body determination unit that inputs the input movement information to the determiner and determines a traffic congestion state of each of the at least one moving body;
a summary information generating unit that generates summary information by correlating the congestion determination result of the at least one mobile object determined by the mobile object determining unit with link identification information that identifies a link on which the at least one mobile object is located;
a link determination unit that determines a congestion state of each of at least one of the links based on the aggregated information,
the link determination unit determines a congestion state of each of the at least one links based on a congestion determination result of each mobile body corresponding to the link identification information in the aggregation information;
The congestion state includes a partial congestion state in which congested lanes and non-congested lanes are mixed.
Congestion detection program.