Center device, signal disregarding detection system, and signal disregarding detection method

Abstract

To more accurately suppress erroneous detection of signal ignorance.SOLUTION: According to an aspect of the embodiments, a center device includes a controller configured to receive, from an in-vehicle device, information including position information at which the in-vehicle device has detected signal ignoring, calculate a traffic volume of vehicles in a predetermined period at a point where signal ignoring frequently occurs based on the information, calculate an occurrence probability of signal ignoring at the point from the number of occurrences of signal ignoring based on the information and the traffic volume, and determine that the point where the occurrence probability is equal to or greater than a threshold value is an abnormality detection point of signal ignoring.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The disclosed embodiments relate to a center device, a red light running detection system, and a red light running detection method. [Background technology]

[0002] Conventionally, a technology is known that detects red lights from an image in front of a vehicle taken by an onboard camera, determines whether the red light has been ignored based on the vehicle's speed information, and issues an alarm to the driver of the vehicle if a red light has been ignored.

[0003] However, even with this technology, depending on road conditions, such as when roads intersect diagonally and multiple traffic lights are located close together, or when warning signals are installed, a traffic light may be mistakenly detected as having run a red light even when it is not.

[0004] Therefore, for example, in the case of a warning signal, a technology has been proposed that prevents false detection of red light violations by not determining that the vehicle has violated the signal if the vehicle has not crossed the stop line or crosswalk after detecting a red light (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-180309 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the above-mentioned conventional technology has room for further improvement in terms of more accurately suppressing false detection of red light ignoring.

[0007] For example, when the technology disclosed in Patent Document 1 is used, it is not possible to prevent false detection of red light ignoring when stop lines or crosswalks are unclear or when stop lines or crosswalks do not exist in the first place.

[0008] Furthermore, even if the technology disclosed in Patent Document 1 is used, it is not possible to suppress false detection of red light violations in road conditions such as when roads intersect diagonally and multiple traffic lights are installed close to each other, as described above.

[0009] One aspect of the embodiment has been made in consideration of the above, and aims to provide a center device, a signal running detection system, and a signal running detection method that can more accurately suppress false detection of signal running. [Means for solving the problem]

[0010] The center device according to one aspect of the embodiment includes a controller that receives information from an on-board device, including location information at which the on-board device detects red light ignition, calculates the volume of vehicles passing through a location where red light ignition is frequent based on the information, calculates the probability of red light ignition occurring at the location from the number of red light ignitions based on the information and the volume of vehicles passing through, and determines that a location where the probability of red light ignition occurring is equal to or greater than a threshold is a location where a red light ignition abnormality has been detected. [Effects of the Invention]

[0011] According to one aspect of the embodiment, an anomaly detection point where the probability of red light running is abnormally high and the possibility of a false positive is considered to be extremely high can be identified based on the probability of a red light running event occurring. In other words, according to one aspect of the embodiment, it is possible to more accurately suppress false positive detection of red light running. [Brief explanation of the drawings]

[0012] [Figure 1] FIG. 1 is a diagram (part 1) outlining a method for detecting red light violation according to an embodiment. [Figure 2]FIG. 2 is a schematic explanatory diagram (part 2) of the red light running detection method according to the embodiment. [Figure 3] FIG. 3 is a diagram (part 3) outlining the traffic light running detection method according to the embodiment. [Figure 4] FIG. 4 is a diagram illustrating an example of the configuration of a red light violation detection system according to the embodiment. [Figure 5] FIG. 5 is a diagram illustrating an example of the configuration of a drive recorder according to the embodiment. [Figure 6] FIG. 6 is a diagram illustrating an example of the configuration of a center device according to the embodiment. [Figure 7] FIG. 7 is a diagram (part 1) illustrating step S3 executed by the controller of the center device. [Figure 8] FIG. 8 is a diagram (part 2) illustrating step S3 executed by the controller of the center device. [Figure 9] FIG. 9 is a diagram (part 3) illustrating step S3 executed by the controller of the center device. [Figure 10] FIG. 10 is an explanatory diagram (part 4) of step S3 executed by the controller of the center device. [Figure 11] FIG. 11 is a diagram (part 5) illustrating step S3 executed by the controller of the center device. [Figure 12] FIG. 12 is an explanatory diagram (part 6) of step S3 executed by the controller of the center device. [Figure 13] FIG. 13 is a diagram showing a processing sequence executed by the red light running detection system according to the embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, embodiments of the center device, the red light running detection system, and the red light running detection method disclosed in the present application will be described in detail with reference to the accompanying drawings. Note that the present invention is not limited to the following embodiments.

[0014] In the following description, the vehicle-mounted device according to the embodiment is assumed to be a drive recorder 10 (see FIG. 1). The center device according to the embodiment is assumed to be a center device 100 (see FIG. 1). The red light running detection method according to the embodiment is assumed to be a red light running detection method executed by a controller 103 (see FIG. 6) included in the center device 100.

[0015] The drive recorder 10 is assumed to be a communication type that is capable of communicating with the center device 100. In the following description, the red light running detection system according to the embodiment is assumed to be a red light running detection system 1 (see FIG. 1) that includes the drive recorder 10 and the center device 100.

[0016] In the following, when it is necessary to distinguish between multiple identical elements, a number in the form "-n" (n is a natural number equal to or greater than 1) may be added after the symbol indicating the element. When there is no particular need to distinguish between them, this numbering will not be used.

[0017] Furthermore, the expressions "predetermined," "specific," and "fixed" in the following description may be read as "predetermined."

[0018] First, an overview of a traffic light running detection method according to an embodiment will be described with reference to Figs. 1 to 3. Fig. 1 is a diagram (part 1) outlining the traffic light running detection method according to an embodiment. Fig. 2 is a diagram (part 2) outlining the traffic light running detection method according to an embodiment. Fig. 3 is a diagram (part 3) outlining the traffic light running detection method according to an embodiment. As shown in Fig. 1, a traffic light running detection system 1 includes drive recorders 10-1, 10-2, ... 10-m (m is a natural number equal to or greater than 3) and a center device 100.

[0019] The drive recorder 10 is a video recording device mounted on a vehicle. The drive recorder 10 according to the embodiment includes a camera 12a. The camera 12a is provided so as to be able to capture at least an external image of the area in front of the vehicle.

[0020] While the vehicle is running, the drive recorder 10 executes operation recording by recording a certain period of operation record data, including images of the exterior of the vehicle captured by the camera 12a, in a ring buffer memory in an overwritable manner. The certain period is, for example, 24 hours. In addition to the images of the exterior of the vehicle, the operation record data includes at least the date and time, position information (latitude and longitude), direction (heading), and vehicle speed.

[0021] In addition, while recording the driving operation, the drive recorder 10 performs image recognition processing on the vehicle exterior video using, for example, an AI model for image recognition. The AI ​​model is, for example, a DNN (Deep Neural Network) model trained using a machine learning algorithm. This AI model is trained in advance so as to be able to recognize the type, position, color, etc. of each object shown in the vehicle exterior video. The AI ​​model is configured to be able to recognize at least traffic lights and the color of their lights shown in the vehicle exterior video.

[0022] The drive recorder 10 is also configured to be able to detect at least a red light running event as a specific event. As a result of image recognition, the drive recorder 10 determines that a red light has been run when, for example, the traffic light ahead of the vehicle is red and the vehicle speed is equal to or greater than a certain level, and detects this as a red light running event. In this way, the drive recorder 10 records driving and detects red light running events (step S1).

[0023] Furthermore, the drive recorder 10 may be configured to be able to detect various specific events, such as the occurrence of an accident, a near miss, or reaching a specified location, based on, for example, changes in vehicle speed, changes in G-value, changes in latitude and longitude, etc.

[0024] When the drive recorder 10 detects a specific event, it sets the operation record data and event data for a certain period of time before and after the detection time to be overwritten-protected. Alternatively, the drive recorder 10 records the operation record data and event data for a certain period of time before and after the detection time on a separate recording medium. This overwriting protection setting or recording on a separate recording medium may be performed in response to an instruction from the center device 100.

[0025] Among the event data, the red light running event data includes at least the date and time, latitude, longitude, and direction when the red light running event occurred.

[0026] The drive recorder 10 also transmits the recorded operation record data and red light running event data to the center device 100 periodically or in real time (step S2). The operation record data and red light running event data transmitted to the center device 100 are stored in the center device 100 in association with a vehicle ID, which is identification information of the vehicle in which the drive recorder 10, which is the transmission source, is installed.

[0027] If the drive recorder 10 detects a red light running event in step S1, it issues a warning to the driver via an HMI (Human Machine Interface) unit 13 (see FIG. 5) including a display or a speaker. At this time, the drive recorder 10 controls the notification based on area information 14c delivered from the center device 100. The area information 14c is information about each area that serves as a counting unit for red light running frequent locations, which will be described later. The area information 14c corresponds to an example of "information about a location." The area information 14c includes the "direction" of the vehicle in each area and the "area type."

[0028] The drive recorder 10 suppresses notification if the "direction" set in the area information 14c for the area corresponding to the latitude and longitude where the red light running event occurred matches the direction of travel of the vehicle, and if a value indicating "red light running suppression" is set in the "area type".

[0029] Regarding the setting of the area information 14c, the center device 100 determines an abnormality detection point based on the occurrence probability of red light running at a red light frequent occurrence point, based on the driving record data and red light running event data acquired from the drive recorder 10 (step S3).

[0030] Specifically, the center device 100 periodically (for example, once a month) analyzes the accumulated driving record data and red light running event data to identify anomaly detection points where the probability of red light running occurring is abnormally high. The center device 100 then sets the aforementioned "red light running suppression" for the area type of the area corresponding to the anomaly detection point. This allows the center device 100 to control notification suppression when the drive recorder 10 detects a red light running at the anomaly detection point.

[0031] In this embodiment, the center device 100 defines each of the aforementioned areas as a mesh divided into, for example, 50m squares based on latitude and longitude.The center device 100 then calculates the number of vehicles per day at which red light running events occur and the number of vehicles per day passing through in each direction for each mesh.The center device 100 then calculates the occurrence probability of red light running events in each direction for each mesh based on the calculated number of vehicles per day at which red light running events occur and the calculated number of vehicles per day passing through, and determines the anomaly detection point based on this occurrence probability.

[0032] More specifically, the center device 100 first performs a "frequent occurrence point calculation process" in step S3. In the "frequent occurrence point calculation process," the vehicle IDs, latitude, longitude, and direction of red light running events over a one-month period are tallied for each mesh, and the number of occurrences (number of events) per day for each direction in each mesh is calculated. Alternatively, the number of times a vehicle that detected a red light running event passed by may be used.

[0033] Next, the center device 100 performs a "traffic volume calculation process." In the "traffic volume calculation process," the vehicle IDs and latitudes and longitudes of the operation record data for one month are tallied for each mesh, and the number of vehicles passing through each mesh per day for each direction is calculated. The number of vehicles passing through may also be the number of times of passage.

[0034] Next, the center device 100 performs an "occurrence probability calculation process." In the "occurrence probability calculation process," the center device 100 calculates the event occurrence probability by calculating the "number of occurrences per day divided by the number of passing vehicles per day" for each direction in each mesh.

[0035] The center device 100 then performs an "anomaly detection point determination process." In this process, it is determined whether the occurrence probability calculated for each direction in each mesh is equal to or greater than a predetermined threshold (for example, 50%). If it is equal to or greater than the threshold, the center device 100 updates the area information 102d (see FIG. 6) (step S4). The center device 100 sets the corresponding direction as the "direction" of the area indicated by the latitude and longitude range of the mesh where the occurrence probability is equal to or greater than the threshold, and sets "red light running prevention" as the "area type."

[0036] Then, the center device 100 distributes the area information 102d to the drive recorder 10 of each vehicle (step S5), and stores it as area information 14c in each drive recorder 10. That is, the controller 103 of the center device 100 transmits the area information 102d determined to be an abnormality detection point to the drive recorder 10, and causes the drive recorder 10 to suppress, based on the area information 102d, the notification of an alarm to the vehicle occupants when a red light ignorance is detected at that point.

[0037] The difference in operation before and after updating the area information 14c is shown in Figures 2 and 3. As shown in the left diagram of Figure 2, in a latitude and longitude range corresponding to a certain mesh, assume that the traffic light detected when a vehicle is traveling north is a warning signal, which makes it easy for a false detection of a red light running event to occur. If the area information 14c does not have a setting to suppress notifications corresponding to this false detection, a red light running warning would be sent to the HMI unit 13, as shown in the left diagram of Figure 2.

[0038] In contrast, with the red light running detection method according to the embodiment, if the direction of the area indicated by the latitude range of the corresponding mesh is set to "north" and the area type is set to "red light running suppression," the notification of an alarm to the HMI unit 13 will be suppressed, as shown in the right diagram of Figure 2.

[0039] Also, as shown in the left diagram of FIG. 3, in a latitude and longitude range corresponding to a certain mesh, assume that there is a traffic light in an intersecting direction near a traffic light that is detected when a vehicle is traveling north. Therefore, even if the traffic light in the traveling direction is green, a false detection of a red light running event is likely to occur due to the red light of the traffic light in the intersecting direction. If the area information 14c does not have a setting to suppress notifications corresponding to this false detection, a red light running warning will be sent to the HMI unit 13, as shown in the left diagram of FIG. 3.

[0040] In contrast, with the red light running detection method according to the embodiment, if the direction of the area indicated by the latitude range of the corresponding mesh is set to "north" and the area type is set to "red light running suppression," the notification of an alarm to the HMI unit 13 will be suppressed, as shown in the right diagram of Figure 3.

[0041] As described above, in the red light running detection method according to the embodiment, the controller 103 of the center device 100 receives, from the drive recorder 10, red light running event data including the latitude and longitude at which the drive recorder 10 detected a red light running and operation record data. Furthermore, the controller 103 calculates the volume of vehicle traffic over a certain period at a location where red light running is frequent based on this data. Furthermore, the controller 103 calculates the probability of red light running at the location from the number of red light running incidents based on the red light running event data and the operation record data and the traffic volume. Furthermore, the controller 103 determines that the location where the occurrence probability is equal to or greater than a threshold is a red light running anomaly detection location.

[0042] Therefore, according to the signal running detection method of the embodiment, it is possible to identify anomaly detection points where the probability of signal running is abnormally high and the possibility of false detection is considered to be extremely high, based on the probability of signal running events occurring. In other words, according to the signal running detection method of the embodiment, it is possible to more accurately suppress false detection of signal running.

[0043] An example of the configuration of the red light running detection system 1 including the center device 100 to which the red light running detection method according to the above-described embodiment is applied will be described in more detail below.

[0044] 4 is a diagram illustrating an example of the configuration of a red light running detection system 1 according to an embodiment. As illustrated in FIG. 4, the red light running detection system 1 includes drive recorders 10-1, 10-2, ... 10-m (m is a natural number equal to or greater than 1) and a center device 100.

[0045] Each drive recorder 10 and the center device 100 are connected to each other so as to be able to communicate with each other via a network N1, which may be the Internet, a mobile phone network, a C-V2X (Cellular Vehicle to Everything) communication network, or the like.

[0046] As described above, the drive recorder 10 executes driving record and traffic light running event detection, and records driving record data for a certain period of time in a ring buffer memory in an overwritable manner.

[0047] Furthermore, as described above, the drive recorder 10 is configured to perform image recognition processing on the outside image of the vehicle using, for example, an AI model for image recognition in parallel with recording the driving operation, and to be able to detect at least a red light ignoring event as a specific event.

[0048] When the drive recorder 10 detects a specific event, it sets the operation record data for a certain period of time before and after the detection time and the event data including the red light running event data to be protected from overwriting. The drive recorder 10 also transmits the recorded operation record data and event data to the center device 100 periodically or in real time.

[0049] The center device 100 is realized as, for example, a private cloud. The center device 100 is managed by, for example, a business operator (such as an insurance company) that operates a data center. The center device 100 collects and stores driving record data and event data (at least red light running event data) transmitted from each drive recorder 10.

[0050] The center device 100 also determines an abnormality detection point based on the probability of red light running occurring at a red light running frequent occurrence point, based on the accumulated driving record data and red light running event data. The center device 100 also sets the area type of the area including the determined abnormality detection point to "red light running suppression" for the area information 102d, and distributes it to each drive recorder 10.

[0051] Next, a configuration example of the drive recorder 10 will be described. Fig. 5 is a diagram showing a configuration example of the drive recorder 10 according to the embodiment. As shown in Fig. 5, the drive recorder 10 has a communication unit 11, a sensor unit 12, an HMI unit 13, a storage unit 14, and a controller 15.

[0052] The communication unit 11 is realized by a network adapter etc. The communication unit 11 is wirelessly connected to the network N1, and transmits and receives information to and from the center device 100 via the network N1.

[0053] The sensor unit 12 is a group of various sensors mounted on the drive recorder 10. The sensor unit 12 includes, for example, a camera 12a, a G sensor 12b, a GPS (Global Positioning System) sensor 12c, and a vehicle speed sensor 12d.

[0054] As described above, the camera 12a is provided so as to be able to capture at least an external image in front of the vehicle. The camera 12a is attached near the windshield, the dashboard, etc. The camera 12a may also be attached near the rear window, etc., so as to be able to capture an image behind the vehicle.

[0055] The G sensor 12b measures acceleration (G) applied to the drive recorder 10. The GPS sensor 12c measures the GPS position (latitude and longitude) of the vehicle. Note that the sensor unit 12 may include various sensors other than the camera 12a, the G sensor 12b, the GPS sensor 12c, and the vehicle speed sensor 12d.

[0056] In addition to the sensor unit 12, the drive recorder 10 is also connected to an in-vehicle sensor 5, which is a group of various sensors mounted on the vehicle. The in-vehicle sensor 5 includes, for example, an accelerator sensor and a brake sensor. The in-vehicle sensor 5 is connected to the drive recorder 10 via an in-vehicle network such as a CAN (Controller Area Network).

[0057] The HMI unit 13 is a component that provides interface components related to input and output to the driver or the like who uses the drive recorder 10. The HMI unit 13 includes an input interface that accepts input operations from the driver or the like. The HMI unit 13 also includes an output interface that presents visual information and audio information to the driver or the like.

[0058] For example, the HMI unit 13 includes a display and a speaker. The display is realized by, for example, a touch panel display. The touch panel display corresponds to the input interface and output interface described above. The touch panel display displays, for example, operation components for operating the drive recorder 10.

[0059] Note that these operation components may not be displayed as software components on the touch panel display, but may be provided as hardware components in the HMI unit 13. The speaker corresponds to an output interface that outputs guidance voices and the like from the drive recorder 10.

[0060] The storage unit 14 is realized by a storage device such as a read-only memory (ROM), a random access memory (RAM), a flash memory, etc. In the example of Fig. 5, the storage unit 14 stores operation record data 14a, an image recognition model 14b, area information 14c, and red light running event data 14d.

[0061] The operation record data 14a and the red light running event data 14d correspond to the "operation record data" and the "red light running event data" respectively described with reference to Fig. 1, and therefore a description thereof will be omitted here. The area information 14c has also been described, and therefore a description thereof will be omitted here.

[0062] The image recognition model 14b corresponds to the AI ​​model described above. After being read into the controller 15 as an AI model, the image recognition model 14b is configured to be able to detect various objects and the like included in each frame when each frame of an outside-vehicle video is input to the controller 15.

[0063] The image recognition model 14b is provided so that when each frame of the outside-of-vehicle image is input, it can recognize the type, position, color, etc. of each of the above-mentioned objects shown in each frame.

[0064] The controller 15 corresponds to a so-called processor. The controller 15 is realized by a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphical Processing Unit), or the like. The controller 15 executes a program according to an embodiment (not shown) stored in the storage unit 14, using RAM as a work area. The controller 15 can also be realized by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array).

[0065] The controller 15 executes information processing according to the processing sequence shown in Fig. 13. The explanation using Fig. 13 will be given later.

[0066] Next, a configuration example of the center device 100 will be described. Fig. 6 is a diagram showing a configuration example of the center device 100 according to the embodiment. As shown in Fig. 6, the center device 100 includes a communication unit 101, a storage unit 102, and a controller 103. An HMI unit 50 is also connected to the center device 100.

[0067] The HMI unit 50 is a component that provides interface components related to input and output to an operator or the like who operates the center device 100. The HMI unit 50 includes an input interface that accepts input operations from an operator or the like. The input interface is realized by, for example, a touch panel. The input interface may also be realized by a keyboard, a mouse, a pen tablet, a microphone, or the like. The input interface may also be realized by software components.

[0068] The HMI unit 50 also includes an output interface that presents visual information and audio information to an operator, etc. The output interface is realized by, for example, a display, a speaker, etc. The HMI unit 50 may also be configured to provide the operator, etc. with an input interface and an output interface as a single unit by, for example, a touch panel display.

[0069] The communication unit 101 is realized by a network adapter etc. The communication unit 101 is connected to the network N1 by wire or wirelessly, and transmits and receives information to and from the drive recorder 10 via the network N1.

[0070] The storage unit 102 is realized by a storage device such as a ROM, a RAM, a flash memory, an HDD (Hard Disk Drive), etc. In the example of Fig. 6, the storage unit 102 stores a collected data DB (Database) 102a, a map information DB 102b, an analysis model 102c, and area information 102d.

[0071] The collected data DB 102a is a database that stores the driving record data and red light running event data collected from each drive recorder 10. The driving record data and red light running event data collected from each drive recorder 10 are linked to the vehicle ID of the vehicle that sent the data and stored in the collected data DB 102a.

[0072] The map information DB 102b is a database of map information that is the basis for mesh division. The analysis model 102c is a mathematical model including various calculation formulas used in the above-mentioned step S3, etc. The analysis model 102c may be an AI model trained to output anomaly detection points based on the driving record data and red light running event data. The area information 102d is master information of the area information 14c of each drive recorder 10.

[0073] The controller 103 corresponds to a so-called processor. The controller 103 is realized by a CPU, an MPU, a GPU, or the like. The controller 103 executes a program according to an embodiment (not shown) stored in the storage unit 102, using a RAM as a work area. The controller 103 can also be realized by an integrated circuit such as an ASIC or an FPGA.

[0074] The controller 103, like the above-mentioned controller 15, executes information processing according to the processing sequence shown in Fig. 13. This information processing includes the above-mentioned step S3. An explanation using Fig. 13 will be given later, but before that, the above-mentioned step S3 will be explained using specific data examples. Figs. 7 to 12 are explanatory diagrams (part 1) to (part 6) of step S3 executed by the controller 103 of the center device 100.

[0075] First, Figure 7 shows an example of a mesh. As shown in Figure 7, there is a mesh indicated by the latitude and longitude range of 34.801774 to 34.802228 latitude and 135.345134 to 135.345563 longitude. For convenience, the mesh ID, which is the identification information of the mesh, is assumed to be "X". Hereinafter, this mesh will be referred to as "Mesh X" where appropriate.

[0076] In this embodiment, each mesh is 50 meters square, but this is merely an example. The size of each mesh may be, for example, in accordance with the so-called "regional mesh" defined by the Ministry of Internal Affairs and Communications as a longitude-latitude grid on a map. Also, as in this embodiment, each mesh may be smaller than the regional mesh. Also, one side of the mesh may be less than 50 meters.

[0077] Next, Fig. 8 shows an example of one month's worth of traffic light running event data. The controller 103 performs a process of calculating frequent incident locations based on such traffic light running event data. In the example of Fig. 8, the latitude and longitude of data D1 and D2 are included in mesh X, but the latitude and longitude of each data item in data D3 are not included in mesh X. The latitude and longitude of each data item in this data D3 are assumed to be included in mesh Y.

[0078] In the frequent incident location calculation process, the controller 103 counts vehicle IDs for the same mesh and the same direction on a daily basis based on one month's worth of traffic light running event data. In the example of FIG. 8, data D1 is counted as one vehicle and is accumulated as data D4 shown in FIG. 9. Data D2 is counted as one vehicle and is accumulated as data D5 shown in FIG. 9. Furthermore, each data item in data D3 with the same vehicle ID on the same day is counted as one vehicle and is accumulated as data D6 shown in FIG. 9.

[0079] Then, based on such a counting result, the controller 103 calculates the daily average number of vehicles in which an event has occurred for each mesh and direction, as shown in the table below in FIG.

[0080] Next, Fig. 10 shows an example of one month's worth of operation record data. The controller 103 performs traffic volume calculation processing based on such operation record data. In the example of Fig. 10, the latitude and longitude of each piece of data D7 to D9 are included in mesh X, and the latitude and longitude of each other piece of data are not included in mesh X.

[0081] In the traffic volume calculation process, the controller 103 counts vehicle IDs for the same mesh and the same direction on a daily basis based on one month's worth of operation record data. In the example of FIG. 10, each data item in data D7 is counted as one vehicle and accumulated into data D10 shown in FIG. 11. Also, each data item in data D8 is counted as one vehicle and accumulated into data D11 shown in FIG. 11. Also, although the latitude and longitude of each data item in data D9 are included in mesh X, no traffic light running event has occurred in the direction "north" (see FIG. 8), so it is not counted.

[0082] Then, based on such aggregation results, the controller 103 calculates the daily average number of passing vehicles for each mesh and direction, as shown in the lower table of Fig. 11. Then, the controller 103 divides the number of event-occurring vehicles (daily average) calculated in the frequent occurrence point calculation process by the number of passing vehicles (daily average) calculated in the traffic volume calculation process, to calculate the event occurrence probability, as shown in Fig. 12.

[0083] Then, the controller 103 compares the event occurrence probability with a threshold value to determine an abnormality detection point. In the example of Fig. 12, if the threshold value is, for example, 50%, the controller 103 determines that the direction "west" of mesh X and the direction "north" of mesh Z are abnormality detection points, and updates the area information 102d as shown in the table below in Fig. 12.

[0084] Next, a description will be given of information processing according to a processing sequence executed by the red light running detection system 1 according to the embodiment. Fig. 13 is a diagram showing a processing sequence executed by the red light running detection system 1 according to the embodiment.

[0085] The controller 15 of the drive recorder 10 executes driving record while the vehicle is running (step S101). Also, the controller 15 determines whether or not a red light running event has been detected while the vehicle is running (step S102).

[0086] If a red light running event is not detected (step S102, No), the controller 15 repeats the process from step S101. If a red light running event is detected (step S102, Yes), the controller 15 records the red light running event data (step S103) and sets the operation record data and red light running event data for a certain period before and after the detection time to be overwritten. Then, the controller 15 repeats the process from step S101.

[0087] The controller 15 also transmits the recorded operation record data and red light running event data to the center device 100 periodically or in real time (step S104). The controller 103 of the center device 100 accumulates the received operation record data and red light running event data (step S105).

[0088] Next, the controller 103 periodically (for example, once a month) analyzes the accumulated driving record data and red light running event data. Specifically, the controller 103 calculates red light running event frequent occurrence points based on the accumulated red light running event data (step S106).

[0089] The controller 103 also calculates the traffic volume for each mesh based on the accumulated operation record data (step S107).The controller 103 then calculates the occurrence probability for each mesh based on the results of steps S106 and S107 (step S108).The controller 103 then determines whether there is a mesh where the occurrence probability is equal to or greater than a threshold (step S109).

[0090] If there is a mesh with an occurrence probability equal to or greater than the threshold (step S109, Yes), the controller 103 updates the area information 102d by setting the corresponding mesh as a red light ignoring suppression range (step S110). Then, the controller 103 distributes the updated area information 102d to each drive recorder 10 (step S111). If there is no mesh with an occurrence probability equal to or greater than the threshold (step S109, No), the controller 103 repeats the process from step S105.

[0091] The controller 15 of the drive recorder 10 updates the area information 14c with the area information 102d received from the center device 100 (step S112). Then, the controller 15 repeats the process from step S101. After delivering the area information 102d, the controller 103 of the center device 100 repeats the process from step S105.

[0092] As described above, the center device 100 according to the embodiment includes the controller 103. The controller 103 receives, from the drive recorder 10 (which corresponds to an example of an "on-vehicle device"), red light running event data including the latitude and longitude at which the drive recorder 10 detected a red light running, and operation record data (which corresponds to an example of "information including location information"). The controller 103 also calculates the volume of vehicle traffic over a certain period at a location where red light running is frequent, based on the red light running event data and the operation record data. The controller 103 also calculates the probability of red light running at the location from the number of red light running incidents and the traffic volume based on the red light running event data and the operation record data. The controller 103 also determines that a location where the occurrence probability is equal to or greater than a threshold is a red light running anomaly detection location.

[0093] Therefore, the center device 100 according to the embodiment can identify anomaly detection points where the probability of red light running is abnormally high and the possibility of false detection is considered to be extremely high, based on the probability of red light running events. In other words, the center device 100 according to the embodiment can more accurately suppress false detection of red light running.

[0094] In the above-described embodiment, the controller 103 mainly identifies locations where red light ignitions frequently occur in units of vehicle travel directions in a mesh and calculates the occurrence probability for each travel direction. By calculating the occurrence probability according to each vehicle travel direction in this way, it is possible to accurately distinguish between differences in detection results of red light ignitions that appear depending on the travel direction, even within the same mesh, for example.

[0095] In the above-described embodiment, only four directions of travel, i.e., east, west, north, and south, are mentioned, but the number of directions is not limited to four. The number of directions of travel may be set appropriately depending on road conditions, etc.

[0096] On the other hand, it is not necessary to calculate the occurrence probability according to the direction of travel of each mesh. In other words, the controller 103 may identify points where red light ignoring frequently occurs in units of meshes and calculate the occurrence probability for each such mesh.

[0097] Although this results in lower accuracy than when the direction of travel is also taken into account, it can reduce false detections of ignoring traffic signals, such as when there are warning signals, when roads intersect diagonally and multiple traffic lights are located close together, or when multiple traffic lights appear close together at an overpass.

[0098] Furthermore, in the above-described embodiment, an example was given in which each mesh has a fixed size of 50m square, but the size of the mesh may be variable depending on the characteristics of the area where the road is located. For example, the controller 103 may divide the mesh based on map information so that the mesh area is smaller in urban areas than in suburban areas. This is because suburban areas generally have fewer intersections and less vehicular traffic than urban areas. On the other hand, urban areas often have more intersections and shorter distances between traffic lights along a single road than suburban areas. By varying the size of the mesh in this way, it is possible to suppress false detections depending on the characteristics of the area, road conditions, etc.

[0099] Furthermore, in the above-described embodiment, an example was given in which each mesh is square, but the shape of each mesh can be any shape.

[0100] In the above-described embodiment, the controller 103 determines that a high occurrence point where the occurrence probability is equal to or greater than a threshold is a red light running anomaly detection point. However, even if the occurrence probability is less than the threshold, if the occurrence probability is equal to or greater than a certain value, the detection of red light running may be determined to be correct. In this case, the controller 103 transmits the area information 102d to the drive recorder 10 so that the drive recorder 10 issues a notification when a red light running is detected at the relevant red light running anomaly detection point. For example, the controller 103 transmits the area information 102d to the drive recorder 10, with a value indicating "red light running notification required" set as the area type. As a result, if a red light running is detected at a high occurrence point that is not an anomaly detection point, a notification corresponding to the red light running can be appropriately issued.

[0101] Furthermore, instead of the traffic volume calculation process, occurrence probability calculation process, and anomaly detection point determination process in step S3 in the above-described embodiment, a method can be considered in which, instead of the number of vehicles passing through the hotspot, an image of the hotspot is transmitted from the drive recorder 10 to the center device 100, and the hotspot is determined by human visual inspection or advanced image analysis, etc. In this case, instead of transmitting an image of the hotspot from the drive recorder 10 to the center device 100, an existing image provided by a public map viewing service may be used.

[0102] Furthermore, when performing the traffic volume calculation process in step S3, the controller 103 may not use the operation record data of the drive recorder 10 that detected the red light running event, but may register, for example, "operation record acquisition" as area information 102d for the mesh and direction in which the red light running event occurred, and distribute this to each drive recorder 10. In this case, the controller 103 will use the operation record data sent from each drive recorder 10, for example, when each drive recorder 10 passes through that mesh in the corresponding direction.

[0103] As another modification, for example, the controller 103 may receive from the drive recorder 10 an external image of the area in front of the vehicle at a frequent location determined to be an abnormality detection location, and train an image recognition model based on this external image so as not to recognize red light running as occurring. In this case, the controller 103 distributes the trained image recognition model to the drive recorder 10, and causes the drive recorder 10 to detect red light running using the new image recognition model 14b. This improves the image recognition accuracy on the drive recorder 10 side, thereby enabling more accurate suppression of false detections.

[0104] Further advantages and modifications will readily occur to those skilled in the art. Therefore, the invention in its broader aspects is not limited to the specific details and representative embodiments shown and described above. Accordingly, various modifications may be made without departing from the spirit or scope of the general inventive concept as defined by the appended claims and their equivalents. [Explanation of symbols]

[0105] 1. Red light running detection system 5. In-vehicle sensors 10 Drive Recorder 11 Communications Department 12 Sensor section 13 HMI section 14 Storage section 15 Controller 50 HMI section 100 Center Device 101 Communications Department 102 Storage section 103 Controller

Claims

1. receiving information from the in-vehicle device including location information where the in-vehicle device detected the red light ignoring; Calculating the volume of vehicles passing through a location where red light violations frequently occur for a predetermined period based on the information; calculating the probability of red light violations occurring at the location from the number of red light violations occurring based on the information and the traffic volume; a controller that determines that the point where the occurrence probability is equal to or greater than a threshold is a traffic light ignoring anomaly detection point; A center device comprising:

2. The controller transmitting information about the location determined to be the abnormality detection location to the in-vehicle device; and causing the in-vehicle device to suppress a notification of an alarm to a vehicle occupant when a red light violation is detected at the location based on information about the location. The center device according to claim 1 .

3. the information about the location includes a type of the location, The controller causing the in-vehicle device to suppress the notification by setting a value indicating the suppression of the notification for the type of the location; The center device according to claim 2 .

4. the location information includes latitude and longitude; The controller Identifying the location in units of a mesh divided based on the latitude and longitude, and calculating the occurrence probability for each mesh. The center device according to claim 1 .

5. The controller Identifying the point in units of the vehicle's traveling direction in the mesh, and calculating the occurrence probability for each traveling direction. The center device according to claim 4.

6. The controller dividing the meshes based on map information so that the area of ​​the meshes is smaller in urban areas than in suburban areas; The center device according to claim 4.

7. The threshold is a first threshold, The controller If the occurrence probability is less than the first threshold but equal to or greater than a second threshold that is smaller than the first threshold, transmitting information about the location to the in-vehicle device so as to cause the in-vehicle device to perform the notification when a red light ignoring is detected at the location. The center device according to claim 2 .

8. The controller receiving from the in-vehicle device an outside image of the area in front of the vehicle at the point determined to be the abnormality detection point; An image recognition model is trained based on the image outside the vehicle so as not to recognize a red light running, Distributing the trained image recognition model to the in-vehicle device. The center device according to any one of claims 1 to 7.

9. The system includes an in-vehicle device and a center device, The in-vehicle device Detects red light violations based on external video footage of the area in front of the vehicle and vehicle speed. transmitting information including location information of the detected red light violation to the center device; The center device receiving the information from the in-vehicle device; Calculating the volume of vehicles passing through a location where red light violations frequently occur for a predetermined period based on the information; calculating the probability of red light violations occurring at the location from the number of red light violations occurring based on the information and the traffic volume; The point where the occurrence probability is equal to or greater than a threshold is determined to be a traffic light ignoring anomaly detection point. Red light running detection system.

10. A method for detecting red light violations executed by a controller, comprising: receiving information from the in-vehicle device, the information including location information where the in-vehicle device detected the red light ignoring; Calculating the volume of vehicles passing through a location where red light violations frequently occur for a predetermined period of time based on the information; calculating a probability of red light violations occurring at the location from the number of red light violations occurring based on the information and the traffic volume; determining that the point where the occurrence probability is equal to or greater than a threshold is a signal violation anomaly detection point; A method for detecting red light violation.

Images