Traffic safety support system and its learning method

Abstract

To provide a traffic safety support system capable of improving safety, convenience and smoothness of traffic for a plurality of traffic participants present in a target traffic area.SOLUTION: A traffic safety support system includes: a target traffic area recognition unit for acquiring recognition information on traffic participants or the like in a target traffic area; a prediction unit 62 for predicting a risk in the target traffic area based on the recognition information; and a cooperation support information notification unit 65 for transmitting cooperation support information to support targets. The prediction unit 62 includes: an area risk prediction unit 620 for extracting a high-risk area from a plurality of local areas obtained by subdividing the target traffic area based on information obtained by performing statistical processing on the recognition information; and a traffic participant risk prediction unit 625 for predicting future risk of traffic participants in the high-risk area based on information related to the high-risk area in the recognition information.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to a traffic safety support system and a learning method thereof, and more particularly to a traffic safety support system that supports the safe movement of traffic participants as people or mobile bodies, and a learning method thereof. [Background technology]

[0002] In public transportation, various traffic participants, such as moving objects such as four-wheeled automobiles, motorcycles, and bicycles, as well as pedestrians, travel at different speeds based on their own will. As a technology for improving the safety and convenience of traffic participants in such public transportation, for example, Patent Document 1 discloses a driving assistance device that assists vehicle drivers in safe driving.

[0003] The driving assistance device disclosed in Patent Document 1 includes a risk prediction unit that predicts the risk level of the vehicle based on information about the vehicle's driving state and the surrounding environment, and a warning control unit that warns the driver by voice, text display, etc., based on the evaluation result of the predicted risk level. When some kind of risk is predicted, the driving assistance device disclosed in Patent Document 1 can urge the driver to perform driving operations to avoid the predicted risk, thereby supporting safe driving by the driver. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent Publication No. 2021-136001 Summary of the Invention [Problem to be solved by the invention]

[0005] However, the invention disclosed in Patent Document 1 predicts the level of danger based on information about the surrounding environment obtained by onboard sensors such as cameras and radars installed in the vehicle, and therefore cannot grasp potential risks that exist outside the detection range of the onboard sensors.

[0006] Therefore, in order to provide appropriate support for such potential risks, it is conceivable that, for example, information regarding traffic participants in a specified target traffic area could be collected on a server that is communicatively connected to each traffic participant, and the server could then obtain an overview of the flow of traffic participants in this target traffic area.

[0007] However, if the information on the huge number of traffic participants in the target traffic area is aggregated on a server in this way, the processing load on the server will increase, and there is a risk that it will not be possible to provide appropriate assistance to each traffic participant in the target traffic area in real time.

[0008] An object of the present invention is to provide a traffic safety support system that can improve the safety, convenience, and smoothness of traffic by multiple traffic participants in a target traffic area. [Means for solving the problem]

[0009] (1) The traffic safety support system of the present invention comprises a recognition means for recognizing recognition objects including traffic participants as people or moving objects in a target traffic area and the traffic environment of each traffic participant, and for acquiring recognition information regarding these recognition objects; a prediction means for predicting risks in the target traffic area based on the recognition information; and a transmission means for transmitting support information generated based on the recognition information and the prediction results by the prediction means to a support target selected from a plurality of traffic participants in the target traffic area, wherein the prediction means comprises an area risk prediction means for extracting at least one of a plurality of local areas obtained by subdividing the target traffic area as a high-risk area based on information obtained by performing statistical processing on the recognition information; and a traffic participant risk prediction means for predicting future risks to traffic participants in the high-risk area based on information related to the high-risk area in the recognition information.

[0010] (2) In this case, it is preferable that the traffic participant risk prediction means does not perform the prediction process for any local area among the plurality of local areas that has not been extracted as a high-risk area by the area risk prediction means.

[0011] (3) In this case, it is preferable that the area risk prediction means estimates the degree of risk for each of the local areas, and the transmission means transmits first support information generated based on the prediction results by the traffic participant risk prediction means to support targets among the multiple support targets that are located within the high-risk area, and transmits second support information generated based on the estimation results by the area risk prediction means to support targets that are located in a low-risk area outside the high-risk area.

[0012] (4) A learning method for a traffic safety support system according to the present invention is a learning method for a traffic safety support system described in any one of (1) to (3), characterized in that the area risk prediction means extracts the high-risk areas by utilizing a macro risk estimation model that outputs a degree of risk for each of the multiple local areas when information obtained by applying statistical processing to the recognition information is input, and the traffic participant risk prediction means predicts the future risk of traffic participants in the high-risk areas by utilizing a micro risk estimation model that outputs the future risk of traffic participants in a specified local area when information related to a specified local area among the recognition information is input, and the learning method comprises the steps of: preparing learning data by using input data for the macro risk estimation model generated based on the recognition information and the output of the micro risk estimation model when the recognition information is input to the micro risk estimation model; and learning the macro risk estimation model using the learning data.

[0013] (5) A learning method for a traffic safety support system according to the present invention is a learning method for a traffic safety support system described in any one of (1) to (3), wherein the area risk prediction means extracts the high-risk areas by utilizing a macro risk estimation model that outputs a degree of risk for each of the multiple local areas when information obtained by applying statistical processing to the recognition information is input, and the traffic participant risk prediction means predicts the future risk of traffic participants in the high-risk areas by utilizing a micro risk estimation model that outputs the future risk of traffic participants in a specified local area when information related to a specified local area among the recognition information is input, and the learning method comprises the steps of: preparing learning data by using input data for the macro risk estimation model generated based on first recognition information acquired in a specified first period and correct answer data for the output of the micro risk estimation model generated based on second recognition information acquired in a second period immediately after the first period; and learning an overall model that combines the macro risk estimation model and the micro risk estimation model using the learning data. [Effects of the Invention]

[0014] (1) The traffic safety support system of the present invention includes a recognition means for recognizing recognition targets, including traffic participants (including people and moving objects) in a target traffic area and the traffic environment of each traffic participant, and for acquiring recognition information related to these recognition targets, a prediction means for predicting risks in the target traffic area based on the recognition information, and a transmission means for transmitting support information generated based on the recognition information and the prediction results of the prediction means to a designated support target from among multiple traffic participants in the target traffic area. The prediction means further includes an area risk prediction means for extracting at least one of multiple local areas obtained by subdividing the target traffic area as a high-risk area, and a traffic participant risk prediction means for predicting future risks to traffic participants in the high-risk area. When extracting high-risk areas from the multiple local areas, the area risk prediction means uses information obtained by performing statistical processing on the recognition information, thereby enabling high-risk areas to be extracted with less load than when a huge amount of recognition information related to the recognition targets in the target traffic area is used as is. Furthermore, when predicting the future risk of a traffic participant in a high-risk area, the traffic participant risk prediction means uses information related to the high-risk area from the recognition information related to the recognition objects in the entire target traffic area, thereby making it possible to predict the future risk of the traffic participant with less load than when using a huge amount of recognition information related to the recognition objects in the target traffic area as is. Therefore, according to the invention, appropriate support information generated based on the prediction results can be provided to traffic participants in the high-risk area in real time, thereby improving the safety, convenience, and smoothness of traffic in the target traffic area.

[0015] (2) In the present invention, the traffic participant risk prediction means does not predict future risks of traffic participants in other local areas that are not extracted as high-risk areas by the area risk prediction means among the multiple local areas. Therefore, according to the present invention, the traffic participant risk prediction means can reduce the computational load compared to when performing prediction processing for all local areas. Furthermore, according to the present invention, the computational load can be reduced by limiting the number of local areas subjected to prediction processing, thereby improving the accuracy of prediction of risks of traffic participants in high-risk areas. Therefore, according to the present invention, appropriate assistance information generated based on highly accurate prediction results can be provided to traffic participants in high-risk areas in real time, thereby further improving the safety, convenience, and smoothness of traffic in the target traffic area.

[0016] (3) In the present invention, the area risk prediction means estimates the risk level for each local area and extracts high-risk areas from the multiple local areas based on the risk level estimation results. The transmission means transmits first support information generated based on the relatively detailed prediction results by the traffic participant risk prediction means to support targets located in high-risk areas among the multiple support targets in the entire target traffic area. This improves traffic safety, convenience, and smoothness of traffic for traffic participants in high-risk areas. The transmission means transmits second support information generated based on the estimation results for each local area by the area risk prediction means to support targets located in low-risk areas outside the high-risk areas among the multiple support targets in the entire target traffic area. This also improves traffic safety, convenience, and smoothness of traffic for traffic participants in low-risk areas. In this way, the present invention changes the support information according to the risk level for each local area, thereby improving traffic safety, convenience, and smoothness of traffic for traffic participants in the entire target traffic area.

[0017] (4) The learning method for a traffic safety support system according to the present invention prepares learning data by utilizing input data for a macro risk estimation model generated based on recognition information and the output of the micro risk estimation model when this recognition information is input to the micro risk estimation model, and then uses this learning data to train the macro risk estimation model. In general, model learning requires the preparation of correct answer data for evaluating the accuracy of the model output. In contrast, according to the present invention, the output of the micro risk estimation model can be used as learning data for training the macro risk estimation model, making it possible to build a highly accurate macro risk estimation model using a relatively simple method. Therefore, according to the present invention, it is possible to improve the accuracy of the macro risk estimation model while operating a service that provides support information to each traffic participant.

[0018] (5) A learning method for a traffic safety support system according to the present invention prepares learning data by using input data for a macro risk estimation model generated based on first recognition information acquired in a first period and correct answer data for the output of a micro risk estimation model generated based on second recognition information acquired in a second period immediately after the first period, and further uses this learning data to learn an overall model that combines the macro risk estimation model and the micro risk estimation model. Therefore, according to the present invention, the second recognition information acquired in the second period immediately after the first period can be used as data for evaluating the accuracy of the output of the overall model when the first recognition information is input, thereby improving the accuracy of the overall model that combines the macro risk estimation model and the micro risk estimation model. Therefore, according to the present invention, the accuracy of the overall model can be improved while operating a service that provides support information to each traffic participant. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a diagram showing the configuration of a traffic safety support system according to an embodiment of the present invention and a portion of a target traffic area that is a support target of this traffic safety support system. [Figure 2] 1 is a block diagram showing the configuration of a collaboration support device and a plurality of area terminals communicably connected to the collaboration support device. [Figure 3A] 1 is a block diagram showing the configuration of a notification device mounted on a four-wheeled vehicle. [Figure 3B] 1 is a block diagram showing the configuration of a notification device mounted on a motorcycle. [Figure 3C] 10 is a block diagram showing the configuration of a notification device mounted on a portable information processing terminal carried by a pedestrian. FIG. [Figure 4] FIG. 10 is a functional block diagram showing a specific configuration of a prediction unit. [Figure 5] FIG. 10 is a diagram illustrating the concept of risk notification optimization processing in the risk notification setting unit. DETAILED DESCRIPTION OF THE INVENTION

[0020] Hereinafter, a traffic safety support system according to an embodiment of the present invention will be described with reference to the drawings.

[0021] FIG. 1 is a diagram showing a schematic configuration of a traffic safety support system 1 according to this embodiment and a part of a target traffic area 9 in which traffic participants to be supported by this traffic safety support system 1 exist.

[0022] The traffic safety support system 1 recognizes pedestrians 4, who are people moving in the target traffic area 9, and moving bodies such as four-wheeled vehicles 2 and motorcycles 3, as individual traffic participants, and notifies each traffic participant of the support information generated through this recognition, thereby supporting the safe and smooth traffic of each traffic participant in the target traffic area 9 by encouraging communication between each traffic participant moving based on their own will (specifically, for example, mutual recognition between each traffic participant) and recognition of the surrounding traffic environment.

[0023] FIG. 1 illustrates a case where a target traffic area 9 is the vicinity of an intersection 52 in an urban area, which includes a roadway 51, an intersection 52, a sidewalk 53, and traffic lights 54 as traffic infrastructure facilities. FIG. 1 shows a case where a total of seven four-wheeled vehicles 2 and a total of two motorcycles 3 are moving on the roadway 51 and within the intersection 52, and a total of three pairs of pedestrians 4 are moving on the sidewalk 53 and within the intersection 52. FIG. 1 also shows a case where a total of three infrastructure cameras 56 are installed.

[0024] The traffic safety support system 1 comprises a group of on-board devices 20 (including on-board devices mounted on the four-wheeled vehicles 2 as well as portable information processing terminals held or worn by the drivers of the four-wheeled vehicles 2) that travel with each four-wheeled vehicle 2, a group of on-board devices 30 (including on-board devices mounted on the motorcycles 3 as well as portable information processing terminals held or worn by the drivers of the motorcycles 3) that travel with each motorcycle 3, a portable information processing terminal 40 held or worn by each pedestrian 4, a plurality of infrastructure cameras 56 installed in the target traffic area 9, a signal control device 55 that controls traffic lights 54, and a collaboration support device 6 that is communicatively connected to a plurality of terminals (hereinafter simply referred to as "area terminals") present in the target traffic area 9, such as the on-board device groups 20, 30, the portable information processing terminal 40, the infrastructure cameras 56, and the signal control device 55.

[0025] The cooperative support device 6 is configured by one or more computers communicably connected to the above-mentioned multiple area terminals via a base station 57. More specifically, the cooperative support device 6 is configured by a server connected to the multiple area terminals via the base station 57, a network core, and the Internet, an edge server connected to the multiple area terminals via the base station 57 and an MEC (Multi-access Edge Computing) core, etc.

[0026] FIG. 2 is a block diagram showing the configuration of a collaboration support device 6 and a plurality of area terminals connected to the collaboration support device 6 so as to be able to communicate with each other.

[0027] The group of on-board devices 20 mounted on four-wheeled vehicles 2 in the target traffic area 9 includes, for example, an on-board driving assistance device 21 that assists the driver in driving, a notification device 22 that notifies the driver of various information, a driver subject state sensor 23 that detects the state of the driver while driving, an on-board communication device 24 that performs wireless communication between the vehicle and the collaborative assistance device 6 or other vehicles in the vicinity of the vehicle, and a portable information processing terminal 25 owned or worn by the driver.

[0028] The in-vehicle driving assistance device 21 includes an external sensor unit, a vehicle status sensor, a navigation device, a driving assistance ECU, etc. The external sensor unit includes an exterior camera unit that captures images of the surroundings of the vehicle, multiple in-vehicle external sensors mounted on the vehicle, such as a radar unit and a Light Detection and Ranging (LIDAR) unit that detect objects outside the vehicle by using electromagnetic waves, and an external recognition device that acquires information about the surroundings of the vehicle by performing sensor fusion processing on the detection results of these in-vehicle external sensors. The vehicle status sensor includes sensors that acquire information about the driving status of the vehicle, such as a vehicle speed sensor, an acceleration sensor, a steering angle sensor, a yaw rate sensor, a position sensor, and a direction sensor. The navigation device includes, for example, a Global Navigation Satellite System (GNSS) receiver that identifies the current position of the vehicle based on signals received from GNSS satellites, a storage device that stores map information, etc.

[0029] The driving assistance ECU executes driving assistance controls such as lane departure prevention control, lane change control, leading vehicle following control, false start prevention control, collision mitigation brake control, and collision avoidance control based on information acquired by the external sensor unit, the vehicle state sensor, the navigation device, etc. The driving assistance ECU also generates driving assistance information for assisting the driver in safe driving based on the information acquired by the external sensor unit, the vehicle state sensor, the navigation device, etc. and transmits the information to the notification device 22.

[0030] Here, the driving assistance ECU initiates collision mitigation brake control, which automatically operates the braking device of the host vehicle to mitigate damage caused by contact between the host vehicle and the other moving object, on the condition that a moving object that may come into contact with the host vehicle is present within a predetermined collision mitigation brake operation range centered on the host vehicle. The driving assistance ECU also initiates collision avoidance control, which automatically operates the steering device of the host vehicle to avoid contact between the host vehicle and the other moving object, on the condition that a moving object that may come into contact with the host vehicle is present within a predetermined collision avoidance steering operation range centered on the host vehicle. Hereinafter, the collision mitigation brake operation range and the collision avoidance steering operation range are collectively referred to as the "ADAS operation range."

[0031] The driver's subject state sensor 23 is composed of various devices that acquire time-series data of information correlated with the driver's driving ability while driving. The driver's subject state sensor 23 is composed of, for example, an in-vehicle camera that detects the direction of the driver's line of sight and whether the driver's eyes are open while driving, a seat belt sensor attached to the seat belt worn by the driver that detects the driver's pulse and whether the driver is breathing, a steering sensor attached to the steering wheel held by the driver that detects the driver's skin potential, an in-vehicle microphone that detects whether the driver is talking to a passenger, etc.

[0032] The in-vehicle communication device 24 has a function of transmitting information acquired by the driving assistance ECU (including information acquired by the external sensor unit, the vehicle status sensor, and the navigation device, etc., and control information related to driving assistance control currently being performed), and information related to the driver acquired by the driver's status sensor 23, to the collaborative assistance device 6, and a function of receiving collaborative assistance information transmitted from the collaborative assistance device 6 and transmitting the received collaborative assistance information to the notification device 22.

[0033] The notification device 22 is composed of various devices that notify the driver of various information through the driver's hearing, vision, touch, etc. by operating a man-machine interface (hereinafter sometimes abbreviated as "HMI (Human Machine Interface)") in a manner determined based on the driving assistance information transmitted from the in-vehicle driving assistance device 21 and the collaborative assistance information transmitted from the collaborative assistance device 6.

[0034] 3A is a block diagram showing the configuration of notification device 22 mounted on a four-wheeled vehicle. Note that Fig. 3A illustrates only the blocks of notification device 22 that are particularly related to control based on the collaboration support information transmitted from collaboration support device 6.

[0035] The notification device 22 includes an HMI 220 that operates in a manner that can be recognized by the driver, and an HMI control device 225 that operates the HMI 220 based on the collaboration support information transmitted from the collaboration support device 6.

[0036] The HMI 220 includes an acoustic device 221 that operates in a manner that the driver can hear, a head-up display 222 that operates in a manner that the driver can see, and a seat belt control device 223 and a seat vibration device 224 that operate in a manner that the driver can feel.

[0037] The sound device 221 includes a headrest speaker 221a that is provided on the headrest of the driver's seat where the driver sits and is capable of emitting directional binaural sound, and a main speaker 221b that is provided near the driver's seat and the passenger seat. The headrest speaker 221a and the main speaker 221b emit sounds in response to commands from the HMI control device 225. The head-up display 222 displays an image in response to commands from the HMI control device 225 within the field of view of the driver while driving (for example, on the windshield). The seat belt control device 223 changes the tension of the seat belt worn by the driver in response to commands from the HMI control device 225. The seat vibration device 224 vibrates the seat where the driver sits with an amplitude and / or frequency in response to commands from the HMI control device 225.

[0038] The HMI control device 225 includes a soundness control device 226 that performs a soundness notification by operating the HMI 220 in a predetermined manner to sounden the driver's driving ability (particularly, cognitive ability), a risk notification control device 227 that performs a risk notification by operating the HMI 220 in a predetermined manner to make the driver aware of the presence of an imminent risk, and a risk area notification control device 228 that performs a risk area notification by operating the HMI 220 in a predetermined manner to make the driver aware of information about risk on an area-by-area basis. As will be described later, the collaboration support information transmitted from the collaboration support device 6 to the four-wheeled vehicle 2 includes information on a soundness notification setting value for setting on / off of the soundness notification by the soundness control device 226, information on a risk notification setting value for setting on / off of the risk notification by the risk notification control device 227 and the type of notification mode described below, information about an imminent risk for the driver (hereinafter also referred to as "risk information"), risk area information used for the risk area notification by the risk area notification control device 228, etc.

[0039] The health notification setting value input to the health control device 226 is set to either “0” which turns off the health notification by the health control device 226, or “1” which turns on the health notification by the health control device 226.

[0040] When the health notification setting value is "0", the health control device 226 sets the health notification to off. That is, when the health notification setting value is "0", the health control device 226 does not operate the HMI 220. Note that this does not prevent the risk notification control device 227 from operating the HMI 220.

[0041] When the health notification setting value is "1", the health control device 226 sets the health notification to ON. More specifically, the health control device 226 improves the driving ability of the driver by playing music that attracts the driver's interest through, for example, the headrest speaker 221a or the main speaker 221b. At this time, the BPM (Beats Per Minute) of the music may be changed or the bass may be emphasized in order to increase the driver's level of awareness.

[0042] In this way, the health improvement control device 226 activates the HMI 220 to improve the driver's driving ability. Therefore, when the risk notification by the risk notification control device 227 described below is set to on (i.e., when the risk notification setting value is "1" or "2"), the health improvement notification may be turned off so as not to bother the driver. In addition, in this embodiment, the health improvement control device 226 activates the headrest speaker 221a and the main speaker 221b to improve the driver's driving ability mainly through the driver's hearing, but the present invention is not limited to this. The health improvement control device 226 may also activate, for example, the seat belt control device 223 or the seat vibration device 224.

[0043] The risk notification control device 227 can provide risk notification in multiple notification modes, each differing in at least one of the target device and operation mode of the HMI 220. More specifically, the risk notification control device 227 can provide risk notification in at least one of the following notification modes: a warning mode intended to make the driver aware of a potential risk; an analog notification mode intended to make the driver aware of the presence and / or severity of an actual risk; and a predictive assistance notification mode intended to notify the driver of useful information for avoiding a predicted risk. Therefore, the risk notification setting value input to the risk notification control device 227 is set to one of the following values: “0” to turn off risk notification; “1” to turn on risk notification in the warning mode; “2” to turn on risk notification in the analog notification mode; “3” to turn on risk notification in the predictive assistance notification mode; “4” to turn on risk notification in both the warning mode and the predictive assistance notification mode; and “5” to turn on risk notification in both the analog notification mode and the predictive assistance notification mode.

[0044] The risk notification control device 227 sets the risk notification to off when the risk notification setting value is "0." That is, the risk notification control device 227 does not operate the HMI 220 when the risk notification setting value is "0." Note that this does not prevent the health control device 226 from operating the HMI 220.

[0045] When the risk notification setting value is "1", the risk notification control device 227 sets the notification mode to the presence notification mode and turns on risk notification under the set notification mode.

[0046] If the risk notification setting value is "2", the risk notification control device 227 sets the notification mode to the analog notification mode and turns on risk notification under the set notification mode.

[0047] When the risk notification setting value is "3", the risk notification control device 227 sets the notification mode to the predicted assistance notification mode and turns on risk notification under the set notification mode.

[0048] When the risk notification setting value is "4", the risk notification control device 227 sets the notification mode to the presence notification mode and the prediction support notification mode, and turns on risk notification under these set notification modes.

[0049] Furthermore, when the risk notification setting value is "5", the risk notification control device 227 sets the notification mode to the analog notification mode and the predicted assistance notification mode, and turns on risk notification under these set notification modes.

[0050] Here, when the notification mode is set to the predictive support notification mode, the risk notification control device 227 generates risk avoidance support information that is useful for the driver to avoid an imminent risk based on the risk information transmitted from the collaboration support device 6, and activates the audio device 221 and head-up display 222 of the HMI 220 in a manner that allows the driver to perceive this risk avoidance support information audibly or visually. Here, the risk avoidance support information includes information on the location of traffic participants (hereinafter also referred to as "risk targets") that may come into contact with the vehicle, information on points (hereinafter also referred to as "risk occurrence points") where the vehicle may come into contact with the risk targets, and information that alerts the driver to the risk targets.

[0051] More specifically, when a motorcycle driven by a rider in an unhealthy state is present ahead of a four-wheeled vehicle driven by a driver, the risk notification control device 227 issues a message such as "Watch out for the dangerous right turn of the motorcycle" as risk avoidance support information for avoiding contact with the motorcycle by sounding it through the audio device 221 or displaying it on the head-up display 222. In addition, at this time, the risk notification control device 227 may also display an image of an arrow pointing to the current position or predicted position of the motorcycle on the head-up display 222 as risk avoidance support information for avoiding contact with the motorcycle.

[0052] Furthermore, when the notification mode is set to the presence notification mode, the risk notification control device 227 operates the HMI 220 in a manner that does not bother the driver, thereby allowing the driver to naturally recognize the presence of a risk target extracted from the risk information transmitted from the collaboration support device 6. In this manner, in the presence notification mode, in order to naturally allow the driver to recognize the presence of a risk target without feeling bothered, it is preferable that the risk notification control device 227 activates the headrest speaker 221a, which particularly appeals to the driver's hearing, among the multiple devices included in the HMI 220. More specifically, when the notification mode is set to the presence notification mode, the risk notification control device 227 causes the headrest speaker 221a to emit a familiar sound effect at a low volume using directional binaural sound directed toward the location of the risk target or the location of the risk occurrence point, thereby naturally directing the driver's gaze toward the location of the risk target or the risk occurrence point.

[0053] Furthermore, when the notification mode is set to the analog notification mode, the risk notification control device 227 operates the HMI 220 in a manner different from that of the presence notification mode described above, thereby making the driver more aware of the presence of a risk object extracted from the risk information transmitted from the collaboration support device 6 and the degree of risk associated with this risk object. In this manner, in the analog notification mode, in order to make the driver more aware of the presence of a risk object, the risk notification control device 227 operates the HMI 220 in a manner with a higher notification intensity than that specified in the presence notification mode. Here, notification intensity refers to the strength of the notification that attracts the driver's interest and attention. More specifically, when the notification mode is set to the analog notification mode, the risk notification control device 227 causes the headrest speaker 221a and the main speaker 221b to emit a buzzer sound or pulse sound at a volume higher than the sound effects emitted in the presence notification mode. These buzzer sounds and pulse sounds are louder and less familiar to the driver than the sound effects emitted in the presence notification mode, and therefore have a higher notification intensity than the sound effects emitted in the presence notification mode.

[0054] In this embodiment, the risk notification control device 227 activates the sound device 221 when the notification mode is set to the analog notification mode, but the present invention is not limited to this. When the notification mode is set to the analog notification mode, the risk notification control device 227 may activate the seat belt control device 223 to change the tension of the seat belt or activate the seat vibration device 224 to vibrate the seat instead of activating the sound device 221. In this way, the seat belt control device 223 and the seat vibration device 224 operate in a manner that appeals to the driver's tactile sense, and therefore have a higher notification intensity than the sound effect emitted in the presence notification mode. Furthermore, when the notification mode is set to the analog notification mode, the risk notification control device 227 may activate the sound device 221, the seat belt control device 223, and the seat vibration device 224 in combination.

[0055] Furthermore, in the analog notification mode as described above, in order to strongly notify the driver of the presence of a risk object as well as the degree of risk associated with this risk object, it is preferable that the risk notification control device 227 varies the notification intensity according to the degree of risk associated with the risk object (e.g., the predicted time to collision with the risk object) extracted from the risk information transmitted from the collaboration support device 6. Specifically, the risk notification control device 227 may increase the volume of the buzzer sound, increase the volume of the pulse sound, or shorten the interval between the pulse sounds, thereby increasing the notification intensity, as the risk degree increases (i.e., as the predicted time to collision decreases). When the seat belt control device 223 is activated as described above, the risk notification control device 227 may increase the tension of the seat belt and increase the notification intensity as the risk degree increases. Furthermore, when the seat vibration device 224 is activated as described above, the risk notification control device 227 may increase the amplitude of the seat vibration and increase the notification intensity as the risk degree increases.

[0056] Furthermore, when the risk notification control device 227 changes the notification intensity according to the degree of risk in this manner, it is preferable to operate the HMI 220 so that the notification intensity is maximized at the point when the above-mentioned driving assistance ECU begins to execute collision mitigation braking control or collision avoidance steering control, in other words, at the point when the risk object enters the ADAS operating range of the vehicle.

[0057] The risk area notification control device 228 operates the HMI 220 based on the risk area information successively transmitted from the collaboration support device 6, thereby performing risk area notification to notify the driver of information related to the current risk area. As will be described later, the risk area information includes information related to the current risk level for each local area obtained by subdividing the target traffic area. Therefore, when the risk area notification control device 228 determines based on the risk area information that the vehicle is traveling in a local area with a high risk level, it issues a message such as "You are traveling in a high-risk area. Please be careful of your surroundings" through the main speaker 221b or displays it on the head-up display 222.

[0058] Returning to FIG. 2 , the mobile information processing terminal 25 is configured, for example, by a wearable terminal worn by the driver of the four-wheeled vehicle 2, a smartphone carried by the driver, or the like. The wearable terminal has a function to measure the driver's biological information, such as heart rate, blood pressure, and blood oxygen saturation, and transmit the measurement data of this biological information to the collaboration support device 6, and a function to receive collaboration support information transmitted from the collaboration support device 6 and notify the driver of a message corresponding to the collaboration support information by means of an image, voice, a warning sound, a melody, a vibration, or the like. The smartphone also has a function to transmit information related to the driver, such as the driver's location information, movement acceleration, and schedule information, to the collaboration support device 6, and a function to receive collaboration support information transmitted from the collaboration support device 6 and notify the driver of a message corresponding to the collaboration support information by means of an image, voice, a warning sound, a melody, a vibration, or the like.

[0059] The group of on-board devices 30 mounted on motorcycles 3 in the target traffic area 9 includes, for example, an on-board driving assistance device 31 that assists the rider in driving, a notification device 32 that notifies the rider of various information, a rider state sensor 33 that detects the state of the rider while driving, an on-board communication device 34 that wirelessly communicates between the vehicle and the cooperative assistance device 6 or other vehicles in the vicinity of the vehicle, and a portable information processing terminal 35 that is owned or worn by the rider.

[0060] The in-vehicle driving assistance device 31 includes an external sensor unit, a vehicle status sensor, a navigation device, and a driving assistance ECU. The external sensor unit includes an exterior camera unit that captures images of the vehicle's surroundings, multiple on-board external sensors such as a radar unit or a lidar unit that detects objects outside the vehicle using electromagnetic waves, and an external recognition device that acquires information about the vehicle's surroundings by performing sensor fusion processing on the detection results from these on-board external sensors. The vehicle status sensor includes sensors that acquire information about the vehicle's driving status, such as a vehicle speed sensor and a 5-axis or 6-axis inertial measurement unit. The navigation device includes, for example, a GNSS receiver that identifies the vehicle's current location based on signals received from GNSS satellites, a storage device that stores map information, and the like.

[0061] The driving assistance ECU executes driving assistance controls such as lane keeping control, lane departure prevention control, lane change control, leading vehicle following control, false start prevention control, and collision mitigation brake control based on information acquired by the external sensor unit, the vehicle status sensor, the navigation device, etc. The driving assistance ECU also generates driving assistance information to assist the rider in safe driving based on the information acquired by the external sensor unit, the vehicle status sensor, the navigation device, etc. and transmits the information to the notification device 32.

[0062] Here, the driving assistance ECU initiates collision mitigation brake control, which automatically operates the vehicle's braking device to reduce damage caused by contact between the vehicle and the other moving object, provided that there is a moving object that may come into contact with the vehicle within a predetermined collision mitigation brake operation range (hereinafter referred to as the "ADAS operation range" together with the term defined for four-wheeled vehicle 2) centered on the vehicle.

[0063] The rider condition sensor 33 is made up of various devices that acquire information correlated with the driving ability of the rider while driving. The rider condition sensor 33 is made up of, for example, a seat sensor that is provided on the seat on which the rider sits and detects the pulse, presence or absence of breathing, etc. of the rider, and a helmet sensor that is provided on the helmet worn by the rider and detects the pulse, presence or absence of breathing, skin potential, etc. of the rider.

[0064] The in-vehicle communication device 34 has a function of transmitting information acquired by the driving assistance ECU (including information acquired by the external sensor unit, the vehicle status sensor, and the navigation device, etc., and control information related to driving assistance control currently being performed), and information related to the rider acquired by the rider status sensor 33, to the collaborative assistance device 6, and a function of receiving collaborative assistance information transmitted from the collaborative assistance device 6 and transmitting the received collaborative assistance information to the notification device 32.

[0065] The notification device 32 is composed of various devices that notify the rider of various information through the rider's hearing, vision, touch, etc. by operating the HMI in a predetermined manner based on the driving assistance information transmitted from the in-vehicle driving assistance device 21 and the collaborative assistance information transmitted from the collaborative assistance device 6.

[0066] Fig. 3B is a block diagram showing the configuration of notification device 32 mounted on a motorcycle. Note that Fig. 3B illustrates only the blocks of notification device 32 that are particularly related to control based on the collaboration support information transmitted from collaboration support device 6.

[0067] The notification device 32 includes an HMI 320 that operates in a manner that can be recognized by the rider, and an HMI control device 325 that operates the HMI 320 based on the collaboration support information transmitted from the collaboration support device 6.

[0068] The HMI 320 includes a head-mounted speaker 321 that operates in a manner that the rider can hear, and a head-up display 322 that operates in a manner that the rider can see.

[0069] The head-mounted speaker 321 is mounted on a helmet worn by the rider and is capable of emitting directional binaural sound. The head-mounted speaker 321 emits sound in response to commands from the HMI control device 325. The head-up display 322 displays an image in response to commands from the HMI control device 325 within the field of view of the rider while driving (for example, on the helmet shield).

[0070] The HMI control device 325 includes a soundness control device 326 that issues a soundness notification by operating the HMI 320 in a predetermined manner to soundness the rider's driving ability (particularly, cognitive ability), a risk notification control device 327 that issues a risk notification by operating the HMI 320 in a predetermined manner to make the rider aware of the presence of an imminent risk, and a risk area notification control device 328 that issues a risk area notification by operating the HMI 320 in a predetermined manner to make the rider aware of information about risk on an area-by-area basis. As will be described later, the collaboration assistance information transmitted from the collaboration assistance device 6 to the motorcycle 3 includes information on a soundness notification setting value for setting on / off of the soundness notification by the soundness control device 326, information on a risk notification setting value for setting on / off of the risk notification by the risk notification control device 327 and the type of notification mode, risk information about an imminent risk for the rider, risk area information used for the risk area notification by the risk area notification control device 328, etc.

[0071] The health notification setting value input to the health control device 326 is set to either “0” which turns off the health notification by the health control device 326, or “1” which turns on the health notification by the health control device 326.

[0072] When the health notification setting value is "0", the health control device 326 sets the health notification to off. In other words, when the health notification setting value is "0", the health control device 326 does not operate the HMI 320. Note that this does not prevent the risk notification control device 327 from operating the HMI 320.

[0073] When the fitness notification setting value is "1", the fitness control device 326 sets the fitness notification to ON. More specifically, the fitness control device 326 improves the rider's driving ability by playing music that attracts the rider's interest, for example, through the head-mounted speaker 321. At this time, the BPM of the music may be changed or the bass may be emphasized in order to increase the rider's level of awareness.

[0074] Since the improvement control device 326 operates the HMI 320 to improve the rider's driving ability in this way, if the risk notification by the risk notification control device 327 described below is set to on (i.e., if the risk notification setting value is "1" or "2"), the improvement notification may be turned off so that the rider does not feel bothered.

[0075] The risk notification control device 327 can provide risk notification in multiple notification modes, each differing in at least one of the target device and operation mode of the HMI 320. More specifically, the risk notification control device 327 can provide risk notification in at least one of the following notification modes: a warning sign mode intended to make the rider aware of the existence of a potential risk; an analog notification mode intended to make the rider aware of the existence of an actual risk and / or the severity of that risk; and a predictive assistance notification mode intended to notify the rider of useful information for avoiding a predicted risk. For this reason, the risk notification setting value input to the risk notification control device 327 is set to one of the following values: “0” to turn off the risk notification; “1” to turn on the risk notification in the warning sign mode; “2” to turn on the risk notification in the analog notification mode; “3” to turn on the risk notification in the predictive assistance notification mode; “4” to turn on the risk notification in both the warning sign mode and the predictive assistance notification mode; and “5” to turn on the risk notification in the analog notification mode and the predictive assistance notification mode.

[0076] The risk notification controller 327 sets the risk notification to off when the risk notification setting value is "0." That is, the risk notification controller 327 does not operate the HMI 320 when the risk notification setting value is "0." Note that this does not prevent the health control device 326 from operating the HMI 320.

[0077] When the risk notification setting value is "1", the risk notification control device 327 sets the notification mode to the presence notification mode and turns on risk notification under the set notification mode.

[0078] If the risk notification setting value is "2", the risk notification control device 327 sets the notification mode to the analog notification mode and turns on risk notification under the set notification mode.

[0079] If the risk notification setting value is "3", the risk notification control device 327 sets the notification mode to the predicted assistance notification mode and turns on risk notification under the set notification mode.

[0080] When the risk notification setting value is "4", the risk notification control device 327 sets the notification mode to the presence notification mode and the prediction support notification mode, and turns on risk notification under these set notification modes.

[0081] In addition, when the risk notification setting value is "5", the risk notification control device 327 sets the notification mode to analog notification mode and predicted assistance notification mode, and turns on risk notification under these set notification modes.

[0082] Here, when the notification mode is set to the predictive assistance notification mode, the risk notification control device 327 generates risk avoidance support information that is useful for the rider to avoid an approaching risk based on the risk information transmitted from the collaboration support device 6, and activates the head-mounted speaker 321 and head-up display 322 of the HMI 320 in a manner that allows the rider to perceive this risk avoidance support information audibly and visually. Here, the risk avoidance support information includes information on the location of risk objects that may come into contact with the vehicle, information on the location of risk occurrence, and information that alerts the rider to the risk objects.

[0083] More specifically, when there is a four-wheeled vehicle driven by a driver in an unhealthy state ahead of the motorcycle driven by the rider, the risk notification control device 327 issues a message such as "Watch out for the dangerous right turn of the four-wheeled vehicle" as risk avoidance support information for avoiding contact with the four-wheeled vehicle by voicing it through the head-mounted speaker 321 or displaying it on the head-up display 322. In this case, the risk notification control device 327 may also display an image of an arrow pointing to the current position or predicted position of the four-wheeled vehicle on the head-up display 322 as risk avoidance support information for avoiding contact with the four-wheeled vehicle.

[0084] Furthermore, when the notification mode is set to the presence notification mode, the risk notification control device 327 operates the HMI 320 in a manner that does not bother the rider, thereby allowing the rider to naturally recognize the presence of a risk target extracted from the risk information transmitted from the collaboration support device 6. In this manner, in the presence notification mode, in order to naturally allow the rider to recognize the presence of a risk target without feeling bothered, it is preferable that the risk notification control device 327 activates, among the multiple devices included in the HMI 320, the head-mounted speaker 321, which particularly appeals to the rider's hearing. More specifically, when the notification mode is set to the presence notification mode, the risk notification control device 327 causes the head-mounted speaker 321 to emit a familiar sound effect at a low volume using directional binaural sound directed toward the location of the risk target or the location of the risk occurrence point, thereby naturally directing the rider's gaze toward the location of the risk target or the risk occurrence point.

[0085] Furthermore, when the notification mode is set to the analog notification mode, the risk notification control device 327 operates the HMI 320 in a manner different from that of the presence notification mode described above, thereby making the rider more aware of the presence of a risk object extracted from the risk information transmitted from the collaboration support device 6 and the degree of risk associated with this risk object. In this manner, in the analog notification mode, in order to make the rider more aware of the presence of a risk object, the risk notification control device 327 operates the HMI 320 in a manner with a higher notification intensity than that determined in the presence notification mode. More specifically, when the notification mode is set to the analog notification mode, the risk notification control device 327 causes the head-mounted speaker 321 to emit a buzzer sound or pulse sound at a volume higher than the sound effects emitted in the presence notification mode. These buzzer sounds and pulse sounds are louder and less familiar to the rider than the sound effects emitted in the presence notification mode, and therefore have a higher notification intensity than the sound effects emitted in the presence notification mode.

[0086] Furthermore, in the analog notification mode as described above, in order to make the rider more aware of the risk level associated with a risk object in addition to the presence of the risk object, it is preferable that the risk notification control device 327 varies the notification intensity according to the risk level associated with the risk object (for example, the predicted time to collision with the risk object) extracted from the risk information transmitted from the collaboration support device 6. Specifically, the risk notification control device 327 may increase the volume of the buzzer sound, increase the volume of the pulse sound, or shorten the interval between the pulse sounds as the risk level increases (i.e., as the predicted time to collision decreases), thereby increasing the notification intensity.

[0087] Furthermore, when the risk notification control device 327 changes the notification intensity according to the degree of risk in this manner, it is preferable to operate the HMI 320 so that the notification intensity is maximized at the time when the above-mentioned driving assistance ECU begins to execute collision mitigation brake control, in other words, at the time when the risk object enters the ADAS operating range of the vehicle.

[0088] The risk area notification control device 328 performs risk area notification to notify the driver of information related to the current risk area by operating the HMI 320 based on the risk area information successively transmitted from the collaboration support device 6. When the risk area notification control device 328 determines based on the risk area information that the vehicle is traveling in a local area with a high degree of risk, it issues a message such as "You are traveling in a high-risk area. Please be careful of your surroundings" through the head-mounted speaker 321 or displays it on the head-up display 322.

[0089] Returning to FIG. 2 , the mobile information processing terminal 40 owned or worn by the pedestrian 4 in the target traffic area 9 is configured, for example, as a wearable terminal worn by the pedestrian 4 or a smartphone held by the pedestrian 4. The wearable terminal has the function of measuring biometric information of the pedestrian 4, such as heart rate, blood pressure, and blood oxygen saturation, transmitting the measurement data of this biometric information to the collaboration support device 6, and receiving collaboration support information transmitted from the collaboration support device 6. The smartphone also has the function of transmitting pedestrian information about the pedestrian 4, such as location information, movement acceleration, and schedule information of the pedestrian 4, to the collaboration support device 6, and receiving collaboration support information transmitted from the collaboration support device 6.

[0090] The mobile information processing terminal 40 also includes a notification device 42 that notifies pedestrians of various information through their hearing, vision, touch, etc. by operating the HMI in a manner determined based on the received collaborative support information.

[0091] 3C is a block diagram showing the configuration of notification device 42 installed in mobile information processing terminal 40. Note that FIG. 3C illustrates only the blocks of notification device 42 that are particularly involved in control based on collaboration support information transmitted from collaboration support device 6.

[0092] The notification device 42 includes an HMI 420 that operates in a manner that can be recognized by pedestrians, and an HMI control device 425 that operates the HMI 420 based on the collaboration support information transmitted from the collaboration support device 6.

[0093] The HMI 420 includes a speaker 421 that operates in a manner that can be heard by pedestrians, and a vibration device 424 that operates in a manner that can be heard by pedestrians through their sense of touch.

[0094] The speaker 421 produces a sound in response to a command from the HMI control device 425. The vibration device 424 vibrates the main body of the portable information processing terminal 40 with an amplitude and / or a frequency in a manner in response to a command from the HMI control device 425.

[0095] As will be explained later, the collaborative support information sent from the collaborative support device 6 to the mobile information processing terminal 40 carried by the pedestrian includes information regarding the risk notification setting value for turning risk notifications on / off by the HMI control device 425 and setting the type of notification mode, as well as risk information regarding risks approaching the pedestrian.

[0096] The HMI control device 425 can provide risk notifications under multiple notification modes that differ in at least one of the target device and operation mode of the HMI 420. More specifically, the HMI control device 425 can provide risk notifications under at least one of a presence notification mode intended to make pedestrians aware of the presence of a potential risk and an analog notification mode intended to make pedestrians aware of the presence and / or severity of an actual risk. For this reason, the risk notification setting value input to the HMI control device 425 is set to one of the following values: “0” that turns off the risk notification by the HMI control device 425; “1” that turns on the risk notification by the HMI control device 425 and sets the notification mode to the presence notification mode; and “2” that turns on the risk notification by the HMI control device 425 and sets the notification mode to the analog notification mode.

[0097] The HMI control device 425 sets the risk notification to off when the risk notification setting value is "0." That is, the HMI control device 425 does not operate the HMI 420 when the risk notification setting value is "0."

[0098] If the risk notification setting value is "1", the HMI control device 425 sets the notification mode to the presence notification mode and turns on risk notification under the set notification mode.

[0099] Furthermore, if the risk notification setting value is "2", the HMI control device 425 sets the notification mode to the analog notification mode and turns on risk notification under the set notification mode.

[0100] Here, when the notification mode is set to the presence notification mode, the HMI control device 425 operates the HMI 420 in a manner that does not bother the pedestrian, thereby allowing the pedestrian to naturally recognize the presence of a risk target extracted from the risk information transmitted from the collaboration support device 6. More specifically, when the notification mode is set to the presence notification mode, the HMI control device 425 operates the vibration device 424 to vibrate the main body of the mobile information processing terminal 40 at a predetermined amplitude and frequency.

[0101] Furthermore, when the notification mode is set to the analog notification mode, the HMI control device 425 operates the HMI 420 in a manner different from the above-described presence notification mode, thereby making pedestrians more aware of the presence of a risk condition extracted from the risk information transmitted from the collaboration support device 6 and the degree of risk associated with this risk object. In this way, in the analog notification mode, in order to make pedestrians more aware of the presence of a risk object, the HMI control device 425 operates the HMI 420 in a manner with a higher notification intensity than the manner determined in the presence notification mode. More specifically, when the notification mode is set to the analog notification mode, the HMI control device 425 causes the speaker 421 to emit a buzzer sound, a pulse sound, a message indicating the presence of a risk, or the like.

[0102] Furthermore, in the analog notification mode as described above, in order to make pedestrians more aware of the presence of a risk object as well as the degree of risk associated with this risk object, it is preferable that the HMI control device 425 changes the notification intensity according to the degree of risk associated with the risk object (for example, the predicted time to collision with the risk object) extracted from the risk information transmitted from the collaboration support device 6. Specifically, the HMI control device 425 may increase the notification intensity by increasing the volume of the buzzer sound, increasing the volume of the pulse sound, shortening the interval between pulse sounds, increasing the volume of the message, or changing the content of the message as the degree of risk increases (i.e., the shorter the predicted time to collision).

[0103] Returning to Figure 2, the infrastructure camera 56 captures images of traffic infrastructure facilities including roadways, intersections, and sidewalks in the target traffic area, as well as moving objects and pedestrians moving on these roadways, intersections, sidewalks, etc., and transmits the obtained image information to the collaboration support device 6.

[0104] The signal control device 55 controls the traffic lights and transmits to the cooperation support device 6 traffic light status information relating to the current lighting color of the traffic lights installed in the target traffic area and the timing for changing the lighting color.

[0105] The collaboration support device 6 is a computer that supports safe and smooth traffic for traffic participants in the target traffic area by generating collaboration support information for each traffic participant to be supported, based on information acquired from multiple area terminals present in the target traffic area as described above, to promote communication between traffic participants and awareness of the surrounding traffic environment, and notifying each traffic participant of the information. In this embodiment, the collaboration support device 6 supports traffic participants among the multiple traffic participants present in the target traffic area who are equipped with means (e.g., the in-vehicle device group 20, 30, the mobile information processing terminal 40, the notification device 22, 32, 42) that receive the collaboration support information generated by the collaboration support device 6 and operate their HMI in a manner determined based on the received collaboration support information.

[0106] The collaborative assistance device 6 includes a target traffic area recognition unit 60 that recognizes people and moving objects in the target traffic area as individual traffic participants, a driving subject information acquisition unit 61 that acquires driving subject state information correlated with the driving ability of the driving subject of the moving object recognized as a traffic participant by the target traffic area recognition unit 60, a prediction unit 62 that predicts the future of the traffic participants in the target traffic area, a health notification setting unit 63 that sets the health notification on / off for each traffic participant recognized by the target traffic area recognition unit 60 as a support target, a risk notification setting unit 64 that sets the notification mode of the risk notification for each traffic participant recognized by the target traffic area recognition unit 60 as a support target, a collaborative assistance information notification unit 65 that transmits collaborative assistance information generated for each traffic participant recognized by the target traffic area recognition unit 60 as a support target, a traffic environment database 67 that stores information about the traffic environment in the target traffic area, and a driving history database 68 that stores information about the past driving history of pre-registered driving subjects.

[0107] The traffic environment database 67 stores information related to the traffic environment of traffic participants in the target traffic area, such as map information of the target traffic area that has been registered in advance (for example, roadway width, number of lanes, speed limit, sidewalk width, presence or absence of guardrails between the roadway and sidewalk, and location of crosswalks), and risk area information related to areas of the target traffic area that are particularly high in risk. Hereinafter, the information stored in the traffic environment database 67 will also be referred to as registered traffic environment information. Note that the risk area information stored in this traffic environment database 67 is information that is updated every few hours to several days, whereas the risk area information described below is information that is updated almost in real time.

[0108] The driving history database 68 stores information about the past driving history of pre-registered drivers in association with the registration number of the mobile vehicle owned by the driver. Therefore, if the target traffic area recognition unit 60, which will be described later, can identify the registration number of the recognized mobile vehicle, the driving history of the driver of the recognized mobile vehicle can be obtained by searching the driving history database 68 based on this registration number. Hereinafter, the information stored in the driving history database 68 will also be referred to as registered driving history information.

[0109] The target traffic area recognition unit 60 recognizes recognition objects including each traffic participant who is a person or a moving object in the target traffic area and the traffic environment of each traffic participant in the target traffic area based on information transmitted from the above-mentioned area terminals (the in-vehicle device group 20, 30, the mobile information processing terminal 40, the infrastructure camera 56, and the signal control device 55) in the target traffic area and registered traffic environment information read from the traffic environment database 67, and obtains recognition information regarding these recognition objects.

[0110] The information transmitted from the on-board driving assistance device 21 and on-board communication device 24 included in the on-board device group 20 to the target traffic area recognition unit 60, and the information transmitted from the on-board driving assistance device 31 and on-board communication device 34 included in the on-board device group 30 to the target traffic area recognition unit 60, include information on the status of traffic participants and the traffic environment around the vehicle acquired by the external sensor unit, and information on the status of the vehicle as a traffic participant acquired by the vehicle status sensor, navigation device, etc. The information transmitted from the mobile information processing terminal 40 to the target traffic area recognition unit 60 also includes information on the status of pedestrians as traffic participants, such as their position and movement acceleration. The image information transmitted from the infrastructure camera 56 to the target traffic area recognition unit 60 also includes information on each traffic participant and their traffic environment, such as the appearance of traffic infrastructure facilities in the target traffic area, such as roadways, intersections, and sidewalks, and the appearance of traffic participants moving through the target traffic area. The traffic light status information transmitted from the signal control device 55 to the target traffic area recognition unit 60 also includes information on the traffic environment of each traffic participant, such as the current lighting color of the traffic light and the timing for changing the lighting color. The registered traffic environment information that the target traffic area recognition unit 60 reads from the traffic environment database 67 includes information about the traffic environment of each traffic participant, such as map information of the target traffic area and dangerous area information.

[0111] Therefore, based on the information transmitted from these area terminals, the target traffic area recognition unit 60 can acquire recognition information of each traffic participant in the target traffic area, such as the position, movement speed, movement acceleration, movement direction, vehicle type of the moving body, vehicle class of the moving body, registration number of the moving body, the number of pedestrians, the age group of the pedestrians, etc. (hereinafter also referred to as "traffic participant recognition information"). Furthermore, based on the information transmitted from these area terminals, the target traffic area recognition unit 60 can acquire recognition information of the traffic environment of each traffic participant in the target traffic area (hereinafter also referred to as "traffic environment recognition information"), such as the width of the roadway, the number of lanes, the speed limit, the width of the sidewalk, the presence or absence of guardrails between the roadway and the sidewalk, the color of traffic lights and their switching timing, the weather, illuminance, road surface conditions, and dangerous area information.

[0112] Therefore, in this embodiment, the recognition means for recognizing traffic participants and the traffic environment in the target traffic area is composed of a target traffic area recognition unit 60, an on-board driving assistance device 21, an on-board communication device 24, and a portable information processing terminal 25 included in the on-board device group 20 of the four-wheeled vehicle 2, an on-board driving assistance device 31, an on-board communication device 34, and a portable information processing terminal 35 included in the on-board device group 30 of the motorcycle 3, the portable information processing terminal 40 of the pedestrian 4, an infrastructure camera 56, a traffic light control device 55, and a traffic environment database 67.

[0113] The target traffic area recognition unit 60 transmits the traffic participant recognition information and traffic environment recognition information acquired in the above manner to the driving subject information acquisition unit 61, the prediction unit 62, the health notification setting unit 63, the risk notification setting unit 64, and the collaborative support information notification unit 65, etc.

[0114] The driving subject information acquisition unit 61 acquires driving subject state information and driving subject characteristic information that are correlated with the current driving ability of the driving subject of the mobile body recognized as a traffic participant by the target traffic area recognition unit 60 based on information transmitted from the area terminals (particularly, the vehicle-mounted device group 20, 30) in the target traffic area and registered driving history information read from the driving history database 68.

[0115] More specifically, when the driver of a four-wheeled vehicle recognized as a traffic participant by the target traffic area recognition unit 60 is a human, the driver subject information acquisition unit 61 acquires information transmitted from the on-board device group 20 mounted on the four-wheeled vehicle as driver subject state information of the driver. Furthermore, when the driver of a motorcycle recognized as a traffic participant by the target traffic area recognition unit 60 is a human, the driver subject information acquisition unit 61 acquires information transmitted from the on-board device group 30 mounted on the motorcycle as driver subject state information of the rider.

[0116] The information transmitted from the driver's subject state sensor 23 and the in-vehicle communication device 24 included in the in-vehicle device group 20 to the driver's subject information acquisition unit 61 includes time-series data on the driver's appearance, such as the driver's gaze direction and whether or not their eyes are open, biological information, such as pulse, whether or not they are breathing, and skin potential, and audio information, such as whether or not they are talking, which is correlated with the driver's driving ability. The information transmitted from the rider's state sensor 33 and the in-vehicle communication device 34 included in the in-vehicle device group 30 to the driver's subject information acquisition unit 61 includes time-series data on the rider's biological information, such as pulse, whether or not they are breathing, and skin potential, which is correlated with the rider's driving ability. The information transmitted from the mobile information processing terminals 25 and 35 included in the in-vehicle device groups 20 and 30 to the driver's subject information acquisition unit 61 also includes the driver's and rider's personal schedule information. For example, when a driver or rider is driving a vehicle under a tight schedule, they may become impatient and their driving ability may decline. Therefore, the driver's and rider's personal schedule information can be considered information correlated with their own driving ability.

[0117] The driver subject information acquisition unit 61 acquires driver subject characteristic information regarding the driver subject's driving characteristics (e.g., excessive sudden lane changes, excessive sudden acceleration / deceleration, etc.) that are correlated with the driver subject's current driving ability while driving, by using both or either of the driver subject state information for the driver subject acquired by the above procedure and the registered driving history information read from the driving history database 68.

[0118] The driver subject information acquisition unit 61 transmits the driver subject state information and driver subject characteristic information of the driver subject acquired in the manner described above to the prediction unit 62, the health notification setting unit 63, the risk notification setting unit 64, and the collaborative support information notification unit 65, etc.

[0119] The prediction unit 62 predicts risks in the entire target traffic area based on the traffic participant recognition information and traffic environment recognition information (hereinafter collectively referred to as "recognition information") acquired by the target traffic area recognition unit 60, and the driving subject state information and driving subject characteristic information (hereinafter collectively referred to as "driving subject information") acquired by the driving subject information acquisition unit 61.

[0120] FIG. 4 is a functional block diagram showing a specific configuration of the prediction unit 62. As shown in FIG. The prediction unit 62 includes an area risk prediction unit 620 that predicts the risk to traffic participants in units of multiple local areas obtained by subdividing the target traffic area, and a traffic participant risk prediction unit 625 that predicts the risk for each individual traffic participant within each local area.

[0121] Here, the target traffic area is a relatively wide traffic area defined, for example, on a city, town, or village basis. In contrast, individual local areas obtained by subdividing the target traffic area are traffic areas that a four-wheeled vehicle traveling at legal speeds can pass through in about several tens of seconds, such as areas near intersections or specific facilities. That is, each local area is set smaller than the target traffic area but larger than the ADAS operation range of the driving assistance ECU installed in each moving object. The range of each local area may be constant or may vary depending on the situation. Furthermore, part of the range of each local area may overlap part of the range of an adjacent local area.

[0122] The area risk prediction unit 620 comprises a statistical processing calculation unit 621, a data pre-processing calculation unit 622, a macro risk estimation model 623, and a high risk area extraction unit 624, and by using these, predicts the risk for each individual local area.

[0123] The statistical processing calculation unit 621 extracts information correlated with the degree of risk for each local area by performing predetermined statistical processing on the recognition information and driver information for the entire target traffic area.

[0124] The data pre-processing calculation unit 622 generates input data for the macro risk estimation model 623 based on the information that has undergone statistical processing by the statistical processing calculation unit 621 , and inputs the data to the macro risk estimation model 623 .

[0125] The macro risk estimation model 623 includes a DNN constructed by machine learning so as to output the degree of risk for each local area when input data that has been processed by the data preprocessing calculation unit 622 is input. Information regarding the degree of risk for each local area calculated by the macro risk estimation model 623 is transmitted to the high-risk area extraction unit 624 and the collaboration support information notification unit 65.

[0126] The high-risk area extraction unit 624 extracts at least one of the multiple local areas constituting the target traffic area as a high-risk area based on the risk degree of each individual local area calculated by the macro risk estimation model 623. More specifically, the high-risk area extraction unit 624 extracts, for example, from the multiple local areas, a local area whose risk degree calculated by the macro risk estimation model 623 is higher than a predetermined threshold as a high-risk area. Information on the high-risk area extracted by the high-risk area extraction unit 624 is transmitted to the traffic participant risk prediction unit 625.

[0127] The traffic participant risk prediction unit 625 includes a monitoring area information extraction unit 626, a data preprocessing calculation unit 627, and a micro risk estimation model 628, and by using these, it sets only the high-risk areas extracted by the area risk prediction unit 620 as monitoring areas and predicts the future risk of each traffic participant in these monitoring areas. In other words, the traffic participant risk prediction unit 625 does not set local areas (hereinafter also referred to as "low-risk areas") that were not extracted as high-risk areas by the area risk prediction unit 620 as monitoring areas, and does not perform the processing described below.

[0128] The monitoring area information extraction unit 626 extracts information related to the monitoring area extracted as a high-risk area by the area risk prediction unit 620 from the recognition information and driving subject information (i.e., information from the recognition information regarding traffic participants and their traffic environment present in the monitoring area, and information from the driving subject information regarding the driving subject of the moving object present in the monitoring area).

[0129] The data pre-processing calculation unit 627 generates input data for the micro risk estimation model 628 based on the recognition information and driving subject information regarding the monitoring area extracted by the monitoring area information extraction unit 626, and inputs the generated input data to the micro risk estimation model 628.

[0130] The micro risk estimation model 628 includes a DNN constructed by machine learning so that, when input data related to a monitoring area that has been processed by the data pre-processing calculation unit 622 is input, the model 628 outputs information related to the future risk of each traffic participant in the monitoring area (more specifically, information related to the movement paths of each traffic participant, information related to the contact risk of each traffic participant, and the predicted time until the contact risk occurs, etc.). The information related to the risk of traffic participants in the monitoring area calculated by the micro risk estimation model 628 is transmitted to the risk notification setting unit 64 and the collaboration support information notification unit 65.

[0131] As described above, the prediction unit 62, which predicts risks in a target traffic area by combining and using the macro risk estimation model 623 and the micro risk estimation model 628, can improve its prediction accuracy by learning the macro risk estimation model 623 and the micro risk estimation model 628 using the following two procedures while operating a service that provides cooperative support information to each traffic participant using the traffic safety support system 1.

[0132] <First learning method> The first learning method comprises the steps of preparing learning data by using input data for the macro risk estimation model 623 generated based on recognition information and driving subject information acquired during the operation of the service, and the output of the micro risk estimation model 628 when input data prepared using the same recognition information and driving subject information is input to the micro risk estimation model 628, and using the learning data to learn the macro risk estimation model 623. This first learning method uses the output of the micro risk estimation model 628 as correct answer data, and is therefore useful when a highly accurate micro risk estimation model 628 has been obtained in advance.

[0133] <Second learning method> The second learning method includes the steps of preparing learning data by using input data for the macro risk estimation model 623 generated based on the first recognition information and the first driving subject information acquired during a predetermined first period during the operation of the service, and correct answer data for the output of the micro risk estimation model 628 generated based on the second recognition information and the second driving subject information acquired during a second period immediately following the first period, and using the learning data to learn an overall model that combines the macro risk estimation model 623 and the micro risk estimation model 628. This second learning method requires manual preparation of correct answer data based on the second recognition information, and therefore requires more time and effort than the first learning method, but it is possible to learn both the macro risk estimation model 623 and the micro risk estimation model 628, and therefore the prediction accuracy of the entire prediction unit 62 can be improved compared to the first learning method.

[0134] 2, the soundness notification setting unit 63 sets as setting targets traffic participants who are recognized as support targets and moving bodies by the target traffic area recognition unit 60 among multiple traffic participants present in the target traffic area, and sets the soundness notification on / off for each of the setting targets. Note that, as will be described in detail later, traffic participants who are parties to a contact risk predicted to occur by the prediction unit 62 are targets for setting risk notifications by the risk notification setting unit 64. For this reason, it is preferable to exclude targets for setting by the risk notification setting unit 64 from the setting targets of the soundness notification setting unit 63.

[0135] More specifically, the soundness notification setting unit 63 first acquires driver-subject state information and driver-subject characteristic information associated with the driver of each setting target, which is a moving object, from the driver-subject information acquisition unit 61. The soundness notification setting unit 63 then calculates the current soundness of the driver for each setting target based on the acquired driver-subject state information and driver-subject characteristic information. If the soundness calculated for each setting target is less than a predetermined soundness threshold, the soundness notification setting unit 63 determines that the driver of that setting target is in an unsound state and sets the soundness notification setting value for that setting target to "1" to turn on the soundness notification for that setting target. If the soundness calculated for each setting target is equal to or greater than the soundness threshold, the soundness notification setting unit 63 determines that the driver of that setting target is in a sound state and sets the soundness notification setting value for that setting target to "0" to turn off the soundness notification for that setting target.

[0136] The health notification setting unit 63 sets the health notification for multiple setting targets in the target traffic area to on or off by the above procedure. Information on the health notification setting value set for each setting target by the health notification setting unit 63 is transmitted to the collaboration support information notification unit 65.

[0137] The risk notification setting unit 64 sets as setting targets traffic participants who are recognized as support targets by the target traffic area recognition unit 60 among multiple traffic participants present in a monitoring area extracted as a high-risk area from the target traffic area by the prediction unit 62, and sets the operation mode of risk notification (i.e., the type of notification mode and on / off of risk notification) for each setting target based on the prediction results by the traffic participant risk prediction unit 625 of the prediction unit 62, the recognition information acquired by the target traffic area recognition unit 60, and the driving subject information acquired by the driving subject information acquisition unit 61, etc.

[0138] More specifically, the risk notification setting unit 64 sets the operation mode of risk notification for each set target present in the monitoring area based on the information related to the monitoring area among the recognition information acquired by the target traffic area recognition unit 60, the information related to the monitoring area among the driving subject information acquisition unit 61, and the prediction result for the monitoring area by the traffic participant risk prediction unit 625. That is, the risk notification setting unit 64 sets the risk notification setting value for each set target to any one of "0", "1", "2", "3", and "4".

[0139] In this way, the risk notification setting unit 64 sets the operation mode of the risk notification for each individual set target present in the monitoring area, so that, for example, if the traffic participant risk prediction unit 625 predicts the occurrence of a collision risk involving multiple set targets within the monitoring area, it is possible to turn on / off the risk notification at different times for each of the multiple predicted parties predicted to be involved in the collision risk, or to simultaneously send risk notifications in different notification modes for each of them. Hereinafter, the process in the risk notification setting unit 64 of setting the operation mode of the risk notification appropriate for each individual set target is also referred to as the "risk notification optimization process."

[0140] 5 is a diagram schematically illustrating the concept of the risk notification optimization process in the risk notification setting unit 64. Note that the procedure of the risk notification optimization process will be described below using an example in which the traffic participant risk prediction unit 625 predicts the occurrence of a contact risk between two parties (i.e., a first set target (moving body) and a second set target (moving body)), but the present invention is not limited to this. It is easy to generalize to cases in which a contact risk is predicted where one of the two parties is a pedestrian, or to cases in which a contact risk is predicted between three parties, and therefore the description will be omitted.

[0141] The left side of FIG. 5 schematically illustrates the transition of the risk notification operation mode for the first set target, and the right side of FIG. 5 schematically illustrates the transition of the risk notification operation mode for the second set target. The two arrows at the top of FIG. 5 conceptually illustrate the time it takes from when the traffic participant risk prediction unit 625 first predicts that a collision risk will occur until the first set target and the second set target collide, i.e., the collision prediction time. However, because these two arrows merely conceptually illustrate the collision prediction time, they do not mean that the risk notification optimization process in the risk notification setting unit 64 cannot be executed unless the collision prediction time has been clearly calculated by the traffic participant risk prediction unit 625. The risk notification optimization process in the risk notification setting unit 64 can be executed from a stage before the traffic participant risk prediction unit 625 clearly calculates the collision prediction time. FIG. 5 also illustrates a case where the risk notification for the first set target and the second set target is set to off (i.e., the risk notification setting value is "0") when the traffic participant risk prediction unit 625 first predicts that a collision risk will occur.

[0142] When the traffic participant risk prediction unit 625 predicts the occurrence of a collision risk involving multiple support targets within a monitoring area, the risk notification setting unit 64 first sets priorities for the multiple predicted parties involved in the collision risk (in the example of FIG. 5, the first set target and the second set target) based on the content of the collision risk predicted by the traffic participant risk prediction unit 625. As will be described in detail later, this priority stipulates the order in which risk notifications (particularly risk notifications under the presence notification mode) are set to on, and risk notifications are set to on for set targets with higher priorities before set targets with lower priorities. Note that FIG. 5 illustrates a case in which the priority of the first set target is set higher than the priority of the second set target.

[0143] Here, the risk notification setting unit 64 sets a priority for each set target so that the predicted contact risk is avoided from manifesting or occurring and traffic flow between these set targets is not disrupted. More specifically, the risk notification setting unit 64 may identify a risk inducer who induces a contact risk from among multiple predicted parties involved in the contact risk, for example, by referring to the prediction result by the traffic participant risk prediction unit 625, the recognition information by the target traffic area recognition unit 60, and the driver information by the driver information acquisition unit 61, and set a higher priority for this risk inducer than for other predicted parties excluding this risk inducer. By setting a higher priority for such a risk inducer and turning on risk notifications before turning on risk notifications for other set targets, the risk inducer's behavior can be modified before risk notifications for the other set targets are turned on, thereby avoiding the originally predicted contact risk from manifesting or occurring.

[0144] Here, risk inducers include those who perform actions that are highly likely to induce the above-mentioned risks of contact (for example, sudden acceleration, sudden deceleration, sudden lane changes, cutting in, closing the distance to a vehicle in front or behind, continuing to drive across lanes, snaking, driving the wrong way, ignoring traffic lights, driving at a speed that is more than a predetermined speed faster than surrounding moving objects, driving at a speed that is more than a predetermined speed slower than surrounding moving objects, driving at a speed that is more than a predetermined speed faster than the speed limit, driving at a speed that is more than a predetermined speed slower than the speed limit, and actions that obstruct the movement of surrounding traffic participants).

[0145] The risk notification setting unit 64 may also set priorities based on the traffic environment of each set target. More specifically, a predicted party who is in a traffic environment where it is difficult for the other predicted parties to recognize the presence of the other predicted parties may be given a higher priority than the other predicted parties, and the risk notification may be set to on before the other set targets. This can improve the cognitive ability of the set target who is set to a high priority, thereby avoiding the actualization or occurrence of the initially predicted contact risk.

[0146] After the risk notification setting unit 64 sets priorities for each set target according to the above procedure in response to the occurrence of a collision risk predicted by the traffic participant risk prediction unit 625, the risk notification setting unit 64 periodically determines whether the initially predicted collision risk has materialized. More specifically, for example, if the traffic participant risk prediction unit 625 predicts the occurrence of a collision risk and the collision prediction time for this collision risk is equal to or greater than a predetermined materialization threshold (including cases where the traffic participant risk prediction unit 625 has not calculated a clear collision prediction time), the risk notification setting unit 64 determines that the collision risk has not materialized (i.e., the collision risk is potential). Furthermore, for example, if the collision prediction time calculated by the traffic participant risk prediction unit 625 falls below the materialization threshold, the risk notification setting unit 64 determines that the collision risk has materialized. Here, the materialization threshold, which is a threshold for the collision prediction time, is set to be wider than the ADAS operation range, as shown in FIG. 5 , in other words, longer than the collision prediction time at which the driving assistance ECU installed in each vehicle starts to execute collision mitigation braking control, collision avoidance steering control, or the like.

[0147] Furthermore, before it is determined that the initially predicted contact risk has materialized, i.e., while it is determined that the contact risk is potential, the risk notification setting unit 64 starts risk notification in the presence notification mode, starting with the set target with the highest priority (the first set target in the example of FIG. 5). That is, the risk notification setting unit 64 sets the risk notification setting value to "1" or "3" starting with the set target with the highest priority. As a result, a driver of the set target who has received a risk notification in this presence notification mode may recognize the presence of a moving object (the second set target in the example of FIG. 5) that may contact his or her vehicle and take action to avoid the predicted contact risk. If a driver who has received such a risk notification takes action to avoid the contact risk, the traffic participant risk prediction unit 625 may predict that the initially predicted contact risk will not occur before it materializes.

[0148] Furthermore, for a set target set to a low priority (the second set target in the example of FIG. 5), the risk notification setting unit 64 starts risk notification in the presence notification mode for a set target set to a high priority, and then starts risk notification in the presence notification mode a predetermined time later. That is, the risk notification setting unit 64 sets the risk notification setting value for the set target set to a high priority to "1" or "3", and then sets the risk notification setting value for the set target set to a low priority to "1" or "3" a predetermined time later. Note that, in order to prevent disruption of traffic flow for a set target set to a low priority, the risk notification setting unit 64 may not perform risk notification in the presence notification mode for this set target set to a low priority until a contact risk becomes apparent. Furthermore, since the occurrence of a contact risk may be avoided by providing a risk notification in the presence notification mode in advance to a set target that has been set to a high priority as described above, if the driver of the set target does not take any action to avoid the contact risk even after a predetermined time has elapsed after the risk notification setting unit 64 has started a risk notification in the presence notification mode for a set target that has been set to a low priority.

[0149] Furthermore, after determining that the initially predicted collision risk has materialized, the risk notification setting unit 64 starts issuing risk notifications in analog notification mode to all predicted parties involved in the collision risk. That is, after determining that the collision risk has materialized, the risk notification setting unit 64 sets the risk notification setting value for all predicted parties to "2" or "4." As described above, in analog notification mode, the shorter the collision prediction time, the higher the notification intensity, so that all predicted parties involved in the collision risk can be made to feel a sense of crisis about the approaching collision risk and can take action to avoid the collision risk.

[0150] Returning to Figure 2, the collaborative assistance information notification unit 65 generates collaborative assistance information for each traffic participant recognized as a support target by the target traffic area recognition unit 60 to encourage communication with surrounding traffic participants and awareness of the surrounding traffic environment based on the recognition information acquired by the target traffic area recognition unit 60, the driving subject information acquired by the driving subject information acquisition unit 61, the prediction results for the monitoring area by the traffic participant risk prediction unit 625, information on the risk level for each individual local area by the area risk prediction unit 620 (hereinafter also referred to as "risk area information"), information on the health setting value set by the health notification setting unit 63, and information on the risk notification setting value set by the risk notification setting unit 64, and transmits the generated collaborative assistance information to each traffic participant.

[0151] Here, the collaborative assistance information notification unit 65 transmits collaborative assistance information to assistance targets that are located within the monitoring area (i.e., high-risk area) targeted by the traffic participant risk prediction unit 625, among the multiple assistance targets present throughout the target traffic area, including information regarding the risk notification setting value set based on the prediction result by the traffic participant risk prediction unit 625 and risk area information generated based on the estimation result by the area risk prediction unit 620.

[0152] In addition, the collaborative support information notification unit 65 transmits collaborative support information including risk area information generated based on the estimation results by the area risk prediction unit 620 to support targets that are located in low-risk areas that were not extracted as high-risk areas by the area risk prediction unit 620, among the multiple support targets located throughout the target traffic area.

[0153] The traffic safety support system 1 and the learning method thereof according to this embodiment have the following advantages. (1) The traffic safety support system 1 includes a target traffic area recognition unit 60 that recognizes recognition targets, including traffic participants (including people and moving objects) in a target traffic area 9 and the traffic environment of each traffic participant, and acquires recognition information related to these recognition targets; a prediction unit 62 that predicts risks in the target traffic area 9 based on the recognition information; and a cooperative support information notification unit 65 that transmits support information generated based on the recognition information and the prediction results by the prediction unit 62 to a designated support target from among multiple traffic participants in the target traffic area 9. In addition, the prediction unit 62 uses an area risk prediction unit 620 to extract at least one of multiple local areas obtained by dividing the target traffic area 9 as a high-risk area, and a traffic participant risk prediction unit 625 to predict future risks to traffic participants in the high-risk area. Here, when extracting high-risk areas from the multiple local areas, the area risk prediction unit 620 uses information obtained by performing statistical processing on the recognition information, thereby enabling extraction of high-risk areas with less load than when a huge amount of recognition information related to recognition targets in the target traffic area 9 is directly used. Furthermore, the traffic participant risk prediction unit 625 sets the high-risk area as a monitoring area, and when predicting the future risk of traffic participants in this monitoring area, by using information related to the monitoring area from the recognition information on the recognition objects in the entire target traffic area 9, it is possible to predict the future risk of traffic participants with less load than when using a huge amount of recognition information on the recognition objects in the target traffic area 9 as is. Therefore, the traffic safety support system 1 can provide traffic participants in high-risk areas with appropriate support information generated based on the prediction results in real time, thereby improving the safety, convenience, and smoothness of traffic in the target traffic area 9.

[0154] (2) In the traffic safety support system 1, the traffic participant risk prediction unit 625 does not predict future risks of traffic participants in low-risk areas that are not extracted as high-risk areas by the area risk prediction unit 620 among the multiple local areas. Therefore, the traffic safety support system 1 can reduce the computational load compared to when the traffic participant risk prediction unit 625 performs prediction processing for all local areas. Furthermore, the traffic safety support system 1 can reduce the computational load by limiting the number of local areas for which prediction processing is performed, thereby improving the accuracy of predictions of risks of traffic participants in high-risk areas. Therefore, the traffic safety support system 1 can provide appropriate support information generated based on the highly accurate prediction results of the traffic participant risk prediction unit 625 to traffic participants in high-risk areas in real time, thereby further improving the safety, convenience, and smoothness of traffic in the target traffic area 9.

[0155] (3) In the traffic safety support system 1, the area risk prediction unit 620 estimates the risk level for each local area and extracts high-risk areas from among the multiple local areas based on the risk level estimation results. The collaboration support information notification unit 65 transmits collaboration support information including information on the risk notification setting value generated by the risk notification setting unit 64 based on the relatively detailed prediction results by the traffic participant risk prediction unit 625 to support targets located in high-risk areas among the multiple support targets in the entire target traffic area 9. This makes it possible to improve traffic safety, convenience, and smoothness of traffic for traffic participants in high-risk areas. The collaboration support information notification unit 65 also transmits collaboration support information including risk area information generated based on the estimation results for each local area by the area risk prediction unit 620 to support targets located in low-risk areas outside the high-risk areas among the multiple support targets in the entire target traffic area 9. This makes it possible to improve traffic safety, convenience, and smoothness of traffic for traffic participants in low-risk areas. In this way, the traffic safety support system 1 can improve the safety, convenience, and smoothness of traffic for traffic participants throughout the entire target traffic area 9 by changing the content of the collaborative support information according to the level of risk in each individual local area.

[0156] (4) The first learning method prepares learning data by utilizing input data for the macro risk estimation model 623 generated based on recognition information and the output of the micro risk estimation model 628 when this recognition information is input to the micro risk estimation model 628, and then uses this learning data to train the macro risk estimation model 623. In general model training, it is necessary to prepare correct answer data for evaluating the accuracy of the model output. In contrast, the first learning method can use the output of the micro risk estimation model 628 as training data for training the macro risk estimation model 623, making it possible to construct a highly accurate macro risk estimation model 623 in a relatively simple manner. Therefore, according to the first learning method, it is possible to improve the accuracy of the macro risk estimation model while operating a service that provides cooperation support information to each traffic participant.

[0157] (5) The second learning method prepares learning data by using input data for the macro risk estimation model 623 generated based on the first recognition information acquired in the first period and correct answer data for the output of the micro risk estimation model 628 generated based on the second recognition information acquired in the second period immediately after the first period, and further uses this learning data to learn an overall model that combines the macro risk estimation model 623 and the micro risk estimation model 628. Thus, according to the second learning method, the second recognition information acquired in the second period immediately after the first period can be used as data for evaluating the accuracy of the output of the overall model when the first recognition information is input, thereby improving the accuracy of the overall model that combines the macro risk estimation model 623 and the micro risk estimation model 628. Therefore, according to the traffic safety support system 1, it is possible to improve the accuracy of the overall model while operating a service that provides cooperation support information to each traffic participant.

[0158] Although one embodiment of the present invention has been described above, the present invention is not limited to this, and the detailed configuration may be modified as appropriate within the scope of the spirit of the present invention. [Explanation of symbols]

[0159] 1. Traffic safety support system 9...Targeted transportation area 2…Four-wheeled vehicle (mobile object, transportation participant) 3…Motorcycles (mobile objects, transportation participants) 4...Pedestrians (people, traffic participants) 6…Coordination support device 60...Target traffic area recognition unit (recognition means) 61...Driver information acquisition unit 62...Prediction unit (prediction means) 620...Area Risk Prediction Unit (Area Risk Prediction Means) 621...Statistical processing calculation unit 623…Macro Risk Estimation Model 624…High-risk area extraction section 625...Transport participant risk prediction unit (transport participant risk prediction means) 626...Monitoring area information extraction unit 628...Micro Risk Estimation Model 63...Health Notification Setting Unit 64...Risk Notification Setting Unit 65... Collaborative support information notification unit (transmission means) 67...Transportation Environment Database 68...Driving history database

Claims

1. recognition means for recognizing recognition objects including traffic participants as people or moving objects in a target traffic area and the traffic environment of each traffic participant, and for acquiring recognition information regarding these recognition objects; a prediction means for predicting a risk in the target traffic area based on the recognition information; a transmitting means for transmitting support information generated based on the recognition information and a prediction result by the prediction means to a support target determined from among a plurality of traffic participants in the target traffic area, The target traffic area is defined on a city, town or village basis and subdivided into multiple local areas, The prediction means an area risk prediction means for calculating a risk level for each of the local areas based on information obtained by performing statistical processing on the recognition information for the entire target traffic area, and extracting at least one of the local areas as a high-risk area based on the risk level; A traffic safety support system comprising: a traffic participant risk prediction means for predicting future risks of traffic participants in the high-risk area based on information related to the high-risk area among the recognition information for the entire target traffic area.

2. The traffic safety support system according to claim 1, characterized in that the traffic participant risk prediction means does not perform the prediction process for local areas among the plurality of local areas that have not been extracted as high-risk areas by the area risk prediction means.

3. The traffic safety support system described in claim 1 or 2, characterized in that the transmitting means transmits first support information generated based on the prediction results by the traffic participant risk prediction means to support targets among the multiple support targets that are located within the high-risk area, and transmits second support information generated based on the estimation results by the area risk prediction means to support targets that are located in a low-risk area outside the high-risk area.

4. A traffic safety support system as described in Claim 1, characterized in that the range of the local area is set wider than the collision mitigation brake operation range in which collision mitigation brake control is initiated in the moving body.

5. The recognition information includes dangerous area information in the target traffic area read by the recognition means from a database, 2. The traffic safety support system according to claim 1, wherein the area risk prediction means calculates the degree of risk for each of the local areas at a cycle shorter than the update cycle of the dangerous area information in the database.

6. A learning method for a traffic safety support system according to any one of claims 1 to 5, comprising: the area risk prediction means extracts the high-risk areas by utilizing a macro risk estimation model that receives input of information obtained by performing statistical processing on the recognition information and outputs a degree of risk for each of the plurality of local areas; the traffic participant risk prediction means predicts the future risk of the traffic participant in the high-risk area by utilizing a micro risk estimation model that outputs the future risk of the traffic participant in the local area when information related to a predetermined local area among the recognition information is input; preparing training data by using input data for the macro risk estimation model generated based on the recognition information and an output of the micro risk estimation model when the recognition information is input to the micro risk estimation model; and a step of training the macro risk estimation model using the training data.

7. A learning method for a traffic safety support system according to any one of claims 1 to 5, comprising: the area risk prediction means extracts the high-risk areas by utilizing a macro risk estimation model that receives input of information obtained by performing statistical processing on the recognition information and outputs a degree of risk for each of the plurality of local areas; the traffic participant risk prediction means predicts the future risk of the traffic participant in the high-risk area by utilizing a micro risk estimation model that outputs the future risk of the traffic participant in the local area when information related to a predetermined local area among the recognition information is input; preparing learning data by using input data for the macro risk estimation model generated based on first recognition information acquired during a predetermined first period, and correct answer data for the output of the micro risk estimation model generated based on second recognition information acquired during a second period immediately following the first period; a step of training the macro risk estimation model and the micro risk estimation model using the training data.

8. recognition means for recognizing recognition objects including traffic participants as people or moving objects in a target traffic area and the traffic environment of each traffic participant, and for acquiring recognition information regarding these recognition objects; a prediction means for predicting a risk in the target traffic area based on the recognition information; a transmitting means for transmitting support information generated based on the recognition information and a prediction result by the prediction means to a support target determined from among a plurality of traffic participants in the target traffic area, The target traffic area is subdivided into a plurality of local areas; The prediction means an area risk prediction means for extracting at least one of a plurality of local areas obtained by subdividing the target traffic area as a high-risk area by utilizing a macro risk estimation model that outputs a degree of risk for each of the plurality of local areas when information obtained by performing statistical processing on the recognition information is input; a traffic participant risk prediction means for predicting the future risk of traffic participants in the high-risk area by utilizing a micro risk estimation model that outputs the future risk of traffic participants in the high-risk area when information related to the high-risk area among the recognition information is input.

Images