Attention warning system and attention warning method

Abstract

To notify the risk of a vehicle coming into contact with a surrounding object in a manner easy for the driver of the vehicle to accept.SOLUTION: A warning system comprises: a notification unit that notifies a prescribed salience level to a driver by an HMI device when a risk index indicating the risk of collision with an object ahead falls within a prescribed range; a determination unit that determines a notification parameter set including a parameter that defines a notification issuance timing and the salience level; an action recording unit that records whether or not the driver has reacted to a notification made by using the notification parameter set; and a reaction information recording unit that stores reaction information in a storage device that is associated with a reaction ratio which is the ratio of the number of notifications that were reacted to the number of notifications issued, for each of the mutually different notification parameter sets. The determination unit calculates and determines, on the basis of reaction information, a notification parameter set with which the reaction ratio is estimated to be maximum.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to an attention calling system and an attention calling method for calling the attention of a vehicle driver. [Background technology]

[0002] In recent years, efforts to provide access to sustainable transport systems that take into consideration vulnerable transport participants have become more active. To achieve this, we are focusing on research and development into preventive safety technologies to further improve road safety and convenience.

[0003] Patent document 1 describes a moving object recognition system that calculates the risk of a moving object outside the vehicle colliding with the vehicle based on the relative approach angle of the moving object relative to the camera, which is estimated from multiple images taken from different viewpoints by an on-board camera, and alerts the driver to the risk.

[0004] Patent Document 2 describes a method of recognizing whether a driver is in a normal state or not from the driver's biometric information, and extracting the parts of the driving state data, such as the driver's accelerator operation, braking operation, and steering operation, that indicate the driver is driving in a normal state, to create a driver model. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-29604 [Patent Document 2] Japanese Patent Application Laid-Open No. 2007-272834 Summary of the Invention [Problem to be solved by the invention]

[0006] Incidentally, in preventive safety technology, it is important to notify the driver of the risk of contact with objects around the vehicle as early as possible so that the driver can reflect this in their driving behavior. When notifying the driver of the risk of contact early, the notification is made in a situation where the risk of contact has not yet increased sufficiently. Therefore, it is important that the stimuli (visual, auditory, and / or tactile stimuli) given to the driver for the notification are easy for the driver to recognize and accept, while minimizing the inconvenience to the driver.

[0007] To resolve this trade-off, it is conceivable to switch the mode of presented stimuli according to the driver's attention state and driving behavior. However, since human sensitivity to stimuli (for example, the degree of acceptability and discomfort) varies from person to person, simply considering the driver's attention state and driving behavior does not sufficiently consider the acceptability of notifications to the driver.

[0008] The present application aims to solve the above-mentioned problems by notifying the driver of the vehicle of the risk of contact between the vehicle and a surrounding object in a manner that is easy for the driver to accept, thereby contributing to the development of a sustainable transportation system. [Means for solving the problem]

[0009] One aspect of the present invention is an attention warning system comprising: a risk calculation unit that calculates a risk index indicating a collision risk between a vehicle and an object ahead of the vehicle based on the distance between the vehicle and the object; and a notification unit that issues a notification of a predetermined saliency level to a driver of the vehicle by an HMI device when the value of the calculated risk index falls within a predetermined value range, the attention warning system comprising: a determination unit that determines a notification parameter set including a parameter that specifies the timing of issuing the notification and a parameter that specifies the saliency level of the notification issued by the HMI device; and a notification unit that notifies the driver of the vehicle of the notification according to the issuing timing and the saliency level indicated by the determined notification parameter set. and a reaction information recording unit that calculates a reaction ratio, which is the ratio of the number of times the driver reacted to the notification to the number of times the notification was made, for each of the different notification parameter sets, and stores reaction information that associates the notification parameter set with the reaction ratio in a storage device.The attention warning system further includes: an action recording unit that records whether or not the driver's driving behavior includes a reaction behavior to the notification when the notification is made; and a reaction information recording unit that calculates a reaction ratio, which is the ratio of the number of times the driver reacted to the notification to the number of times the notification was made, for each of the different notification parameter sets, and stores reaction information that associates the notification parameter set with the reaction ratio in a storage device.The determination unit calculates and determines an optimal parameter set, which is the notification parameter set that is estimated to maximize the reaction ratio, based on the relationship between the notification parameter set indicated by the reaction information and the reaction ratio. According to another aspect of the present invention, the determination unit further determines an update period, which is the period until the next update of the determined optimal parameter set, and at least one data collection parameter set, which is the notification parameter set used exclusively for data collection, and during the update period, in cooperation with the notification unit, the behavior recording unit, and the reaction information recording unit, performs the notification selectively using the determined optimal parameter set and the data collection parameter set, and obtains the reaction information for the used optimal parameter set and data collection parameter set. According to another aspect of the present invention, the update period is defined as a period of a predetermined length of time or a period during which the notification unit issues the notification a predetermined number of times. According to another aspect of the present invention, the determination unit generates execution order information that defines the order in which the notification unit will execute the notification using the optimal parameter set and the notification using the data collection parameter set in multiple notifications within the update period, and the notification unit selects and uses the optimal parameter set and the data collection parameter set in accordance with the execution order information to execute the notification. According to another aspect of the present invention, a learning unit is provided that uses machine learning to train a driver response model to learn the relationship between the notification parameter set and the response ratio based on the response information, and the determination unit uses the driver response model to determine the optimal parameter set that is estimated to maximize the response ratio. According to another aspect of the present invention, the object in front of the vehicle is a preceding vehicle traveling in front of the vehicle, and the risk index is a time gap, which is the distance between the vehicle and the preceding vehicle divided by the speed of the vehicle. According to another aspect of the present invention, the presence or absence of the reactive behavior is determined based on the presence or absence of changes in the accelerator release operation, accelerator on operation, accelerator pedal depression amount, brake operation, and / or brake pedal depression amount in the vehicle. According to another aspect of the invention, the distance between the vehicle and the object in front of the vehicle is calculated based on information from a camera, a lidar, and / or a radar mounted on the vehicle. According to another aspect of the invention, the HMI device includes a tactile HMI device, an auditory HMI device, and / or a visual HMI device. According to another aspect of the present invention, the system includes a server device communicatively connected to the vehicle, the server device comprising the determination unit and the reaction information recording unit, and the vehicle comprising the risk calculation unit and the notification unit. According to another aspect of the present invention, the determination unit, the reaction information recording unit, the risk calculation unit, and the notification unit are provided in the vehicle. Another aspect of the present invention is an attention-drawing method executed by a computer of an attention-drawing system that notifies a driver of a vehicle using an HMI device, the method comprising: a risk calculation step of calculating a risk index indicating a collision risk between the vehicle and an object ahead of the vehicle based on the distance between the vehicle and the object; and a notification step of notifying the driver of the vehicle using an HMI device when the value of the calculated risk index falls within a predetermined value range, the method comprising: a determination step of determining a notification parameter set including a parameter that specifies the timing of issuing the notification and a parameter that specifies a saliency level of the notification issued by the HMI device; and a notification parameter set indicating the timing of issuing the notification and the saliency level of the notification indicated by the determined notification parameter set. This is an attention warning method further comprising: a behavior recording step for recording whether or not the driver's driving behavior involved a reaction to the notification when the notification was made in the notification step based on the saliency level; and a reaction information recording step for calculating a reaction ratio, which is the ratio of the number of times the driver made the reaction to the notification to the number of times the notification was made, for each of the different notification parameter sets, and storing reaction information associating the notification parameter set with the reaction ratio in a storage device.In the determination step, an optimal parameter set, which is the notification parameter set that is estimated to maximize the reaction ratio, is calculated and determined from the relationship between the various notification parameter sets and the reaction ratio. [Effects of the Invention]

[0010] According to the present invention, the risk of contact between a vehicle and a surrounding object can be notified in a manner that is easily acceptable to the driver of the vehicle. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a diagram showing the configuration of an attention calling system according to one embodiment of the present invention. [Figure 2] FIG. 2 is a diagram showing the configuration of a vehicle equipped with an attention calling device that constitutes the attention calling system. [Figure 3]FIG. 3 is a diagram showing the configuration of the interior of a vehicle in which an attention-attracting device is installed. [Figure 4] FIG. 4 is a diagram showing an example of a scene in which the attention calling system operates. [Figure 5] FIG. 5 is a diagram showing the configuration of an attention calling device that constitutes the attention calling system. [Figure 6] FIG. 6 is a diagram showing the configuration of an optimization server that constitutes the alert system. [Figure 7] FIG. 7 is an explanatory diagram for explaining the operation of the attention calling system. [Figure 8] FIG. 8 is a flowchart showing the procedure of the attention calling method executed by the attention calling system. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. Fig. 1 is a diagram showing the configuration of an attention calling system 1 according to one embodiment of the present invention. Fig. 2 is a diagram showing the configuration of a vehicle 2 in which the attention calling system 1 is installed, and Fig. 3 is a diagram showing the configuration of the interior of the vehicle 2.

[0013] [Configuration of the warning system] The attention warning system 1 includes an attention warning device 3 mounted on a vehicle 2, and an optimization server 4, which is a server device provided outside the vehicle 2. The attention warning device 3 and the optimization server 4 are connected to each other so as to be able to communicate with each other via a communication network 5 such as the Internet.

[0014] [Vehicle configuration] The vehicle 2 is equipped with an HMI device 6 (FIG. 5). In this embodiment, the HMI device 6 includes a tactile HMI device 6a that stimulates the driver's tactile sense, an auditory HMI device 6b that outputs sounds as auditory stimuli to the driver, and a visual HMI device 6c that conveys visual information to the driver. However, this is just one example, and the HMI device 6 may be any one of the tactile, auditory, and visual HMI devices, or may include any two of them.

[0015] The tactile HMI device 6a is, for example, an electric seat belt that is provided in the driver's seat 20 of the vehicle 2 and that changes the tension of the seat belt 21 worn by the driver to provide the driver with a tactile stimulus. However, the electric seat belt is just one example, and the tactile HMI device 6a may be any device that can provide a tactile stimulus to the driver. For example, the tactile HMI device 6a may be a vibration actuator that provides vibrations to the driver's hands via the steering wheel 22, or a massage seat that is provided as the driver's seat 20 and provides stimuli such as tapping, kneading, pressing, or vibrating to a body part of the driver, such as the back.

[0016] In this embodiment, the auditory HMI device 6b is a single speaker, or alternatively, the auditory HMI device 6b may be a speaker system made up of multiple speakers. In this embodiment, the visual HMI device 6c is a meter display device provided on the instrument panel 23 of the vehicle 2. However, the meter display device is just one example, and the visual HMI device 6c may be a general-purpose display device 24 that displays something other than a meter, or a head-up display (not shown) that displays an image on the windshield 25, or the like.

[0017] The vehicle 2 also includes an object detection device 7 that detects objects in the environment surrounding the vehicle 2, an exterior camera 8 (FIG. 5) that captures images of the environment, and a GPS device 9. The object detection device 7 may be, for example, a radar, a lidar, and / or a sonar that is provided at the front of the body of the vehicle 2 and detects objects in the environment ahead of the vehicle 2 (hereinafter also referred to as forward objects). In addition, the object detection device 7 may include multiple cameras, radars, lidars, and / or sonars that are distributed over the body of the vehicle 2 and detect objects in the entire environment surrounding the vehicle 2, including the left and right sides and rear of the vehicle 2.

[0018] The exterior camera 8 includes a front camera 8a mounted at the front of the vehicle 2 to capture images of the environment in front of the vehicle 2, a left side camera 8b and a right side camera 8c mounted on the left and right door mirrors 13a and 13b to capture images of the environments to the left and right of the vehicle 2, and a rear camera 8d mounted at the rear of the vehicle to capture images of the environment behind the vehicle 2.

[0019] Vehicle 2 further includes a vehicle control device 10 that detects the operation state of vehicle 2's steering controls such as the accelerator pedal, brake pedal, turn signal lights, steering wheel, etc., as well as the vehicle speed, acceleration, yaw rate, etc. of vehicle 2.

[0020] The vehicle 2 also includes an interior camera 11 that captures images of the interior of the vehicle, including the driver, and a room mirror 12 that enables the driver to view the area behind the vehicle 2. The room mirror 12 may be a rearview mirror that displays an image captured by a rear camera 8d that captures an image behind the vehicle 2.

[0021] [Outline of the warning system operation] In the attention warning system 1, the attention warning device 3 and the optimization server 4 work together to notify the driver of the vehicle 2 of a risk of collision between the vehicle 2 and a preceding object. In this embodiment, the preceding object is a preceding vehicle traveling ahead of the vehicle 2. For example, as shown in FIG. 4 , when the vehicle 2 traveling on a road 14 approaches a preceding vehicle 15 traveling ahead, the attention warning system 1 notifies the driver of the vehicle 2 by outputting a tactile stimulus, an auditory stimulus (e.g., sound), and / or visual information (e.g., light) from the HMI device 6.

[0022] Specifically, when the risk of collision between the vehicle 2 and the preceding vehicle 15 reaches a predetermined level, the attention warning device 3 notifies the driver of the vehicle 2 via the HMI device 6 in a manner defined by a notification parameter set (described later) received from the optimization server 4. The HMI device 6 used for this notification may be any one of a tactile HMI device 6a, an auditory HMI device 6b, and a visual HMI device 6c, or a combination of two or more thereof.

[0023] The optimization server 4 optimizes the notification parameter set so as to increase the driver's receptivity to the notifications made by the attention calling device 3 using the HMI device 6. Here, in this embodiment, the degree of receptivity is a response ratio, which is the ratio of the number of times the driver reacted to the notifications made by the HMI device 6 to the number of times the notifications were made.

[0024] The optimization server 4 executes a data collection operation for optimizing the notification parameter set, and an operational operation for performing notification using a more optimal notification parameter set.

[0025] In the data collection operation, the optimization server 4 generates a data collection parameter set, which is a notification parameter set used exclusively for data collection. The optimization server 4 generates various data collection parameter sets with different values, and obtains reaction information indicating the relationship between each of these data collection parameter sets and the reaction ratio.

[0026] The optimization server 4 uses the acquired response information to train the driver response model on the relationship between the notification parameter set and the response ratio through machine learning. For example, every time response information is acquired through the data collection operation, the optimization server 4 performs additional learning on the driver response model using the acquired response information.

[0027] In operation, the optimization server 4 uses a driver response model that has been additionally trained using the acquired response information to determine an optimal parameter set, which is a notification parameter set that is estimated to maximize the driver's response rate. The optimization server 4 transmits the determined optimal parameter set to the attention warning device 3, and the attention warning device 3 uses the optimal parameter set to issue a notification via the HMI device 6.

[0028] The optimization server 4 also acquires response information indicating the relationship between the optimal parameter set and the response ratio during operational operation. The optimization server 4 also uses the acquired response information to further perform additional learning of the driver model.

[0029] This enables the optimization server 4 to determine the optimal parameter set so that, as additional learning of the driver response model is repeated, notifications with a higher response rate (i.e., notifications that are more acceptable to the driver) can be provided.

[0030] [Configuration of the warning device] FIG. 5 is a diagram showing the configuration of the attention calling device 3 provided in the vehicle 2. As shown in FIG. The attention calling device 3 has a processor 30, a memory 31, and a first communicator 32. The first communicator 32 is a transceiver for the attention calling device 3 to communicate with the optimization server 4. The memory 31 is configured, for example, with a volatile and / or non-volatile semiconductor memory and / or a hard disk drive. A reaction log 37 is stored in the memory 31 by a behavior recording unit 36, which will be described later.

[0031] The processor 30 is, for example, a computer equipped with a CPU, etc. The processor 30 may have a configuration including a ROM in which a program is written, a RAM for temporarily storing data, etc. The processor 30 includes a risk calculation unit 33, a notification unit 34, a determination unit 35, and a behavior recording unit 36 ​​as functional elements or units.

[0032] These functional elements of the processor 30 are realized, for example, by the processor 30, which is a computer, executing a program 38 stored in the memory 31. The program 38 can be stored in any computer-readable storage medium. Alternatively, all or part of the functional elements of the processor 30 can be configured by hardware including one or more electronic circuit components.

[0033] The risk calculation unit 33 repeatedly calculates, for example, at predetermined time intervals, a risk index indicating the risk of collision between the vehicle 2 and an object ahead of the vehicle 2 (forward object) based on the distance between the vehicle 2 and the object ahead of the vehicle 2. In this embodiment, for example, the forward object is another vehicle traveling ahead of the vehicle 2, and the risk index is THW (Time to Headway), which is a value obtained by dividing the distance between the vehicle 2 and the forward object by the vehicle speed of the vehicle 2.

[0034] Here, the distance between the vehicle 2 and the preceding vehicle, which is a forward object, can be calculated based on information from an object detection device 7, which can be, for example, a radar, a lidar, and / or a sonar, and / or an exterior camera 8.

[0035] When the value of the risk index calculated by the risk calculation unit 33 falls within a predetermined value range in accordance with the notification parameter set (i.e., the above-mentioned data collection parameter set or the optimized parameter set) received from the optimization server 4, the notification unit 34 notifies the driver of the vehicle 2 of the existence of a collision risk via the HMI device 6. Furthermore, this notification is executed at a saliency level defined by the parameters of the notification parameter set.

[0036] In this embodiment, the notification parameter set includes values ​​of three parameters, as shown in equation (1): threshold a, which is a parameter that determines the timing of issuing a notification, and constants b and c, which are parameters that determine the salience level of the notification. Notification parameter set = (a, b, c) (1)

[0037] When the value th of THW, which is a risk index calculated by the risk calculation unit 33 based on the threshold a included in the notification parameter set, falls within a value range less than the threshold a, i.e., when equation (2) is satisfied, the notification unit 34 starts issuing a notification via the HMI device 6. This notification is executed at a saliency level SL given by equation (3) using constants b and c, which are parameters that define the saliency level of the notification and are included in the notification parameter set. th ​SL=cb×th (3)

[0038] Here, the salience level refers to the degree to which the stimulus output by the HMI device 6 attracts or attracts a person's attention. For example, if the tactile stimulus is output as tension in an electric seatbelt, the salience level of the tactile stimulus corresponds to the magnitude of the tension: the greater the tension, the higher the salience level of the tactile stimulus provided by the tension. Furthermore, for example, the salience level of an auditory stimulus corresponds to the intensity, frequency, repetition period, and / or intensity or frequency change period of the sound that is the auditory stimulus. The greater the intensity of the sound, the higher the frequency, the shorter the repetition period, and / or the shorter the intensity or frequency change period, the higher the salience level of the auditory stimulus provided by that sound.

[0039] Alternatively, if the tactile stimulus is vibration imparted to the driver's hands from a vibration actuator provided on the steering wheel 22 as described above, the salience level of the tactile stimulus corresponds to the vibration intensity, frequency, repetition period, and / or intensity or frequency change period of the vibration. The greater the vibration intensity, the higher the frequency, the shorter the repetition period, and / or the shorter the intensity or frequency change period, the higher the salience level of the tactile stimulus imparted by the vibration.

[0040] Alternatively, when the visual information is output as a graphic element such as a character or a figure displayed on a display device, the salience level of the visual information may be determined by the brightness, the period of brightness change, the period of blinking, or the color of the displayed graphic element. For example, the higher the brightness, the shorter the period of brightness change or blinking, or the closer the color is to a warm color from a cool color, the higher the salience level of the visual information.

[0041] In addition, the correspondence between the details of the aspects of the tactile stimuli, auditory stimuli, and / or visual information (such as the intensity, frequency, tension, brightness, brightness changes, etc. of the various stimuli or visual information described above) and the magnitude of the saliency level SL can be determined in advance and stored in memory 31. The notification unit 34 can refer to the correspondence information and notify the driver from the HMI device 6 using the tactile stimulation, auditory stimulation, and / or visual information in a manner corresponding to the magnitude of the determined salience level SL.

[0042] As can be seen from the above equation (3), the salience level of the notification increases as the value th of THW for the preceding vehicle decreases while the value th satisfies equation (2). This allows the stimulus output by, for example, the HMI device 6 to become stronger as the vehicle 2 approaches the preceding vehicle.

[0043] As described above, in this embodiment, there are two types of notification parameter sets: a data collection parameter set and an optimization parameter set. The notification unit 34 notifies the driver via the HMI device 6 using the data collection parameter set or the optimization parameter set in accordance with an instruction from the optimization server 4 (for example, in accordance with execution order information, which will be described later).

[0044] The judgment unit 35 judges whether or not the driver's steering behavior regarding the vehicle 2 includes a responsive behavior to the notification when the notification unit 34 issues a notification using the HMI device 6 based on the issuance timing and salience level indicated by the notification parameter set determined by the optimization server 4.

[0045] Specifically, the determination unit 35 acquires, as the reactive behavior of the driver in response to the notification, a change in the steering behavior of the vehicle 2 from when the notification unit 34 issues a notification via the HMI device 6 until a predetermined time (for example, two seconds from the start of issuance of the notification) has elapsed. In this embodiment, for example, the determination unit 35 determines that the reactive behavior has occurred when there is a predetermined amount of change in one or more of the steering behaviors, such as accelerator release operation, accelerator on operation, accelerator pedal depression amount, brake operation, and / or brake pedal depression amount.

[0046] The behavior recording unit 36 ​​records whether or not the driver's steering behavior regarding the vehicle 2 included any reactive behavior in response to the notification when the notification unit 34 issued the notification using the HMI device 6 based on the issuance timing and salience level indicated by the notification parameter set determined by the optimization server 4.

[0047] Specifically, based on the judgment result of the judgment unit 35 as to whether or not a reactive behavior has occurred, the behavior recording unit 36 ​​saves a reaction record in the reaction log 37 stored in the memory 31, which associates the notification parameter set used for the notification with the judgment result as to whether or not a reactive behavior has occurred in response to the notification.

[0048] Then, the behavior recording unit 36 ​​transmits the reaction record saved in the reaction log 37 to the optimization server 4 at the end of an update period, which will be described later.

[0049] [Optimization Server Configuration] FIG. 6 is a diagram showing the configuration of the optimization server 4. The optimization server 4 has a processor 40, a memory 41, and a second communicator 42. The second communicator 42 is a transceiver that enables the optimization server 4 to communicate with the attention warning device 3. The memory 41 is configured, for example, with a volatile and / or non-volatile semiconductor memory and / or a hard disk drive. The memory 41 stores reaction information 46 generated by a reaction information recording unit 43 (described later) and a driver reaction model 47 that is machine-learned by a learning unit 44.

[0050] The processor 40 is, for example, a computer equipped with a CPU, etc. The processor 40 may have a configuration including a ROM in which a program is written, a RAM for temporarily storing data, etc. The processor 40 includes a reaction information recording unit 43, a learning unit 44, and a determination unit 45 as functional elements or units.

[0051] These functional elements of processor 40 are realized, for example, by processor 40, which is a computer, executing program 48 stored in memory 41. Program 48 can be stored in any computer-readable storage medium. Alternatively, all or part of the functional elements of processor 40 can be configured by hardware including one or more electronic circuit components.

[0052] The reaction information recording unit 43 calculates, for each of the different notification parameter sets, a reaction ratio, which is the ratio of the number of times the driver reacted to the notification to the number of times the notification was made using that notification parameter set, based on the reaction record transmitted from the attention warning device 3 of the vehicle 2. The reaction information recording unit 43 also generates, for each of the different notification parameter sets, reaction information 46 that associates that notification parameter set with the reaction ratio for the notification using that notification parameter set. The reaction information recording unit 43 then stores the generated reaction information 46 in the memory 41.

[0053] The learning unit 44 uses the response information 46 stored in the memory 41 by the response information recording unit 43 to cause the driver response model 47 stored in the memory 41 to learn the relationship between the notification parameter set and the response ratio through machine learning. The driver response model 47 can be configured, for example, by a neural network according to conventional technology.

[0054] Every time the reaction information recording unit 43 stores new reaction information 46 based on the reaction record received from the attention warning device 3 of the vehicle 2 (for example, every time an update period described below ends), the learning unit 44 performs additional learning on the driver reaction model 47 using the newly stored reaction information 46. As a result, the driver reaction model 47 stored in the memory 41 is updated so that more accurate predictions can be made every time new reaction information is generated based on new reaction records received from the attention warning device 3.

[0055] The determination unit 45 determines a notification parameter set that defines the notification operation performed by the notification unit 34 of the attention warning device 3 provided in the vehicle 2 through the HMI device 6. As shown in the above formula (1), in this embodiment, the notification parameter set includes a threshold value a, which is a parameter that defines the timing of issuing a notification, and constants b and c, which are parameters that define the saliency level of the notification.

[0056] In this embodiment, the determination unit 45 determines, as the notification parameter sets, a data collection parameter set which is a notification parameter set for data collection, and an optimum parameter set which is a notification parameter set for operation. The data collection parameter set is a notification parameter set that is used exclusively to perform data collection operations in the attention warning device 3 provided in the vehicle 2. In the data collection operations, reaction information is collected using, for example, randomly, various data collection parameter sets with different values.

[0057] For example, the determination unit 45 randomly determines each of the data collection parameter sets. That is, the determination unit 45 randomly determines the value of the threshold a and the values ​​of the constants b and c that constitute the notification parameter set from predetermined value ranges, and generates a data collection parameter set that is configured by the determined threshold a and the constants b and c.

[0058] The optimal parameter set is a notification parameter set for causing the attention warning device 3 provided in the vehicle 2 to perform an operational operation that provides notification in a manner that is easily acceptable to the driver. Specifically, the determination unit 45 calculates a notification parameter set that is estimated to maximize the response rate from the relationship between various notification parameter sets and the response rate, and determines the calculated notification parameter set as the optimal parameter set. In this embodiment, a driver response model 47 that has learned the relationship between various notification parameter sets and response ratios is used to calculate the notification parameter set that is estimated to maximize the response ratio, and the calculated notification parameter set is determined as the optimal parameter set.

[0059] The determination unit 45 determines a predetermined period and selectively executes, during the predetermined period, a data collection operation using the determined data collection parameter set and an operation operation using the determined optimal parameter set in cooperation with the attention calling device 3. In this embodiment, the predetermined period is an update period until the determined optimal parameter set is next updated (i.e., a period until a new optimal parameter set is determined).

[0060] Then, during the update period, the determination unit 45 cooperates with the notification unit 34, the behavior recording unit 36, and the reaction information recording unit 43 to provide notifications selectively using the determined optimal parameter set and the multiple data collection parameter sets, and obtains reaction information about the optimal parameter set and the data collection parameter set used in these notifications.

[0061] Specifically, the determination unit 45 first determines an optimal parameter set using the driver response model 47, and then determines an update period, which is the period until the determined optimal parameter set is next updated. The determination unit 45 also generates multiple data collection parameter sets to be used in the update period. Here, the update period can be determined as a period of a predetermined length or a period during which the notification unit 34 issues a predetermined number of notifications.

[0062] The determination unit 45 further generates execution order information that determines the order in which the notification unit 34 of the warning device 3 will execute notifications using the above-mentioned optimal parameter set and notifications using any of the above-mentioned multiple data collection parameter sets in multiple notifications within the above-mentioned update period.

[0063] For example, the determination unit 45 determines the number of notifications using the data collection parameter set (hereinafter referred to as data collection notifications) to be executed within the update period, and the number of notifications using the optimal parameter set (hereinafter referred to as operation notifications).The determination unit 45 then generates execution order information that randomly determines the order of the data collection notifications and operation notifications.The execution order information may specify whether the nth (n>1) notification from the start of the update period should be a data collection notification or an operation notification, for example, such that the first notification from the start of the update period is a data collection notification, the second and third notifications are operation notifications, the fourth notification, and so on.

[0064] Here, the number of data collection notifications and the number of operational notifications determined when generating the execution order information can be determined, for example, according to the learning status (e.g., the number of reaction information used in learning up to that point) of driver response model 47. More specifically, the number of data collection notifications and the number of operational notifications can be determined so that the ratio of the number of data collection notifications to the number of operational notifications decreases as learning of driver response model 47 progresses (e.g., as the number of reaction information used in learning up to that point increases).

[0065] The determination unit 45 transmits the determined or generated optimal parameter set, data collection parameter set, execution order information, and information on the update period to the attention warning device 3. This starts the determined update period, and the notification unit 34 of the attention warning device 3 executes a data collection notification or an operation notification in accordance with the execution order information. Furthermore, the behavior recording unit 36 ​​of the attention warning device 3 stores a reaction record indicating whether or not the driver has performed a reaction behavior when a data collection notification or an operation notification is issued in a reaction log 37. Note that if the number of notifications occurring within the update period exceeds the number of notifications included in the execution order information, the notification unit 34 can return to the beginning of the execution order information and repeatedly use the execution order information.

[0066] When the update period expires, the behavior recording unit 36 ​​transmits the reaction record saved in the reaction log 37 to the optimization server 4. The reaction information recording unit 43 of the optimization server 4 receives the reaction record and generates reaction information. Furthermore, the learning unit 44 of the optimization server 4 performs additional learning of the driver reaction model 47 using the generated reaction information.

[0067] Next, the determination unit 45 determines a new optimal parameter set (i.e., updates the optimal parameter set) using the additionally trained driver response model 47, and determines or generates a new next update period, a new data collection parameter set, and new execution order information. The determination unit 45 transmits the new optimal parameter set, the new data collection parameter set, the new execution order information, and the new update period information to the attention warning device 3, and operation for the new update period begins.

[0068] Thereafter, the update period is repeated in the same manner as described above, and a mixture of data collection notifications and operational notifications is executed in each update period (i.e., data collection operations and operational operations are repeated in each update period).

[0069] [Warning system operation] Fig. 7 is an explanatory diagram for explaining the operation of the attention warning system 1 having the above-described configuration. In Fig. 7, the horizontal axis represents time, and the vertical axis represents the value th of the THW between the vehicle 2 and the preceding vehicle. Furthermore, line 60 represents the change over time in the value th of the THW calculated by the risk calculation unit 33 of the attention warning device 3. Although Fig. 7 shows two update periods -1 and -2 as an example, in reality, a plurality of update periods are repeated along the time axis.

[0070] At the time up to time t1, the determination unit 45 of the optimization server 4 determines an optimal parameter set (ao1, bo1, co1) using the driver response model 47 trained using the response information acquired up to that point, and determines the update period until the next new optimal parameter set is determined as update period-1. For example, update period-1 is determined as a period having a predetermined length of time.

[0071] The determination unit 45 also uses random values ​​to determine, for example, three data collection parameter sets (a1, b1, c1), (a2, b2, c2), and (a3, b3, c3). Furthermore, the determination unit 45 generates execution order information-1 that randomly determines the order in which the three data collection parameter sets and the optimal parameter set are used within the update period-1.

[0072] Then, the determination unit 45 transmits the determined or generated optimal parameter set, data collection parameter set, execution order information-1, and information on update period-1 to the attention calling device 3 at time t1. As a result, update period-1 starts, for example, from time t1.

[0073] The notification unit 34 of the attention-calling device 3 issues a notification in the order defined by the execution order information-1 received from the optimization server 4, using any one of the three data collection parameter sets and the optimum parameter set.

[0074] For example, in accordance with execution order information-1, the notification unit 34 first uses the data collection parameter set (a1, b1, c1) to issue a notification indicating the existence of a risk of collision with a preceding vehicle at time t2 when the value th of THW crosses the threshold a1 and decreases, via the HMI device 6, at a saliency level defined by the values ​​b1 and c1 of the data collection parameter set. In addition, the behavior recording unit 36 ​​of the attention warning device 3 saves a record of the driver's response to the notification in a response log in the memory 31.

[0075] Next, in accordance with execution order information-1, the notification unit 34 uses the data collection parameter set (a2, b2, c2) to issue a notification indicating the existence of a risk of collision with a preceding vehicle at the saliency level specified by the values ​​b2 and c2 of the data collection parameter set at time t3 when the value th of THW crosses the threshold a2 and decreases.

[0076] Next, the notification unit 34 sequentially uses the optimization parameter set (ao1, bo1, co1), the data collection parameter set (a3, b3, c3), and the optimization parameter set (ao1, bo1, co1) in the same manner as above to issue a notification at the saliency level defined by the corresponding notification parameter set at times t4, t5, and t6 when the value th of THW crosses the thresholds ao1, a3, and ao1 and drops, respectively. The behavior recording unit 36 ​​of the attention warning device 3 saves a record of the driver's reaction in the reaction log in the memory 31 each time a notification is issued.

[0077] Thereafter, the notification unit 34 continues to issue notifications according to the execution order information-1, and the behavior recording unit 36 ​​stores records of the driver's reactions to those notifications in the reaction log of the memory 31. Then, at time t7, when the time specified as the update period-1 has elapsed since time t1, the update period-1 ends. In response to this, the behavior recording unit 36 ​​of the attention warning device 3 transmits the reaction records stored in the reaction log 37 of the memory 31 to the optimization server 4.

[0078] The reaction information recording unit 43 of the optimization server 4 calculates the reaction ratio in each notification of the three data collection parameter sets and the optimal parameter set based on the reaction record, and generates reaction information 46 that associates each of the three data collection parameter sets and the optimal parameter set with the calculated reaction ratio.

[0079] Furthermore, the learning unit 44 of the optimization server 4 uses the generated response information 46 to perform additional learning of the driver response model 47. Next, the determination unit 45 of the optimization server uses the driver response model 47 to determine the optimal parameter set (ao2, bo2, co2) that is estimated to maximize the response ratio, and then determines the update period until the next optimal parameter set is newly determined as update period-2. For example, update period-2 is determined to be a period having a predetermined length of time.

[0080] The determination unit 45 also uses random values ​​to determine, for example, three data collection parameter sets (a4, b4, c4), (a5, b5, c5), and (a6, b6, c6). Furthermore, the determination unit 45 generates execution order information-2 within the update period-2.

[0081] Then, the determination unit 45 transmits the determined or generated optimal parameter set, data collection parameter set, execution order information-2, and information on update period-2 to the attention calling device 3 at time t8. As a result, update period-2 starts, for example, from time t8.

[0082] In update period-2, similar to the operation in update period-1 described above, the notification unit 34 uses one of the three data collection parameter sets (a4, b4, c4), (a5, b5, c5), and (a6, b6, c6) and the optimal parameter set (ao2, bo2, co2) in the order specified by the execution order information-2 received from the optimization server 4. At times t9, t10, t11, t12, and t13 when the value th of THW crosses and drops below the thresholds a4, a5, a6, or ao2 of the respective notification parameter sets, the notification unit 34 issues a notification indicating the existence of a collision risk with the preceding vehicle via the HMI device 6 at the saliency level specified by the corresponding notification parameter set. In addition, the behavior recording unit 36 ​​of the attention alert device 3 stores a record of the driver's response in the response log in the memory 31 each time a notification is issued.

[0083] Then, at time t14, when the time defined as update period-2 has elapsed since time t8, update period-2 ends. Thereafter, the optimization server 4 repeatedly determines new update periods in the same manner as above, and repeats the same operations as above.

[0084] [Operation flow of the warning system] Next, we will explain the operation procedure of the attention calling system 1. Figure 8 is a flow diagram showing the operation procedure of the attention calling method executed by the computer of the attention calling system 1 (i.e., the processor 30 of the attention calling device 3 and the processor 40 of the optimization server 4). The process shown in Figure 8 is executed repeatedly.

[0085] When the process starts, first, the determination unit 45 of the optimization server 4 uses the driver response model 47 to calculate and determine an optimal parameter set, which is a notification parameter set that is estimated to maximize the driver's response rate (S100). The determination unit 45 also determines an update period until the next update of the determined optimal parameter set, and multiple data collection parameter sets, which are notification parameter sets to be used exclusively for data collection during the update period (S102). The determination unit 45 transmits information on the determined or generated optimal parameter set, data collection parameter set, and update period to the attention warning device 3. This starts the update period in the attention warning device 3. When transmitting the optimal parameter set, etc., the determination unit 45 may generate execution order information that defines the order in which the optimal parameter set and the data collection parameter set will be used during the update period, and transmit this information to the attention warning device 3.

[0086] Next, the notification unit 34 of the attention-calling device 3 executes either a data collection notification, which is a notification using a parameter set for data collection, or an operational notification, which is a notification using an optimal parameter set, according to, for example, the execution order information (S104). The notification is executed by the HMI device 6 at a saliency level SL given by equation (3) when the value th of the THW between the vehicle 2 and the preceding vehicle satisfies the above equation (2) according to each notification parameter set.

[0087] Then, every time a notification is output from the HMI device 6, the behavior recording unit 36 ​​of the attention calling device 3 saves and stores a reaction record, which is a record of whether or not the driver has performed a reaction behavior, in the reaction log 37 of the memory 31 based on the determination result of the determination unit 35 as to whether or not the driver has performed a reaction behavior (S104). Here, whether or not the driver has performed a reaction behavior in response to the notification is determined based on whether or not the driver has performed a reaction behavior within a first predetermined time (for example, 2 seconds) from the start of the notification.

[0088] If no reaction behavior is observed within the first predetermined time, the notification unit 34 may change the notification parameter set and repeat the notification until the driver performs the reaction behavior. For example, the notification unit 34 determines whether the driver performs the reaction behavior within a second predetermined time (e.g., 10 seconds) from the notification (S108). If the driver does not perform the reaction behavior within the second predetermined time (S108, NO), the notification unit 34 may return to step S104 and repeat the process.

[0089] On the other hand, if the driver performs a reaction behavior within the second predetermined time (S108, YES), the notification unit 34 determines whether the update period has ended (S110). If the update period has not ended (S110, NO), the notification unit 34 returns to step S104 and repeats the process.

[0090] On the other hand, when the update period has ended (S110, YES), the behavior recording unit 36 ​​transmits the reaction records collected during that update period and stored in the reaction log 37 to the optimization server 4. The reaction information recording unit 43 of the optimization server 4 generates reaction information 46 based on the received reaction records and stores it in the memory 41 (S112). The learning unit 44 uses the reaction information 46 stored in the memory 41 to perform additional learning on the driver reaction model 47, and this process ends.

[0091] After this process is completed, the alert system 1 restarts the process from step S100 and repeats the above operations.

[0092] [Other embodiments] In the above-described embodiment, the risk index indicating the risk of collision between vehicle 2 and a preceding vehicle, which is a forward object, is THW, but other values ​​such as the distance between vehicle 2 and the forward object or TTC (time to collision) may also be used.

[0093] Furthermore, in the above-described embodiment, a data collection parameter set is determined for each update period, but some or all of the data collection parameter sets may be used across multiple update periods.

[0094] In the above-described embodiment, the determination unit 45 randomly determines the execution order of the data collection parameter sets and the optimal parameter sets to generate execution order information, and the notification unit 34 issues a notification using one of the multiple data collection parameter sets and the optimal parameter set in accordance with the execution order information. Alternatively, the determination unit 45 may not generate the execution order information, and the notification unit 34 itself may randomly determine which of the multiple data collection parameter sets and the optimal parameter set to use and issue a notification.

[0095] Furthermore, in the above-described embodiment, the determination unit 45 determines a plurality of data collection parameter sets for each update period, but may also determine at least one data collection parameter set for each update period.

[0096] In the above-described embodiment, the determination unit 45 of the optimization server 4 selectively executes a mixture of data collection operations using a data collection parameter set and operational operations using an optimal parameter set during an update period. However, the data collection operations and operational operations do not necessarily need to be selectively executed during an update period; the data collection operations and operational operations may be executed in separate periods. The period may be a predetermined length of time or a period during which the notification unit 34 issues a predetermined number of notifications.

[0097] Furthermore, in the above-described embodiment, the determination unit 45 determines the optimal parameter set using the driver response model 47 that has undergone machine learning based on the response information generated by the response information recording unit 43. However, using the driver response model 47 is only one method for determining the optimal parameter set, and the determination unit 45 can use any other method that does not use the driver response model 47 to determine the optimal parameter set from the relationship between the notification parameter set indicated by the response information and the response ratio.

[0098] Such other techniques may be, for example, a technique such as multivariate analysis (e.g., analysis of covariance) of each parameter (e.g., threshold a and constants b and c) constituting the notification parameter set and the response rate.

[0099] In the above-described embodiment, the determiner 45 determines the parameter set for data collection by randomly determining the values ​​of each parameter constituting the notification parameter set, but the method for determining the parameter set for data collection is not limited to this method (random value determination). For example, the determiner 45 may use a genetic algorithm to perform operations such as selection, crossover, and / or mutation on the value sets of each parameter constituting one notification parameter set to generate a new generation of the value sets, and determine the notification parameter set having the value sets of the generated generation as the parameter set for data collection.

[0100] Alternatively, the determination unit 45 may use a probabilistic method such as optimal arm identification based on the response information obtained up to that point to determine the notification parameter set that is estimated to result in a greater reward response ratio as the data collection parameter set and / or the optimal parameter set.

[0101] Furthermore, in the above-described embodiment, there is one driver response model 47, but if multiple drivers take turns driving the vehicle 2, a driver response model 47 may be created for each driver.

[0102] Furthermore, in the above-described embodiment, the attention warning system 1 is configured with the attention warning device 3 mounted on the vehicle 2 and the optimization server 4 placed outside the vehicle 2, but is not limited to this configuration. All of the functional elements of the optimization server 4 and the attention warning device 3 described above may be provided in the vehicle 2. For example, the reaction information recording unit 43, learning unit 44, and determination unit 45 provided in the optimization server 4 may be provided in the processor 30 of the attention warning device 3, and the driver reaction model 47 may be stored in the memory 31, so that the attention warning system 1 is realized by the attention warning device 3 alone.

[0103] Alternatively, all of the functional elements of the attention calling device 3 and the optimization server 4 in the above-described embodiment may be provided in a server device external to the vehicle 2. For example, the risk calculation unit 33, the notification unit 34, the determination unit 35, and the behavior recording unit 36 ​​provided in the attention calling device 3 may be provided in the processor 40 of the optimization server 4. In this case, the optimization server 4 and each device of the vehicle 2 (such as the HMI device 6, the object detection device 7, and the exterior camera 8) may be communicatively connected.

[0104] The present invention is not limited to the configurations of the above-described embodiments, and can be embodied in various forms without departing from the spirit and scope of the present invention.

[0105] [Configuration supported by the above embodiment] The above-described embodiment supports the following configurations.

[0106] (Configuration 1) An attention warning system including: a risk calculation unit that calculates a risk index indicating a collision risk between a vehicle and an object ahead of the vehicle based on the distance between the vehicle and the object; and a notification unit that issues a notification of a predetermined saliency level to a driver of the vehicle by an HMI device when the value of the calculated risk index falls within a predetermined value range, wherein the system includes a determination unit that determines a notification parameter set including a parameter that specifies the timing of issuing the notification and a parameter that specifies the saliency level of the notification issued by the HMI device; and the notification unit that issues the notification according to the issuing timing and the saliency level indicated by the determined notification parameter set. and a response information recording unit that calculates, for each of the different notification parameter sets, a response ratio that is the ratio of the number of times the driver performed the reaction behavior in response to the notification to the number of times the notification was made, and stores response information that associates the notification parameter sets with the response ratios in a storage device, wherein the determination unit calculates and determines an optimal parameter set that is the notification parameter set that is estimated to maximize the response ratio based on the relationship between the notification parameter set indicated by the response information and the response ratio. According to the warning system of configuration 1, the degree of receptivity of the driver to notifications regarding the risk of contact between the vehicle and surrounding objects is evaluated as the driver's response rate to the notification, and a notification parameter set that specifies the manner of notification is calculated so that the response rate is maximized, thereby making it possible to notify the driver of the vehicle of the risk of contact between the vehicle and surrounding objects in a manner that is easy for the driver to accept.

[0107] (Configuration 2) The determination unit further determines an update period, which is the period until the next update of the determined optimal parameter set, and at least one data collection parameter set, which is the notification parameter set used exclusively for data collection, and during the update period, in cooperation with the notification unit, the behavior recording unit, and the reaction information recording unit, performs the notification selectively using the determined optimal parameter set and the data collection parameter set, and acquires the reaction information about the used optimal parameter set and data collection parameter set, in the alert system described in Configuration 1. According to the attention warning system of configuration 2, the data collection operation and the operational operation are performed separately within the same update period, so that the driver's discomfort regarding the notification using the parameter set for data collection can be reduced.

[0108] (Configuration 3) The alert system according to Configuration 2, wherein the update period is determined as a period of a predetermined length of time or a period during which the notification unit issues the notification a predetermined number of times. According to the alert system of configuration 3, the amount of reaction record data collected during an update period can be controlled by the length of the update period or the number of notifications.

[0109] (Configuration 4) The determination unit generates execution order information that determines the order in which the notification unit will execute the notification using the optimal parameter set and the notification using the data collection parameter set in multiple notifications within the update period, and the notification unit selects and uses the optimal parameter set and the data collection parameter set in accordance with the execution order information to execute the notification, in an alert system described in configuration 2 or 3. According to the alert system of configuration 4, the notification unit can select a notification parameter set in accordance with the execution order information, which simplifies the processing.

[0110] (Configuration 5) An attention warning system described in any of configurations 1 to 4, comprising a learning unit that uses machine learning to train a driver response model to learn the relationship between the notification parameter set and the response ratio based on the response information, and the determination unit uses the driver response model to determine the optimal parameter set that is estimated to maximize the response ratio. According to the alert system of configuration 5, it is possible to easily determine the optimal parameter set that is estimated to maximize the response ratio.

[0111] (Configuration 6) An attention warning system described in any of configurations 1 to 5, wherein the object in front of the vehicle is a preceding vehicle traveling in front of the vehicle, and the risk index is a time gap, which is the value obtained by dividing the inter-vehicle distance between the vehicle and the preceding vehicle by the vehicle speed of the vehicle. According to the alert system of configuration 6, it is possible to easily detect the degree of risk of collision between the vehicle and the preceding vehicle and notify the driver of the existence of the risk of collision.

[0112] (Configuration 7) An attention warning system described in any one of configurations 1 to 6, wherein the presence or absence of the reactive behavior is determined based on the presence or absence of changes in the accelerator off operation, accelerator on operation, accelerator pedal depression amount, brake operation, and / or brake pedal depression amount in the vehicle. According to the warning system of configuration 7, whether or not the driver has responded to the notification of the collision risk can be easily determined from whether or not the driver has performed an acceleration or deceleration operation on the vehicle.

[0113] (Configuration 8) An attention warning system described in any of configurations 1 to 7, wherein the distance between the vehicle and the object in front of the vehicle is calculated based on information from a camera, lidar, and / or radar provided on the vehicle. According to the attention warning system of configuration 8, the distance between the vehicle and the object ahead can be calculated by various sensors.

[0114] (Configuration 9) An attention warning system according to any one of configurations 1 to 8, wherein the HMI device includes a tactile HMI device, an auditory HMI device, and / or a visual HMI device. According to the attention warning system of configuration 9, various types of notifications can be given to the driver using various HMI devices.

[0115] (Configuration 10) An alert system described in any one of configurations 1 to 9, including a server device communicatively connected to the vehicle, the server device comprising the determination unit and the reaction information recording unit, and the vehicle comprising the risk calculation unit and the notification unit. According to the alert system of configuration 10, processing can be distributed between the server device and the in-vehicle device, which reduces the load on the in-vehicle device and reduces vehicle costs, for example.

[0116] (Configuration 11) The attention warning system according to any one of configurations 1 to 9, wherein the determination unit, the reaction information recording unit, the risk calculation unit, and the notification unit are provided in the vehicle. According to the attention warning system of configuration 11, the entire attention warning system can be mounted on the vehicle, allowing for more stable processing that is not affected by factors such as congestion of communication lines outside the vehicle.

[0117] (Configuration 12) An attention warning method executed by a computer of an attention warning system that notifies a vehicle driver using an HMI device, comprising: a risk calculation step of calculating a risk index indicating a collision risk between the vehicle and an object in front of the vehicle based on the distance between the vehicle and the object; and a notification step of notifying the driver of the vehicle using the HMI device when the value of the calculated risk index falls within a predetermined value range, and a determination step of determining a notification parameter set including a parameter that specifies the timing of issuing the notification and a parameter that specifies the prominence level of the notification issued by the HMI device; and a notification step of determining the timing of issuing the notification and the prominence level indicated by the determined notification parameter set. The method further comprises: a behavior recording step of recording whether or not the driver's driving behavior involved a reaction to the notification when the notification was made in the notification step based on the significance level; and a reaction information recording step of calculating, for each of the different notification parameter sets, a reaction ratio, which is the ratio of the number of times the driver performed the reaction to the notification to the number of times the notification was made, and storing reaction information associating the notification parameter sets with the reaction ratios in a storage device; and in the determination step, an optimal parameter set, which is the notification parameter set that is estimated to maximize the reaction ratio, is calculated and determined from the relationship between the various notification parameter sets and the reaction ratios. According to the warning method of configuration 12, the degree of receptivity of the driver to notifications regarding the risk of contact between the vehicle and surrounding objects is evaluated as the driver's response rate to the notification, and a notification parameter set that specifies the manner of notification is calculated so that the response rate is maximized, thereby making it possible to notify the driver of the vehicle of the risk of contact between the vehicle and surrounding objects in a manner that is easy for the driver to accept. [Explanation of symbols]

[0118] 1...attention system, 2...vehicle, 3...attention device, 4...optimization server, 5...communication network, 6...HMI device, 6a...tactile HMI device, 6b...auditory HMI device, 6c...visual HMI device, 7...object detection device, 8...exterior camera, 8a...front camera, 8b...left side camera, 8c...right side camera, 8d...rear camera, 9...GPS device, 10...vehicle control device, 11...interior camera, 12...room mirror, 13a, 13b...door mirror, 14...road, 15...leading vehicle, 20...driver's seat , 21...seat belt, 22...steering wheel, 23...instrument panel, 24...display device, 25...windshield, 30...processor, 31...memory, 32...first communication device, 33...risk calculation unit, 34...notification unit, 35...judgment unit, 36...behavior recording unit, 37...reaction log, 38, 48...program, 40...processor, 41...memory, 42...second communication device, 43...reaction information recording unit, 44...learning unit, 45...decision unit, 46...reaction information, 47...driver reaction model, 60...line.

Claims

1. a risk calculation unit that calculates a risk index indicating a collision risk between a vehicle and an object in front of the vehicle based on a distance between the vehicle and the object; a notification unit that notifies the driver of the vehicle of a predetermined saliency level by an HMI device when the calculated risk index value falls within a predetermined value range; An attention warning system comprising: a determination unit that determines a notification parameter set including a parameter that defines a timing of issuing the notification and a parameter that defines a salience level of the notification issued by the HMI device; a behavior recording unit that records whether or not the driver's driving behavior includes a reaction behavior to the notification when the notification unit issues the notification at the issuance timing and the salience level indicated by the determined notification parameter set; and a reaction information recording unit that calculates a reaction ratio, which is a ratio of the number of times the driver has performed the reaction behavior in response to the notification to the number of times the notification has been made, for each of the different notification parameter sets, and stores reaction information that associates the notification parameter set with the reaction ratio in a storage device; Further provided with the determination unit calculates and determines an optimal parameter set, which is the notification parameter set that is estimated to maximize the reaction rate, based on a relationship between the notification parameter set indicated by the reaction information and the reaction rate. Attention warning system.

2. The determination unit Further determining an update period, which is a period until the determined optimal parameter set is next updated, and at least one data collection parameter set, which is the notification parameter set used exclusively for data collection; during the update period, in cooperation with the notification unit, the behavior recording unit, and the reaction information recording unit, the notification is made by selectively using the determined optimal parameter set and a parameter set for data collection, and the reaction information for the used optimal parameter set and parameter set for data collection is acquired. The warning system according to claim 1 .

3. The update period is determined as a period of a predetermined length of time or a period during which the notification unit issues the notification a predetermined number of times. The warning system according to claim 2 .

4. the determination unit generates execution order information that defines an order in which the notification unit should execute the notification using the optimal parameter set and the notification using the data collection parameter set in the multiple notifications within the update period; the notification unit selects and uses the optimal parameter set and the data collection parameter set in accordance with the execution order information to make the notification. The warning system according to claim 2 .

5. a learning unit that causes a driver response model to learn the relationship between the notification parameter set and the response ratio by machine learning based on the response information; the determination unit determines the optimal parameter set that is estimated to maximize the reaction ratio using the driver reaction model. The warning system according to claim 1 .

6. the object in front of the vehicle is a preceding vehicle traveling in front of the vehicle, The risk index is a time gap between the vehicle and the preceding vehicle, which is a value obtained by dividing the distance between the vehicle and the preceding vehicle by the speed of the vehicle. The warning system according to claim 1 .

7. The presence or absence of the reaction behavior is determined based on the presence or absence of a change in an accelerator-off operation, an accelerator-on operation, an accelerator pedal depression amount, a brake operation, and / or a brake pedal depression amount in the vehicle. The warning system according to claim 1 .

8. The distance between the vehicle and the object in front of the vehicle is calculated based on information from a camera, a lidar, and / or a radar provided on the vehicle. The warning system according to claim 1 .

9. The HMI device includes a tactile HMI device, an auditory HMI device, and / or a visual HMI device. The warning system according to claim 1 .

10. a server device communicably connected to the vehicle; the server device includes the determination unit and the reaction information recording unit; The vehicle includes the risk calculation unit and the notification unit. The warning system according to any one of claims 1 to 9.

11. The determination unit, the reaction information recording unit, the risk calculation unit, and the notification unit are provided in the vehicle. The warning system according to any one of claims 1 to 9.

12. An attention calling method executed by a computer of an attention calling system that notifies a vehicle driver by an HMI device, comprising: a risk calculation step of calculating a risk index indicating a collision risk between the vehicle and an object in front of the vehicle based on a distance between the vehicle and the object; a notification step of notifying the driver of the vehicle by an HMI device when the calculated risk index value falls within a predetermined value range; and determining a notification parameter set including a parameter defining the timing of the notification and a parameter defining the salience level of the notification to be issued by the HMI device; a behavior recording step of recording whether or not the driver's driving behavior includes a reaction behavior to the notification when the notification is made in the notification step based on the issuance timing and the salience level indicated by the determined notification parameter set; a reaction information recording step of calculating a reaction ratio, which is a ratio of the number of times the driver has performed the reaction behavior in response to the notification to the number of times the notification has been made, for each of the different notification parameter sets, and storing reaction information in a storage device that associates the notification parameter set with the reaction ratio; Further provided with In the determination step, an optimal parameter set is calculated and determined, which is the notification parameter set that is estimated to maximize the response rate, based on the relationship between various notification parameter sets and the response rate. How to get attention.

Images