JP6772775B2 - Driving support device and driving support method - Google Patents

Description

The present invention relates to a device that assists the driving of a vehicle.

As a method for detecting dangerous driving such as involuntary driving, for example, a method as described in Patent Document 1 has been proposed. In the device described in Patent Document 1, the statistical distribution of the gaze information of the vehicle driver during casual driving is defined in advance, and the similarity with the statistical distribution of the gaze information of the vehicle driver obtained at the time of evaluation is evaluated. By doing so, the possibility of casual driving is estimated.

Further, as a method for detecting dangerous driving, for example, a method as described in Patent Document 2 has been proposed. In the apparatus described in Patent Document 2, the line-of-sight direction of the vehicle driver is detected for a predetermined time, line-of-sight distribution information indicating the distribution in the line-of-sight direction is generated, and the predetermined position of the distribution indicated by the obtained line-of-sight distribution information is determined. When the vehicle is separated from the standard position in the line-of-sight direction in a non-involuntary driving state by a predetermined threshold value or more, which is indicated by the reference information defined in advance, it is judged to be dangerous driving.

Further, as a method for detecting dangerous driving, for example, a method as described in Patent Document 3 has been proposed. The device described in Patent Document 3 obtains the heartbeat interval RRI from the electrocardiogram (R waveform) of the vehicle driver, stores the heartbeat interval RRI at each time, and detects an abnormal pattern from the time series data, which is dangerous. Driving is detected.

Further, as a method for detecting dangerous driving, for example, a method as described in Patent Document 4 has been proposed. In the device described in Patent Document 4, the risk level is defined based on the relationship such as vehicle position, road condition, vehicle behavior, and mental state of the vehicle driver, and the risk level is equal to or higher than a predetermined threshold value. In some cases, it is determined that the driving is dangerous.

Japanese Unexamined Patent Publication No. 08-178712 Japanese Unexamined Patent Publication No. 2008-243031 Japanese Unexamined Patent Publication No. 2010-162282 Japanese Unexamined Patent Publication No. 2014-206906 Japanese Unexamined Patent Publication No. 2008-058459 Japanese Unexamined Patent Publication No. 2008-139553 Japanese Unexamined Patent Publication No. 03-165737

Driver's fully automatic line-of-sight measurement method using an in-vehicle camera Vol.2013-CVIM-185 No.39 Haruko Takada, Mikio Takada, Ai Kanayama, Significance of LF and HF components of heart rate variability frequency analysis and heart rate variability coefficient: Autonomic nervous function evaluation comprehensive examination by acceleration pulse wave measurement system, Vol. 32, No. 6, pp. 504 -512 (2005). Masao Yamanaka, Shunta Saito, Yoshimitsu Aoki, Behavior recognition method for occupants in the vehicle based on the global structure of the position of human body parts, Image Sensing Symposium 2016 Hiroshi Morinaga, Yasushi Hanatsuka, Yasumichi Wakao, Katsuhiro Kobayashi, Development of Road Surface Condition Judgment System Using Tire Sensor, Automotive Technology 65 (12), 97-98, 2011-12-01s Kanamori, T., Hido, S., & Sugiyama, M., A least-squares approach to direct importance estimation, Journal of Machine Learning Research, vol.10 (Jul.), Pp.1391-1445, 2009. Yamada, M., Suzuki, T., Kanamori, T., Hachiya, H., & Sugiyama, M., Relative density-ratio estimation for robust distribution comparison, Neural Computation, vol.25, no.5, pp.1324 -1370, 2013. Sugiyama, M., Kanamori, T., Suzuki, T., du Plessis, MC, Liu, S., & Takeuchi, I., Density difference estimation, IEICE Technical Report IBISML2012-8, pp.49-56, Kyoto, Japan, Jun. 19-20, 2012.

However, the techniques disclosed in the above-mentioned patent documents have the following problems.

For example, in the invention described in Patent Document 1, the estimation accuracy is lower than that which is not defined in advance as the statistical distribution of the gaze information of the vehicle driver during casual driving.

Further, in the invention described in Patent Document 2, the difference in the line-of-sight distribution information between the state of involuntary driving and the state of not involuntarily driving is not a standard position, but a difference in statistical distribution (for example, kurtosis, skewness, dispersion). Etc.), the estimation accuracy will decrease.

Further, in the invention described in Patent Document 3, the invention is caused by vehicle behavior (for example, speed, acceleration, etc.) or external environment (for example, weather, road surface condition, etc.), which is not expressed as a vital change of the vehicle driver. On the other hand, the estimation accuracy is lowered.

Further, in the invention described in Patent Document 4, the estimation accuracy is lowered for a potential danger (a situation in which the above-mentioned plurality of factors are complicatedly involved) that is difficult for a person to define in advance.

The present invention has been made in consideration of the above problems, and an object of the present invention is to provide a technique for accurately detecting a dangerous driving state.

The driving support device according to the present invention is a driving support device that detects a dangerous driving state of a vehicle. The dangerous driving state refers to a state in which safe driving is not possible, for example, a state in which involuntary driving, dozing driving, overworked driving, or the like occurs. Further, the dangerous driving state does not have to be caused by the driver himself / herself. For example, it also includes a driving state in an environment where the vehicle cannot be operated safely due to the state of the vehicle or the environment.

The driving support device according to the present invention includes a occupant information acquisition means for acquiring occupant information which is information about a vehicle driver, an environmental information acquisition means for acquiring environmental information which is information about the external environment of the vehicle, and the vehicle. The first feature vector, which is a feature vector corresponding to the driving to be evaluated, based on the vehicle information acquisition means for acquiring the vehicle information which is information about the vehicle, the occupant information, the environmental information, and the vehicle information. The first feature vector acquisition means for acquiring the above, the second feature vector acquisition means for acquiring the second feature vector which is the feature vector corresponding to normal operation, the first feature vector, and the second. Based on the difference in the statistical distribution from the feature vector of the above, the vehicle has a risk calculation means for calculating the risk to the driving of the evaluation target.

The occupant information is information obtained by sensing the driver, and typically includes, but is not limited to, line-of-sight information, face orientation, biological information, and classification results of driving behavior.
Further, the environmental information is information obtained by sensing the environment outside the vehicle, and typically includes, but is not limited to, the weather, the condition of the road surface, the position and distance of obstacles, and the like.
Further, the vehicle information is information obtained by sensing the vehicle, and typically includes speed, position information, engine speed, steering angle, braking force, front-rear acceleration, left-right acceleration, yaw rate, and the like. Not limited to these.

Further, the first feature vector acquisition means acquires a feature vector (first feature vector) corresponding to the driving to be evaluated based on the acquired occupant information, environmental information, and vehicle information. The acquired feature vector is compared with the second feature vector, which is a feature vector corresponding to normal operation, and the degree of risk is calculated based on the difference in the statistical distributions (for example, probability density) of both.
According to such a configuration, it is possible to determine whether or not the vehicle is in a dangerous state based on the driver, the surrounding environment, and the vehicle condition, so that the accuracy of determining dangerous driving can be improved.

Further, the occupant information, the environmental information, and the vehicle information may each be a multidimensional vector having a plurality of feature quantities represented by time series data as elements.

By holding a plurality of feature quantities in a time series format, it is possible to make a judgment in consideration of the time change of the state.

Further, the occupant information may be characterized by being a multidimensional vector having at least one of the occupant's line-of-sight position, face orientation, vital value, and action label that classifies driving behavior as elements.

The vital value is the biological information of the driver. The vital value may be, for example, information obtained by directly sensing the living body such as heart rate and respiratory rate, or information indirectly obtained based on the sensing result such as stress level and alertness. May be good. The action label is a value representing the result of classifying the actions performed by the driver and the driving actions.

Further, the environmental information may be characterized as a multidimensional vector having at least one of information on the weather outside the vehicle, the road surface condition, and the position of an obstacle as an element.

The information regarding the position of the obstacle is, for example, the absolute position of the obstacle, the relative position with respect to the own vehicle, the distance to the own vehicle, and the like. Further, the obstacle may be another automobile, a bicycle, a pedestrian, or the like. Further, it may be a fixed object such as a building or a structure.

Further, the vehicle information may be characterized by being a multidimensional vector having at least one of the speed, acceleration, steering angle, and yaw rate of the vehicle as an element.

This information can be obtained from an in-vehicle network such as CAN (Controller Area Network).

Further, the risk calculation means may be characterized in that the difference in the statistical distribution is calculated based on the ratio of the probability densities of the first feature vector and the second feature vector, and the risk may be calculated. The degree calculation means may be characterized in that the difference in the statistical distribution is calculated based on the difference in probability density between the first feature vector and the second feature vector .

By using the ratio or difference of probability densities, the difference in statistical distribution can be obtained accurately.

Further, the driving support device according to the present invention may be further provided with a display means for visualizing the magnitude of the difference in the statistical distribution and presenting the difference to the driver.

For example, the larger the difference, the more conspicuous the display may be. According to such a configuration, it is possible to intuitively convey to the driver how much the current driving state deviates from the normal driving state.

Further, the driving support device according to the present invention may further include a stimulus generating means for stimulating the driver based on the magnitude of the difference in the statistical distribution.
Further, the stimulus generating means may be characterized in that at least a part of the driver's seat is vibrated based on the magnitude of the difference in the statistical distribution.
Further, the stimulus generating means may be characterized in that a warning sound is emitted based on the magnitude of the difference in the statistical distribution.

In this way, by transmitting the magnitude of the difference by vibration or voice, it is possible to intuitively convey to the driver how much the current driving state deviates from the normal driving state. The object to be vibrated may be the seat itself on which the driver's seat is seated, a seat belt, a part of the steering wheel, or the like.

The present invention can be specified as a driving support device including at least a part of the above means. It can also be specified as a driving support method performed by the driving support device. The above processes and means can be freely combined and carried out as long as there is no technical contradiction.

According to the present invention, it is possible to provide a technique for accurately detecting dangerous driving.

It is a system block diagram of the driving support device which concerns on embodiment. It is a figure explaining the occupant information represented in the time series format. It is a figure explaining the line-of-sight detection of a driver. It is a figure explaining the position detection of an obstacle. This is an example of visually displaying the difference in statistical distribution. It is a processing flowchart figure of the driving support device which concerns on embodiment.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.
The driving support device according to the present embodiment is a device that detects that a dangerous driving state has occurred based on the collected information and warns the driver of the automobile. The dangerous driving state refers to a driving state that causes a dangerous state, and does not have to be caused by the driver himself / herself.

FIG. 1 is a system configuration diagram of the driving support device 1 according to the present embodiment. The driving support device 1 according to the present embodiment includes an occupant information acquisition unit 11, an environment information acquisition unit 12, a vehicle information acquisition unit 13, a feature vector calculation unit 14, a normal data storage unit 15, a comparison unit 16, and an image display unit 17. It has a stimulus generator 18.

The driving support device 1 can be configured as an information processing device having a CPU, a main storage device, and an auxiliary storage device. When the program stored in the auxiliary storage device is loaded into the main storage device and executed by the CPU, each means shown in FIG. 1 functions. In addition, all or a part of the illustrated functions may be executed by using a circuit (semiconductor integrated circuit or the like) specially designed.

The occupant information acquisition unit 11 is a means for collecting information about the driver from a first sensor (not shown) provided in the vehicle, converting it into a feature amount, and then outputting it in a time series format. In the present embodiment, the occupant information acquisition unit 11 converts the collected information into a multidimensional vector (referred to as occupant information) having a plurality of feature quantities and outputs the information.

The environmental information acquisition unit 12 is a means for collecting information on the environment outside the vehicle from a second sensor (not shown) provided in the vehicle, converting it into a feature amount, and then outputting it in a time series format. In the present embodiment, the environmental information acquisition unit 12 converts the collected information into a multidimensional vector (referred to as environmental information) having a plurality of feature quantities and outputs the information.

The vehicle information acquisition unit 13 is a means for collecting information on the state of the vehicle from a third sensor (not shown) provided in the vehicle, converting it into a feature amount, and then outputting it in a time series format. In the present embodiment, the vehicle information acquisition unit 13 converts the collected information into a multidimensional vector (referred to as vehicle information) having a plurality of feature quantities and outputs the information.

Here, specific examples of occupant information, environmental information, and vehicle information will be described.
The occupant information is a multidimensional vector P (with elements of the driver's line-of-sight position P ₁ (t), face orientation P ₂ (t), vital value P ₃ (t), and action label P ₄ (t) at time t. t) = {P ₁ (t)
, P ₂ (t), P ₃ (t), P ₄ (t)}.

As shown in FIG. 2, P (t) is a multidimensional vector having T pieces of information that are continuous in time. For example, when T = 3, the occupant information P (0) at t = 0 is {P ₁ (0).
, P ₂ (0), P ₃ (0), P ₄ (0), P ₁ (1), P ₂ (1), P ₃ (1), P ₄ (1),
It is given as a 4 × 3 dimensional vector of P ₁ (2), P ₂ (2), P ₃ (2), P ₄ (2)}.
The occupant information P (1) at t = 1 is {P ₁ (1), P ₂ (1), P ₃ (1), P ₄ (1), P ₁ (2), P ₂ (2). , P ₃ (2), P ₄ (2), P ₁ (3), P ₂ (3), P ₃ (
3), P ₄ (2)}.

Here, the number of T may be determined by trial and error. For example, it may be determined according to the identification performance of dangerous driving required for the entire system, or may be determined according to the processing time required for the entire system.

Each feature amount constituting the occupant information P _n (t) will be described.
The line-of-sight position P ₁ (t) can be obtained by photographing the driver's eyes with a camera provided in the vehicle and based on the obtained image (hereinafter, eye image). For example, FIG. 3 shows an example in which it is determined that the line-of-sight position of the driver at time t is (x _e (t), y _e (t)). It is assumed that the XY coordinates are the coordinates corresponding to the images taken in the front direction of the vehicle (hereinafter referred to as the front image) acquired at the same time.

Here, the line-of-sight direction of the vehicle driver is, for example, to extract the iris portion by comparing the difference image of the two eye images taken by the visible light sensing camera and the infrared light sensing camera with an appropriate threshold value. You can get it at. Then, based on the positional relationship between the camera that captures the eye image and the camera that captures the front image, the line-of-sight position (x _e) of the vehicle driver at time t.
(T), y _e (t)) can be obtained.

The driver's face orientation P ₂ (t) can be obtained by taking a photograph of the driver's face with a camera provided in the vehicle and based on the face image. The face orientation can be expressed by, for example, the inclinations (θ _x (t), θ _y (t), θ _z (t)) with respect to the X-axis, Y-axis, and Z-axis, respectively.
As disclosed in Non-Patent Document 1, for example, the driver's face orientation is a geometric shape composed of the obtained feature points by extracting feature points such as the inner corner of the eye, the corner of the eye, and the nose from the face image. Can be estimated based on. The camera for acquiring the driver's face orientation may also serve as a camera for acquiring the driver's line-of-sight position.

The vital value P ₃ (t) is a value representing the biological information of the driver. For the vital value, for example, an index affecting driving such as stress level, alertness, and fatigue can be used. Further, the vital value can be obtained by wearing, for example, a commercially available wristband type terminal (see Non-Patent Document 2).

The action label P ₄ (t) is a value representing the result of classifying the actions and driving actions performed by the driver. The behavior label can be estimated, for example, by acquiring a full-body image of the seated driver and based on the obtained full-body image.
For example, as disclosed in Non-Patent Document 3, there is a technique of detecting a part of the human body of an in-vehicle occupant using deep learning and estimating an action label based on the obtained global structure of the human body part. Are known.

Next, the environmental information will be described.
Environmental information includes weather E ₁ (t) at time t, road surface condition E ₂ (t), position E ₃ (t) of surrounding vehicles, relative distance E ₄ (t) with surrounding vehicles, and position E _{5 of} pedestrians. (T), a multidimensional vector E (t) = {E ₁ having elements of a relative distance E ₆ (t) to a pedestrian, a position E ₇ (t) of the bicycle, and a relative distance E ₈ (t) to the bicycle. Represented by (t), E ₂ (t), E ₃ (t), E ₄ (t), E ₅ (t), E ₆ (t), E ₇ (t), E ₈ (t)} ..
The environmental information E (t) is also a multidimensional vector having T pieces of information that are continuous in time, like the occupant information P (t).

Weather E ₁ (t) is the weather at the vehicle position at time t (eg, sunny, cloudy, rainy, snowy,
It is a value representing thick fog, etc.). The weather E ₁ (t) can be obtained, for example, by acquiring the current weather information at the vehicle position via the network.

The road surface condition E ₂ (t) is a value representing the road surface condition (for example, dry, semi-humid, wet, snow cover, compressed snow, freezing, sherbet, etc.) at time t. The road surface condition E ₂ (t) can be obtained by using a sensor installed in the vehicle, for example, as disclosed in Non-Patent Document 4.

As shown in FIG. 4, the position E ₃ (t) of the peripheral vehicle is the coordinates of the peripheral vehicle at time t when the lateral direction is the X axis and the traveling direction is the Y axis with respect to the vehicle (x _c (t)). , Y _c (t)). The coordinates of the surrounding vehicles can be obtained, for example, by using a LIDAR sensor installed outside the vehicle.

Further, the relative distance E ₄ (t) with the surrounding vehicle is the Euclidean distance to the surrounding vehicle at time t when the lateral direction is the X axis and the traveling direction is the Y axis with respect to the vehicle as shown in FIG. It is a value representing r _c .

The pedestrian position E ₅ (t), the relative distance to the pedestrian E ₆ (t), and the bicycle position E ₇ (t).
), The relative distance E ₈ to the bicycle may be acquired by the same method as for the surrounding vehicles, so detailed description thereof will be omitted.

Next, vehicle information will be described.
Vehicle information includes vehicle speed C ₁ (t) at time t, vehicle position C ₂ (t), engine speed C ₃ (t), steering angle C ₄ (t), brake oil pressure C ₅ (t), and forward / backward acceleration. C ₆ (t),
Multidimensional vector C (t) = {C ₁ with lateral acceleration C ₇ (t) and yaw rate C ₈ (t) as elements
(T), C ₂ (t), C ₃ (t), C ₄ (t), C ₅ (t), C ₆ (t), C ₇ (t), C ₈ (
It is represented by t)}. This information can be obtained from CAN (Controller Area Network).
The vehicle information C (t) is also a multidimensional vector having T pieces of information that are continuous in time, like the occupant information P (t) and the environmental information E (t).

Returning to the description of each element constituting the driving support device 1.
The feature vector calculation unit 14 is a means for integrating the occupant information output by the occupant information acquisition unit 11, the environmental information output by the environmental information acquisition unit 12, and the vehicle information output by the vehicle information acquisition unit 13.
Specifically, the occupant information P (t), the environmental information E (t), and the vehicle information C (t) are connected, and the feature vector F (t) = (P (t), E (P (t), E) at time t. t), C (t)) (t = 1, 2, ..., T) are calculated.
Since the occupant information, the environmental information, and the vehicle information include a plurality of time steps, the feature vector obtained by the integration is also a multidimensional vector according to the number of time steps. Hereinafter, the feature vector F (t) obtained by the integration will be referred to as evaluation data (first feature vector in the present invention).

The normal data storage unit 15 has a feature vector F'(t) (t = 1,) generated based on a driving performed by a good driver (driving that does not include a dangerous event; hereinafter referred to as normal driving). 2, ..., T') is a means of memorizing. The T'used here may be the same as or different from the value T used for calculating the evaluation data F (t). Hereinafter, the feature vector F'(t) generated based on the normal operation will be referred to as normal data (second feature vector in the present invention).

The normal data F'(t) stored in the normal data storage unit 15 may be generated by having one driver perform normal operation, but it is normal for a plurality of people of various types such as age and gender. It may be generated by driving.
Further, the normal data may be generated by using an external device or may be generated by using the driving support device 1. When the normal data is generated by using the driving support device 1, the feature vector calculated by the feature vector calculation unit 14 may be stored in the normal data storage unit 15 as normal data (dotted line in FIG. 1).

Further, the driver who performs the driving for generating normal data and the driver who is evaluated by the driving support device 1 may be the same person. For example, after a certain driver performs normal operation to generate normal data, the same driver may be evaluated by the driving support device 1. Further, normal data may be acquired or updated via the network.

The evaluation data calculated by the feature vector calculation unit 14 (that is, the feature vector corresponding to the operation to be evaluated) and the normal data stored in the normal data storage unit 15 (that is, the feature vector corresponding to the normal operation) are It is transmitted to the comparison unit 16 and the comparison is performed.

The comparison unit 16 is a means for calculating the difference D in the statistical distribution between the feature vector (normal data) stored in the normal data storage unit 15 and the feature vector (evaluation data) calculated by the feature vector calculation unit 14. ..
In this embodiment, the probability density followed by a sample consisting of T time-series features F (0), F (1), F (2), ..., F (T) and T'time-series features Difference in statistical distribution of two feature vectors based on the probability density followed by a sample consisting of F'(0), F'(1), F'(2), ..., F'(T') D To estimate.

For example, using the density ratio estimation method as described in Non-Patent Documents 5 and 6, the probability densities for T feature vectors F (0), ..., F (T) and T'feature vectors. The ratio of the probability densities to F'(0), ..., F'(T') (density ratio) may be estimated, and the difference D may be calculated based on the obtained density ratio. The probability density ratio can be calculated by, for example, the relative density ratio estimation method described in Non-Patent Document 5, the minimum square density ratio matching method described in Non-Patent Document 6, the probabilistic classification method, the product factor matching method, or the like. ..
Alternatively, the difference D is, for example, the probability density for T feature vectors F (0), ..., F (T) and for T'feature vectors F'(0), ..., F'(T'). The difference in probability density (density difference) may be estimated and calculated based on the obtained density difference. The probability density difference can be calculated by, for example, the probability density difference estimation method described in Non-Patent Document 7.

The image display unit 17 is a means for drawing the difference D calculated by the comparison unit 16 in an intuitive and easy-to-understand figure and presenting it to the vehicle driver. Here, it can be estimated that the larger the difference D calculated by the comparison unit 16, the higher the possibility of dangerous driving, and the smaller the difference D, the higher the possibility of normal driving. Therefore, as shown in FIG. 5, a figure that changes (animates) depending on the magnitude of the difference D may be drawn. By doing so, it is possible to notify the vehicle driver that a dangerous state has occurred and call attention to it. In addition to the graphic, the image display unit 17 may present the difference D to the vehicle driver by outputting characters (text).

Although the example in which the difference D is always presented is given here, the display may be performed only when the magnitude of the difference D is larger than a predetermined threshold value. That is, the presentation by visual information may be performed only when a situation in which there is a high possibility of dangerous driving occurs. The threshold value used here can be calculated using, for example, the intersection confirmation method.

The stimulus generation unit 18 is a means for giving a stimulus (for example, vibration or voice warning) to the vehicle driver based on the magnitude of the difference D obtained in the comparison unit 16. For example, the seat or steering wheel may be vibrated according to the magnitude of the difference D. Further, vibration may be generated by controlling the tightening strength of the seat belt.
Alternatively, a warning sound may be emitted depending on the magnitude of the difference D. Or, depending on the size of the difference D, "It's safe driving. That's how it is.""It seems that you are distracted. Be careful.""Are you tired? Take a break in the next service area. You may call your attention with a voice such as "Why don't you try it?"
It should be noted that the stimulus may be generated only when the magnitude of the difference D is larger than a predetermined threshold value.

Next, the flow of processing performed by the driving support device 1 according to the present embodiment will be described.
FIG. 6A is a flowchart illustrating the generation of normal data.
First, in step S11, the occupant information acquisition unit 11, the environmental information acquisition unit 12, and the vehicle information acquisition unit 13 collect occupant information, environmental information, and vehicle information in normal driving (not dangerous driving), respectively.
Next, in step S12, the feature vector calculation unit 14 calculates the feature vector using the collected information.
Next, in step S13, the feature vector calculation unit 14 stores the feature vector calculated in step S12 as normal data in the normal data storage unit 15.

FIG. 6B is a flowchart illustrating a process of evaluating the operation. The process described in FIG. 6B is periodically executed, for example, while the vehicle is running.
First, in step S21, the occupant information acquisition unit 11, the environment information acquisition unit 12, and the vehicle information acquisition unit 13 collect occupant information, environment information, and vehicle information, respectively.
Next, in step S22, the feature vector calculation unit 14 calculates the feature vector using the collected information.
Next, in step S23, the comparison unit 16 statistically obtains the feature vector (evaluation data) calculated in step S22 and the feature vector (normal data) during normal operation stored in the normal data storage unit 15. Calculate the difference D in the distribution.
Then, in step S24, the image display unit 17 and the stimulus generation unit 18 alert the vehicle driver by outputting an image or giving a stimulus according to the magnitude of the difference D calculated in step S23. I do.

As described above, the driving support device according to the present embodiment calculates the feature vector based on the three pieces of information of the occupant information, the environmental information, and the vehicle information. By calculating the feature vector based on such a plurality of multifaceted information, it is possible to detect a danger involving complicated factors.
Further, the driving support device according to the present embodiment calculates the degree of risk based on the difference between the statistical distribution of the feature vector extracted from the excellent driver and the statistical distribution of the feature vector extracted from the evaluation target driver. To do. With such an approach, it is possible to detect not only overt dangers that can be easily defined, but also potential dangers that are difficult for humans to define directly.
In addition, by stimulating the vehicle driver, it is possible to prevent a dangerous state from occurring.

With the prior art, it has been difficult to detect a pattern in which a plurality of factors are intricately entwined. On the other hand, the driving support device according to the present embodiment calculates the difference in the statistical distribution by using the density ratio estimation method or the density difference estimation method, which is an advanced machine learning method. As a result, it is possible to detect an abnormal pattern in which a plurality of factors are complicatedly entwined, and it is possible to accurately detect dangerous driving.

(Modification example)
The above embodiment is merely an example, and the present invention can be appropriately modified and implemented without departing from the gist thereof.
For example, in the description of the embodiment, the image is displayed and the stimulus is generated based on the difference in the statistical distribution of the time-series features during the evaluation and the normal operation, but when a dangerous driving is detected, a notification is sent. The output may be made to another device.

Further, in the description of the embodiment, it is assumed that the driving support device 1 generates normal data, but the normal data may be generated by another device and stored in the normal data storage unit of the driving support device 1. Good.

Further, in the description of the embodiment, the normal data is held in the device, but the normal data may be acquired from the outside of the device each time. Further, the driving support device 1 does not necessarily have to be mounted on the vehicle. For example, it can be implemented as a server device configured to acquire information from an in-vehicle terminal and transmit the result of comparison to the in-vehicle terminal.

1 ... Driving support device 11 ... Crew information acquisition unit 12 ... Environmental information acquisition unit 13 ... Vehicle information acquisition unit 14 ... Feature vector calculation unit 15 ... Normal data storage unit 16 ...・ Comparison unit 17 ・ ・ ・ Image display unit 18 ・ ・ ・ Stimulation generation unit

Claims (11)

A driving support device that detects dangerous driving conditions of a vehicle.
Crew information acquisition means for acquiring occupant information, which is information about the driver of the vehicle, and
Environmental information acquisition means for acquiring environmental information, which is information on the external environment of the vehicle, and
Vehicle information acquisition means for acquiring vehicle information which is information about the vehicle, and
A first feature vector acquisition means for acquiring a first feature vector, which is a feature vector corresponding to the driving to be evaluated, based on the occupant information, the environmental information, and the vehicle information.
A second feature vector acquisition means for acquiring the second feature vector, which is the feature vector corresponding to normal operation, and
A risk calculation means for calculating the risk to the driving of the evaluation target based on the difference in the statistical distribution between the first feature vector and the second feature vector.
Have a,
The risk calculation means calculates the difference in the statistical distribution based on the ratio or difference of the probability densities between the first feature vector and the second feature vector.
Driving support device. The occupant information, the environmental information, and the vehicle information are multidimensional vectors each having a plurality of feature quantities represented by time series data as elements.
The driving support device according to claim 1. The occupant information is a multidimensional vector having at least one of the occupant's line-of-sight position, face orientation, vital value, and behavior label that classifies driving behavior as elements.
The driving support device according to claim 2. The environmental information is a multidimensional vector having at least one of information on the weather outside the vehicle, the road surface condition, and the position of an obstacle as an element.
The driving support device according to claim 2 or 3. The vehicle information is a multidimensional vector having at least one of the vehicle speed, acceleration, steering angle, and yaw rate as an element.
The driving support device according to any one of claims 2 to 4. Further provided with a display means for visualizing the magnitude of the difference in the statistical distribution and presenting it to the driver.
The driving support device according to any one of claims 1 to 5 . Further provided with a stimulus generating means for stimulating the driver based on the magnitude of the difference in the statistical distribution.
The driving support device according to any one of claims 1 to 6 . The stimulus generating means vibrates at least a part of the driver's seat based on the magnitude of the difference in the statistical distribution.
The driving support device according to claim 7 . The stimulus generating means emits a warning sound based on the magnitude of the difference in the statistical distribution.
The driving support device according to claim 7 or 8 . It is a driving support method performed by a driving support device that detects a dangerous driving state of a vehicle.
The occupant information acquisition step for acquiring occupant information, which is information about the driver of the vehicle, and
An environmental information acquisition step for acquiring environmental information, which is information on the external environment of the vehicle, and
A vehicle information acquisition step for acquiring vehicle information, which is information about the vehicle, and
A first feature vector acquisition step for acquiring a first feature vector, which is a feature vector corresponding to the driving to be evaluated, based on the occupant information, the environmental information, and the vehicle information.
A second feature vector acquisition step for acquiring the second feature vector, which is the feature vector corresponding to normal operation, and
A risk calculation step for calculating the risk to the driving of the evaluation target based on the difference in the statistical distribution between the first feature vector and the second feature vector.
Only including,
In the risk calculation step, the difference in the statistical distribution is calculated based on the ratio or difference of the probability densities between the first feature vector and the second feature vector.
Driving support method. A program that causes a computer to execute the driving support method according to claim 10 .

Images