JP3843513B2 - Vehicle alarm device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an alarm device for a vehicle, and more particularly to generation of an alarm when a driver's concentration is reduced.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, there is known a device that outputs a warning by detecting a decrease in concentration of a vehicle driver. For example, Japanese Patent Laid-Open No. 3-260900 includes alarm generation means for giving an alarm when the relative speed of the host vehicle with respect to the preceding vehicle is equal to or higher than a predetermined allowable relative speed, and detects aside driving or dozing operation. In such a case, a technique is disclosed in which the allowable relative speed is corrected to be low so that an alarm is easily generated.
[0003]
[Problems to be solved by the invention]
However, the driver determines that the current driving state (relative speed with respect to the preceding vehicle, inter-vehicle distance, etc.) is within the range of his / her control and is safe, and temporarily confirms the post-measurement and operates the in-vehicle audio. In many cases, it is determined that the driving is unambiguously determined to be a side-viewing operation, and the alarm is given because there is a problem that the driver feels troublesome because it does not match the driving feeling of the driver.
The present invention has been made in view of the above-described problems of the prior art, and an object of the present invention is to provide an alarm device that can reliably suppress unnecessary alarms and give an alarm to a driver at an appropriate timing. There is.
[0004]
[Means for Solving the Problems]
In order to achieve the above object, the first invention provides a detecting means for detecting a running state quantity, an evaluating means for evaluating a driver's concentration reduction, and a running state quantity at the time when the driver's concentration falls. And a control means for outputting an alarm when the reference running state quantity differs from the current running state quantity by a predetermined value or more.
[0005]
In a second aspect based on the first aspect, the evaluation means comprises: The driver's face is detected based on information on the driver's face direction and eye state obtained by processing driver face image information obtained by photographing the driver's face. The The decrease in concentration is evaluated.
[0006]
In addition, a third invention is characterized in that, in the first invention, the evaluation means evaluates the concentration reduction by a driver's operation of the in-vehicle device.
[0007]
According to a fourth aspect, in the first to third aspects, the detecting means detects an inter-vehicle distance as the travel state quantity.
[0008]
According to a fifth invention, in the first to third inventions, the detecting means detects a curvature or a slope of a road as the running state quantity.
[0009]
According to a sixth invention, in the first to third inventions, the detecting means detects a lateral displacement amount of the vehicle with respect to a travel lane as the travel state amount.
[0010]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0011]
<First Embodiment>
FIG. 1 is a configuration block diagram of the present embodiment. The TV camera 10 is installed in the vicinity of the front panel of the vehicle and photographs the face of the vehicle driver. The TV camera 10 is composed of, for example, a CCD (Charge Coupled Device) solid-state image sensor. In order to obtain a clear image with the TV camera 10, an illumination device may be provided at a predetermined position of the vehicle to illuminate the driver's face. At this time, near-infrared light (wavelength 700 to 1000 μm) is preferably used as illumination light so as not to hinder driving. Ordinary white light may be cut with a visible light cut filter and used as irradiation light. it can. The TV camera 10 may be filtered so that only the illumination light from the illumination device is incident, and it is not always necessary to take a front image of the driver. For example, an image in a predetermined oblique direction may be used. .
[0012]
The driver's face image information photographed by the TV camera 10 is sent to the binarization circuit 12 and binarized to a white or black level using a predetermined threshold. It is also possible to use face image information as analog data. The face image information binarized to the white or black level is further sent to the cutout circuit 14, where a window is set only for a predetermined area to be processed and cut out from the face image information. When a CCD solid-state imaging device or the like is used as the TV camera 10, the face image information is composed of a plurality of pixel information arranged two-dimensionally, and each pixel information is binarized to a white or black level. Will be output. Now, a number is assigned to the plurality of pixels arranged two-dimensionally in this way, the pixel located in the i-th row and j-th column is denoted as aij, and the pixel information is denoted as Iij. The face image information Iij cut out in this way (having a value of either logic level 1 corresponding to the white level or logic level 0 corresponding to the black level) is input to the neural network 16 and processed. The Note that these binarization processing, clipping processing, and neural network processing described below may be performed using a DSP (Digital Signal Processor). As is well known, a neural network has a hierarchical structure of an input layer, an intermediate layer, and an output layer, and each layer is composed of a group of neural units. The neural unit is a unit that converts and outputs a predetermined rule according to the total sum of inputs to which a predetermined weight (coupling coefficient) wij is added. Various functions are used as this rule. In this embodiment, the total sum “net” of the inputs is used.
net = ΣwijIi
When
f = 1 / {1 + exp [− (net + α)]} where α is a constant
I decided to use the sigmoid function. This function has a value range of 0 to 1 and shows a characteristic that approaches 1 as the input value increases and approaches 0 as the input value decreases.
[0013]
Now, the binarized face image information from the cut-out circuit 14, that is, the image information Iij (0 or 1) of the pixel aij constituting the cut-out image is input to each neural unit constituting the input layer of the neural network. Is done. Therefore, when the cut out image information is composed of 240 × 40 pixels, there are 9600 neural units equal to the number of pixels. If you want to reduce the number of neural units in the input layer, divide the clipped image into several small areas, and add the pixel information in the divided small areas to perform the averaging process to calculate the number of neural units. Should be reduced. For example, if the small area is set to 4 × 4, the number of neural units is reduced to 60 × 10 = 600.
[0014]
After the pixel information Iij of each pixel constituting the image cut out in this way is input to each neural unit in the input layer, a predetermined conversion process is performed in the intermediate layer and the output layer made up of similar neural unit groups. That is, if the output value of the i-th neural unit in the input layer input to the j-th neural unit in the intermediate layer is I i and the weighting, that is, the coupling coefficient at this time is w ij, the j-th neural The total netj of input values input to the unit is
netj = Σwij · Ii
The output is as described above using the sigmoid function,
yj = f (netj)
It becomes. The above-described processing is performed in all the neural units existing in the intermediate layer, and the output value is input to each neural unit in the output layer. Also in the output layer neural unit, the same conversion processing as in the intermediate layer is performed. The weighting from the input layer to the intermediate layer and the weighting from the intermediate layer to the output layer are adjusted in advance by learning so that the difference between the actual output value from the output layer and the desired output value is reduced. .
[0015]
By using the neural network in this way, a set of output values corresponding to the set of input image information can be obtained. However, each of the neural units 1, 2,. The predetermined state corresponds, and the direction of the face and the state of the eyes of the vehicle driver can be determined from the output value from the output layer. That is, the output from the first neural unit in the output layer means that the driver's face is facing the front, and the output value from the second neural unit is that the driver's eyes are closed. Corresponds to the state, and the output value from the third neural unit corresponds to the driver facing the instrument panel. Therefore, when the output values from the output layer are (1, 0, 0,..., 0), respectively, it is detected that the face of the vehicle driver faces the front and is in an open state. Then, the driver's concentration is evaluated based on the detected information on the face direction and eye state of the vehicle driver, and the evaluation result is output to an electronic control unit (ECU) that performs electronic control of the vehicle. Note that the driver's concentration is evaluated to be low when, for example, the driver's aside is detected.
[0016]
Then, the ECU 18 activates an alarm device based on the evaluation result of the driver's concentration and information from the sensor 20 that detects the running state to give an alarm to the driver. Specifically, a radar device or the like that detects the inter-vehicle distance from the preceding vehicle is used as the sensor 20, and the ECU 18 temporarily stores the inter-vehicle distance from the preceding vehicle when the driver's concentration is reduced as a reference value. Then, an alarm is given when the inter-vehicle distance is decreased by a predetermined value or more with respect to the reference inter-vehicle distance while the driver's concentration is continuing to decrease. Of course, it is also possible to control the brake actuator instead of or together with the alarm.
[0017]
FIG. 2 shows a processing flowchart in the present embodiment. First, the distance between the preceding vehicle and the preceding vehicle is detected by the sensor 20 (S101), and the driver's look-ahead is detected by the neural network 16 in order to evaluate the driver's (driver) concentration reduction (S102). Next, it is confirmed whether a preceding vehicle exists (S103). This confirmation is determined by whether or not finite inter-vehicle distance data is output from the sensor 20. When there is no preceding vehicle, a predetermined value LW (for example, 80 m) is given as an initial value to the reference inter-vehicle distance LT (S109). If there is a preceding vehicle, it is next determined whether or not the driver is looking aside (S104). When the driver is not looking aside, that is, when the driver's concentration is not reduced, the predetermined value LW is substituted for the reference inter-vehicle distance LT as in S109 (S110), and the current inter-vehicle distance L is calculated. The allowable inter-vehicle distance L0 (for example, 30 m) is compared (S111). Even if the driver's concentration is not reduced, if the current inter-vehicle distance L is smaller than the allowable inter-vehicle distance L0, an alarm is output to alert the driver (S108). On the other hand, if the driver is looking aside, it is further determined whether the driver was looking aside last time (S105). This determination is to confirm whether or not the driver is looking aside. If the driver does not look aside last time and does this for the first time this time (starting to reduce the driver's concentration), The inter-vehicle distance L at the time is set to the reference inter-vehicle distance LT (S106). Further, when looking aside again (continuation state where the driver's concentration is lowered), the process of S106 is not performed, and the value of the reference inter-vehicle distance at the time of concentration reduction is maintained. Then, the current inter-vehicle distance L is compared with the reference inter-vehicle distance LT, and it is determined whether or not the two are different by a predetermined value or more, that is, whether the current inter-vehicle distance L is smaller than LT−ΔL1 (S107). However, ΔL1 is a predetermined allowable change value, and is set to 10 m, for example. When the current inter-vehicle distance L decreases by a predetermined value or more with respect to the reference inter-vehicle distance LT, an alarm is output to alert the driver (S108). On the other hand, if the current inter-vehicle distance L has hardly changed with respect to the reference inter-vehicle distance LT even if the driver's concentration is reduced, it is determined that safety is secured and no alarm is output.
[0018]
As described above, in the present embodiment, the inter-vehicle distance at the time when the driver looks aside is used as a reference, and the current inter-vehicle distance does not change by a predetermined value or more with respect to the reference inter-vehicle distance even if the looking-aside state continues. In this case, since no warning is output, it is possible to prevent an unnecessary warning from being given when the driver judges that the safety is sufficient and temporarily looks aside.
Second Embodiment
In the first embodiment described above, the driver's concentration reduction is evaluated based on whether or not the driver is looking aside. However, in this embodiment, the concentration decreases depending on whether the on-vehicle mobile phone is operated or not. To evaluate. The basic configuration is almost the same as in FIG. 1, but instead of the TV camera 10 and the neural network 16, a circuit for detecting the presence or absence of operation of the mobile phone is provided, and when the mobile phone is in an operating state, the driver's concentration It is different in that it is determined that the power is decreasing.
[0019]
FIG. 3 shows a processing flowchart of the present embodiment. The distance between the preceding vehicle and the preceding vehicle is detected by the sensor 20 (S201), and the operation of the mobile phone is detected in order to evaluate the driver's (driver) concentration reduction (S202). Next, it is confirmed whether there is a preceding vehicle (S203). If there is no preceding vehicle, a predetermined value LW (for example, 80 m) is given as an initial value to the reference inter-vehicle distance LT (S209). If there is a preceding vehicle, it is next determined whether or not the driver is operating the mobile phone (S204). When the driver is not operating the mobile phone, that is, when the driver's concentration is not reduced, the predetermined value LW is substituted for the reference inter-vehicle distance LT as in S209 (S210), and the current inter-vehicle distance L is compared with the allowable inter-vehicle distance L0 (for example, 30 m) (S211). Even if the driver's concentration is not reduced, if the current inter-vehicle distance L is smaller than the allowable inter-vehicle distance L0, an alarm is output to alert the driver (S208). On the other hand, if the driver is operating the mobile phone, it is further determined whether or not the mobile phone was operated last time (S205). This determination is for confirming whether or not the operation of the mobile phone is continued. When the operation is not performed last time and this operation is performed for the first time, the inter-vehicle distance L at that time is set as the reference inter-vehicle distance LT. (S206). Further, when the mobile phone has been operated last time, the process of S206 is not performed, and the value of the reference inter-vehicle distance at the time of concentration reduction is maintained. Then, the current inter-vehicle distance L is compared with the reference inter-vehicle distance LT, and it is determined whether or not both are different by a predetermined value or more, that is, whether the current inter-vehicle distance L is smaller than LT−ΔL 2 (S207). However, ΔL2 is a predetermined allowable change value, for example, set to 8 m. When the current inter-vehicle distance L decreases by a predetermined value or more with respect to the reference inter-vehicle distance LT, an alarm is output to alert the driver (S208). On the other hand, if the current inter-vehicle distance L has hardly changed with respect to the reference inter-vehicle distance LT even if the driver's concentration is reduced, it is determined that safety is secured and no alarm is output.
[0020]
Thus, in the present embodiment, the inter-vehicle distance at the time when the driver operates the mobile phone is a reference, and even if the operation state of the mobile phone continues, the current inter-vehicle distance is a predetermined value with respect to the reference inter-vehicle distance. Since no warning is output when there is no change, it is possible to prevent an unnecessary warning from being given when the driver determines that the safety is sufficient and temporarily operates the mobile phone.
[0021]
<Third Embodiment>
In the second embodiment described above, the driver's concentration reduction is evaluated by operating the mobile phone. However, in this embodiment, whether or not the in-vehicle navigation device is operated, in particular whether the destination is input by voice. Evaluate the decline in concentration based on whether or not The basic configuration is almost the same as that shown in FIG. 1, but a circuit for detecting ON / OFF of voice input for in-vehicle navigation is provided instead of the TV camera 10 and the neural network 16, and the driver's The difference is that it is determined that the concentration is reduced.
[0022]
FIG. 4 shows a processing flowchart of the present embodiment. The distance between the preceding vehicle and the preceding vehicle is detected by the sensor 20 (S301), and in-vehicle navigation voice input is detected in order to evaluate the driver's (driver) concentration reduction (S302). Next, it is confirmed whether there is a preceding vehicle (S303). When there is no preceding vehicle, a predetermined value LW (for example, 80 m) is given as an initial value to the reference inter-vehicle distance LT (S309). If there is a preceding vehicle, it is next determined whether or not the driver inputs a destination or the like by voice (S304). When the driver is not inputting voice, that is, when the driver's concentration is not reduced, the predetermined value LW is substituted for the reference inter-vehicle distance LT as in S309 (S310), and the current inter-vehicle distance L is calculated. The allowable inter-vehicle distance L0 (for example, 30 m) is compared (S311). Even if the driver's concentration is not reduced, if the current inter-vehicle distance L is smaller than the allowable inter-vehicle distance L0, an alarm is output to alert the driver (S308). On the other hand, if the driver is inputting voice, it is further determined whether or not the previous voice input was made (S305). This determination is for confirming whether or not the voice input state is continued. When the voice input is not input last time and the voice is input for the first time this time, the inter-vehicle distance L at that time is set as the reference inter-vehicle distance. Set to LT (S306). In addition, when voice is input last time, the process of S306 is not performed, and the value of the reference inter-vehicle distance at the time when the concentration is lowered is maintained. Then, the current inter-vehicle distance L is compared with the reference inter-vehicle distance LT, and it is determined whether or not the two are different by a predetermined value or more, that is, whether the current inter-vehicle distance L is smaller than LT−ΔL3 (S307). However, ΔL3 is a predetermined allowable change value, for example, set to 9 m. If the current inter-vehicle distance L decreases by a predetermined value or more with respect to the reference inter-vehicle distance LT, an alarm is output to alert the driver (S308). On the other hand, if the current inter-vehicle distance L has hardly changed with respect to the reference inter-vehicle distance LT even if the driver's concentration is reduced, it is determined that safety is secured and no alarm is output.
[0023]
As described above, in the present embodiment, the vehicle-to-vehicle distance at the time when the driver inputs voice for in-vehicle navigation is used as a reference, and even if the voice input state continues, the current vehicle-to-vehicle distance is a predetermined value with respect to the reference vehicle-to-vehicle distance. Since no warning is output when there is no change, it is possible to prevent an unnecessary warning from being given when the driver judges that the safety is sufficient and temporarily inputs voice.
In the present embodiment, it goes without saying that the present invention can be similarly applied to a case where a destination or the like is manually set instead of voice input.
[0024]
<Fourth embodiment>
In the first to third embodiments described above, the inter-vehicle distance from the preceding vehicle is detected as the travel state quantity of the vehicle. However, in this embodiment, the case where the curvature of the road is detected as the travel state quantity is shown. The basic configuration is almost the same as that of FIG. 1 except that a sensor for detecting the curvature of the road is provided as the sensor 20, and the curvature at the time when the driver's concentration is reduced is used as a reference curvature. The curvature sensor can be composed of, for example, map data for in-vehicle navigation and GPS, and the curvature of the road on the map data corresponding to the detected current position may be searched.
[0025]
FIG. 5 shows a processing flowchart of the present embodiment. First, the current position is detected by GPS or the like (S401), and the driver (driver) looking aside is detected by the neural network 16 (S402). Then, it is determined whether or not the person is looking aside (S403). If the person is looking aside, it is further determined whether or not the person was looking aside again (S404). When looking aside for the first time this time (start of the driver's concentration reduction), the curvature ρ at that time is set as the reference curvature ρT (S407). On the other hand, in the case where the last time was a side look (a state in which the concentration reduction continued), it is determined whether or not the curvature ρ1 ahead of the current position (for example, 10 m ahead) is equal to the reference curvature ρT (S405). ). The curvature ahead of the current position can be searched from the map data of the in-vehicle navigation, and if they are not equal, that is, if they differ by more than a predetermined value, an alarm is output to alert the driver (S406). On the other hand, if the curvature of the road does not change from the reference curvature even if the driver's concentration is reduced, it is determined that safety is secured and no warning is output. Therefore, in this embodiment, even if the driver is temporarily looking aside while traveling on a straight road, no warning is given as long as the driver is traveling on a straight road, but he was temporarily looking aside while traveling on a straight road. However, if there is a curved road ahead, an alarm will be given.
[0026]
Also according to the present embodiment, it is possible to prevent a warning that is considered unnecessary for the driver from occurring frequently, and to give a warning at a timing that is considered to be truly necessary.
[0027]
In the present embodiment, the reduction in the driver's concentration can be evaluated by operating a mobile phone or inputting voice.
[0028]
<Fifth Embodiment>
In the fourth embodiment described above, the curvature of the road is detected as the travel state quantity of the vehicle. However, in the present embodiment, a case where the road gradient is detected as the travel state quantity is shown. The basic configuration is almost the same as that in FIG. 1 except that a sensor for detecting a road gradient is provided as the sensor 20, and the gradient at the time when the driver's concentration is reduced is used as a reference gradient. The gradient sensor can be composed of, for example, map data for in-vehicle navigation and GPS, and it is only necessary to search for the gradient of the road on the map data corresponding to the detected current position.
[0029]
FIG. 6 shows a processing flowchart of the present embodiment. First, the current position is detected by GPS or the like (S501), and the driver (driver) looking aside is detected by the neural network 16 (S502). Then, it is determined whether or not the person is looking aside (S503). If the person is looking aside, it is further determined whether or not the person was looking aside last time (S504). When looking aside for the first time this time (start of reduction of driver's concentration), the gradient θ at that time is set as a reference gradient θT (S507). On the other hand, if the previous look was also a side look (a state in which the concentration reduction continued), it is determined whether or not the gradient θ1 that is a predetermined forward (for example, 10 m ahead) from the current position is equal to the reference gradient θT (S505). ). The gradient ahead of the current position can be searched from the map data of in-vehicle navigation, and if they are not equal, that is, if they differ by more than a predetermined value, an alarm is output to alert the driver (S506). On the other hand, if the road gradient does not change from the reference gradient even when the driver's concentration is reduced, it is determined that safety is ensured and no warning is output. Therefore, in this embodiment, even if the driver is temporarily looking aside while driving on a flat road, no warning is given as long as the driver is driving on a flat road, but he was temporarily looking aside while driving on a flat road. However, if there is a downhill (or uphill) ahead, an alarm will be given.
[0030]
Also according to the present embodiment, it is possible to prevent a warning that is considered unnecessary for the driver from occurring frequently, and to give a warning at a timing that is considered to be truly necessary.
[0031]
Also in this embodiment, the driver's concentration reduction can be evaluated with a mobile phone or the like.
[0032]
<Sixth Embodiment>
In each of the above-described embodiments, the inter-vehicle distance, road curvature, or gradient with respect to the preceding vehicle is detected as the travel state amount of the vehicle, but in this embodiment, the lateral displacement amount with respect to the travel lane of the vehicle is detected as the travel state amount of the vehicle. Show the case. The basic configuration is almost the same as in FIG. 1, and a sensor 20 is provided as a sensor 20 for detecting the lateral displacement of the vehicle with respect to the travel lane, and the lateral displacement when the driver's concentration is reduced is used as a reference lateral displacement. The amount is different.
[0033]
FIG. 7 shows a processing flowchart of the present embodiment. The sensor 20 detects the lateral displacement amount ΔX in the travel lane (S601), and the neural network 16 detects the driver (driver) looking aside (S602). Next, it is confirmed whether or not a travel lane has been detected (S603). When the traveling lane cannot be detected (when there is no white line, etc.), a predetermined value ΔXw (for example, 0.4 m) is given as an initial value to the reference left lateral displacement amount ΔXTl and right lateral displacement amount ΔXTr, respectively (S609). If the travel lane can be detected, it is next determined whether or not the driver is looking aside (S604). When the driver is not looking aside, that is, when the driver's concentration is not reduced, the predetermined value ΔXW is substituted for the reference lateral displacement amounts ΔXTl and ΔXTr as in S609 (S610), The displacement amount ΔX is compared with the allowable lateral displacement amount Δx1 (for example, 0.8 m) (S611). If the current lateral displacement amount ΔX exceeds the allowable lateral displacement amount Δx1 even if the driver's concentration is not reduced, an alarm is output to alert the driver (S608). On the other hand, if the driver is looking aside, it is further determined whether the driver was looking aside last time (S605). This determination is for confirming whether or not the look-aside is continued. If the look-ahead is not looked aside last time and the look is looked aside for the first time this time, the amount of lateral displacement ΔX at that time is an allowable amount. Is set as the reference lateral displacement amount (S606). Specifically, it is as follows.
[0034]
<When the vehicle is displaced to the left with respect to the center of the lane>
The reference left lateral displacement amount ΔXTL is ΔXTl = | ΔX | + ΔXww
The reference right lateral displacement amount ΔXtr is ΔXTr = ΔXw
<When the vehicle is displaced to the right with respect to the center of the lane>
The reference amount of left lateral displacement ΔXTl is ΔXTl = ΔXw
The reference right lateral displacement amount ΔXTr is ΔXTr = | ΔX | + ΔXww
However, ΔXww is an allowable amount, and is preferably changed according to ΔX, such as 0.4 m when | ΔX | is small and 0.2 m when | ΔX | is large. In addition, when looking aside last time, the process of S606 is not performed and the value of the reference lateral displacement amount at the time of concentration reduction is maintained. Then, the current lateral displacement amount ΔX and the reference lateral displacement amounts ΔXT1, ΔXTr are compared, and it is determined whether or not the current lateral displacement amount exceeds the reference lateral displacement amount (S607). If the current lateral displacement amount exceeds the reference lateral displacement amount, an alarm is output to alert the driver (S608). On the other hand, if the current lateral displacement has hardly changed relative to the reference lateral displacement even if the driver's concentration is reduced, safety is ensured even if the vehicle is not driving in the center of the driving lane. It is judged that the alarm has been made and no alarm is output.
[0035]
As described above, in this embodiment, even if the aside look state continues, if there is no difference between the lateral displacement amount at the start of looking aside and the current lateral displacement amount, an alarm is not output, so the driver determines that it is sufficiently safe. It is possible to prevent an unnecessary alarm from being given when looking aside.
[0036]
In this embodiment, the driver's concentration reduction may be evaluated by operating a mobile phone or the like. When the lateral displacement amount exceeds the reference lateral displacement amount, the ECU 18 generates an alarm and performs steering. It is also possible to control to return the vehicle to the center of the traveling lane by driving the actuator.
[0037]
Furthermore, Δx1, ΔXw, and ΔXww in the present embodiment can be changed according to the vehicle speed, the lane width, the curvature, and the like.
[0038]
【The invention's effect】
As described above, according to the present invention, it is possible to reliably suppress unnecessary alarms that are bothersome for the driver and to give the driver an alarm at an appropriate timing.
[Brief description of the drawings]
FIG. 1 is a configuration block diagram of an embodiment of the present invention.
FIG. 2 is a process flowchart of the first embodiment.
FIG. 3 is a processing flowchart of the second embodiment.
FIG. 4 is a process flowchart of the third embodiment.
FIG. 5 is a process flowchart of the fourth embodiment.
FIG. 6 is a process flowchart of the fifth embodiment.
FIG. 7 is a processing flowchart of the sixth embodiment.
[Explanation of symbols]
10 TV camera, 12 binarization circuit, 14 clipping circuit, 16 neural network, 18 ECU (electronic control unit), 20 sensor.

Claims (6)

Detecting means for detecting a running state quantity;
An evaluation means for evaluating the driver's concentration reduction,
Control means for outputting a warning when the driving state amount at the time when the driver's concentration is reduced, and when the reference driving state amount is different from the current driving state amount by a predetermined value or more,
A vehicular alarm device comprising: The evaluation means is based on information about the driver's face direction and eye state obtained by processing driver face image information obtained by photographing the driver's face. The vehicle alarm device according to claim 1, wherein the side-by-side detection is detected to evaluate the concentration reduction.   The vehicle alarm device according to claim 1, wherein the evaluation unit evaluates the concentration reduction by a driver's operation of an in-vehicle device.   The vehicle alarm device according to claim 1, wherein the detection unit detects an inter-vehicle distance as the travel state quantity.   The vehicle alarm device according to any one of claims 1, 2, and 3, wherein the detection means detects a curvature or a slope of a road as the travel state quantity.   The vehicle alarm device according to claim 1, wherein the detection unit detects a lateral displacement amount of the vehicle with respect to a travel lane as the travel state amount.

Images