Drowsy driving alarm device

Abstract

Provided is a drowsy driving alarm device comprising a computer that calculates a frequency component of a travel trajectory of a host vehicle on a travel path and outputs an alarm instruction to the alarm device when a drowsiness evaluation value based on the frequency component exceeds a prescribed value. The computer: calculates the curvature radius of the travel path on the basis of the curvature of the travel path and the turning curvature of the host vehicle calculated from operation data of the host vehicle; determines whether the curvature radius is less than a threshold value; and corrects the frequency component when the curvature radius is less than the threshold value.

Description

Drowsy driving warning device

[0001] The present invention relates to a drowsy driving warning device.

[0002] A device is known that uses the lateral position of a vehicle in a lane as input, estimates the driver's level of alertness based on its frequency components, and issues an alert if the level of alertness falls below a certain value (see Patent Document 1, etc.).

[0003] Japanese Patent Application Laid-Open No. 2005-71184

[0004] With the introduction of the Driver Drowsiness and Attention Warning (DDAW) into law, there is a demand for improved accuracy of drowsy driving warning functions such as those described in Patent Document 1. On the other hand, if the sensitivity of the drowsy driving warning function is higher than necessary, it may be issued even when the driver's level of alertness is not low, resulting in a false alarm that differs from the driver's perception and leads to stress for the driver. Therefore, it is necessary to suppress the occurrence of such false alarms.

[0005] In Patent Document 1, taking into consideration that the steering operation performed by the driver on a zigzag road with a series of curves of different curvature directions, such as a mountain road, may match the frequency components characteristic of drowsiness, the system makes it difficult to issue an alarm when it is estimated that the vehicle is traveling on a zigzag road based on the steering angle while traveling, thereby reducing the occurrence of false alarms while traveling on a zigzag road.

[0006] However, when an analysis of scenes in which false alarms actually occur was conducted, it was found that false alarms are more likely to occur not only on zigzag roads but also on simple curved roads where the direction of curvature does not change, compared to when driving on a straight road, particularly when driving on or exiting a curved road.As a result of intensive research by the present inventors, it was found that the corrective steering that the driver makes to align the vehicle's driving trajectory with the curve of the road causes changes in the vehicle's lateral position within the lane to match the frequency components characteristic of drowsiness, which can increase the probability of false alarms occurring.

[0007] An object of the present invention is to provide a drowsy driving warning device that can suppress the occurrence of false warnings caused by the trajectory of the road.

[0008] In order to achieve the above object, the present invention provides a drowsy driving warning device that has a computer that calculates the frequency components of the vehicle's driving trajectory relative to a roadway and outputs an alarm command to a warning device when a drowsiness evaluation value based on the frequency components exceeds a predetermined value, wherein the computer calculates the radius of curvature of the roadway based on the curvature of the roadway and the turning curvature of the vehicle obtained from operating data of the vehicle, determines whether the radius of curvature is less than a threshold value, and corrects the frequency components if the radius of curvature is less than the threshold value.

[0009] According to the present invention, it is possible to suppress the occurrence of false alarms caused by the trajectory of the road.

[0010] Schematic diagram showing an example of a vehicle equipped with a drowsy driving warning device according to one embodiment of the present invention. Schematic diagram showing an example of the configuration of a drowsy driving warning device according to one embodiment of the present invention. Schematic diagram showing an example of the travel trajectory of a vehicle traveling on a straight road. Diagram showing an example of the distribution of frequency component values ​​related to the travel trajectory. Schematic diagram showing an example of the travel trajectory of a vehicle traveling on a curved road. Diagram showing an example of the relationship between the lateral position of the vehicle within the lane while traveling and the turning curvature of the travel trajectory of the vehicle. Diagram showing an example of setting the correction degree of frequency components with respect to the curvature radius of the travel road. Flowchart showing an example of one cycle of warning procedures by a drowsy driving warning device according to one embodiment of the present invention.

[0011] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0012] 1 is a schematic diagram showing an example of a vehicle equipped with a drowsy driving warning device according to one embodiment of the present invention. The vehicle 1 is an automobile or the like, and is configured to include a vehicle control system 2, a communication device 3, a drive device 5, a recognition device 6, an output device 7, an input device 8, and a notification device 9.

[0013] The vehicle control system 2 includes an in-vehicle network 10 and at least one computer 11 connected to the in-vehicle network 10. The in-vehicle network 10 is, for example, a Controller Area Network (CAN) or a CAN with Flexible Data-rate (CANFD). The computer 11 may include, for example, an Electronic Control Unit (ECU) 11A and a Map Positioning Unit (MPU) 11B. The vehicle control system 2 is connected to a communication device 3, a drive device 5, a recognition device 6, an output device 7, an input device 8, and a notification device 9, and communicates with these devices to exchange data.

[0014] The ECU 11A may include an ADAS-ECU (Advanced Driver Assistance Systems) and an AD-ECU (Autonomous Driving). The ADAS-ECU has a driving assistance function and a preventive safety function that implements vehicle speed limiting, brake assistance, etc. based on data from the recognition device 6 and the input device 8. The AD-ECU has an autonomous driving function for the vehicle 1.

[0015] The communication device 3 is a device that performs wireless communication with devices outside the vehicle 1, and includes, for example, a GNSS (Global Navigation Satellite System) mobile station that acquires data on the current position and direction of the vehicle 1, and other communication devices.

[0016] The drive device 5 is a device, such as an actuator, that drives mechanical and electrical equipment (e.g., motor, engine, transmission, wheels, brakes, steering device, etc.) related to the operation of the vehicle (vehicle 1) in response to control commands from the vehicle control system 2.

[0017] The recognition device 6 is a device that acquires data related to the outside world of the vehicle 1 and the state of the vehicle 1. The recognition device 6 includes, for example, external sensors such as an on-board camera (monocular camera or stereo camera), radar, LIDAR, and ultrasonic sensors, dynamic system sensors that recognize the state of the vehicle 1 (operating state, acceleration, vehicle speed, etc.), and other sensors. In this embodiment, in the process described below of recognizing signs and the like by image processing, a stereo camera is used as the on-board camera 21 ( FIG. 2 ) to acquire distance information, but distance measurement is also possible with other recognition devices 6 such as a monocular camera or LIDAR.

[0018] The output device 7 is connected to the in-vehicle network 10 by wire or wirelessly, and displays or outputs message information (e.g., video or sound) necessary for the driver based on data received via the in-vehicle network 10. The output device 7 includes, for example, a liquid crystal display, a warning light, a speaker, an alarm device, and other notification devices.

[0019] The input device 8 is a device for generating input signals that the driver uses to input operational intentions and instructions to the vehicle control system 2, and includes, for example, a steering wheel, accelerator pedal, brake pedal, shift lever, buttons, levers, touch panels, and other operating devices.

[0020] The notification device 9 is a device that the vehicle 1 uses to notify the outside world (following vehicles, surrounding people, etc.) of the state of the vehicle 1. The notification device 9 includes, for example, a lamp, an LED (Light Emitting Diode), a speaker, and other notification devices.

[0021] - Drowsy driving warning device - Figure 2 is a schematic diagram showing an example of the configuration of a drowsy driving warning device according to this embodiment. The drowsy driving warning device 100 has at least one computer that calculates the frequency components (Figure 4, described below) of the travel trajectory of the vehicle 1 (host vehicle) relative to a road recognized from road data of the road, and outputs an alert command to a warning device (for example, a speaker included in the output device 7) when a drowsiness evaluation value (described below) based on the frequency components exceeds a predetermined value. Note that the "road" refers to the road on which the vehicle 1 is traveling, and if the road has no lanes, it refers to the road itself on which the vehicle is traveling, or if the road has lanes, it refers to the lane on which the vehicle is traveling.

[0022] The drowsy driving warning device 100 has an external environment recognition function F1, a map interpretation function F2, and a drowsiness determination function F3. The drowsiness determination function F3 includes a frequency component calculation function F31, a curvature radius calculation function F32, a determination function F33, a correction function F34, an evaluation value calculation function F35, and an alert command function F36. In the example of FIG. 2 , the external environment recognition function F1 is installed in the computer of the vehicle-mounted camera 21, the map interpretation function F2 and the map database DB are installed in the MPU 11B, and the drowsiness determination function F3 is installed in the ECU 11A. However, the location of each function and database can be changed as appropriate, for example, by providing the drowsiness determination function F3 in the computer of the vehicle-mounted camera 21. In some cases, a single computer may have the external environment recognition function F1, the map interpretation function F2, and the drowsiness determination function F3.

[0023] The external environment recognition function F1 is a function that recognizes the external environment from images captured by the on-board camera 21. In this embodiment, the on-board camera 21 is, for example, a stereo camera included in the recognition device 6 ( FIG. 1 ). The on-board camera 21 is equipped with a computer, and the external environment recognition function F1 recognizes the road on which the vehicle is currently traveling based on camera images captured in real time while the vehicle is traveling.

[0024] The map interpretation function F2 is a function for acquiring road shapes from the map database DB based on the current position of the vehicle 1 (host vehicle) input from the GNSS mobile station (communication device 3). The MPU 11B uses the map interpretation function F2 to acquire map data around the vehicle 1 based on the current position of the vehicle 1 and acquires road shape data for the road on which the vehicle is currently traveling. The MPU 11B may also be provided with a function for communicating with a server via the communication device 3 and updating the data in the map database DB.

[0025] In this way, at least one computer constituting the drowsy driving warning device 100 uses at least one of the camera image and map data as roadway data and recognizes the roadway based on this roadway data.

[0026] The drowsiness determination function F3 is a function that determines the alertness level of the driver driving the vehicle 1 and commands an alarm device (e.g., a speaker included in the output device 7) to issue an alert (e.g., output an alarm sound) when drowsiness is detected. As described above, the drowsy driving alarm device 100 has a function that calculates the frequency components of the travel trajectory of the vehicle 1 relative to the travel road and outputs an alert command to the alarm device when a drowsiness evaluation value based on the frequency components exceeds a predetermined value. Among these functions, the frequency component calculation function F31 is a function that calculates the frequency components of the travel trajectory of the vehicle 1 (host vehicle) relative to the travel road recognized from the travel road data obtained by the external environment recognition function F1 and the map interpretation function F2. In addition, the evaluation value calculation function F35 is a function that calculates a drowsiness evaluation value based on the frequency components, and the alarm command function F36 is a function that determines whether the calculated drowsiness evaluation value exceeds a predetermined value and outputs an alert command to the alarm device when the drowsiness evaluation value exceeds the predetermined value.

[0027] The curvature radius calculation function F32 is a function that calculates the curvature radius of the roadway while the vehicle 1 (host vehicle) is traveling, based on the curvature of the roadway and the turning curvature of the travel trajectory of the vehicle 1 (host vehicle) calculated from operational data of the vehicle 1 (host vehicle) (e.g., steering angle and yaw rate). The steering angle can be calculated, for example, from the steering operation amount included in the input device 8. The yaw rate can be calculated, for example, from the rate of change of the azimuth angle of the vehicle 1 obtained from a GNSS mobile station included in the communication device 3. In this way, at least one computer constituting the drowsy driving warning device 100 calculates the turning curvature of the vehicle 1 (host vehicle) using at least one of the steering angle and yaw rate as operational data of the vehicle 1 (host vehicle), and calculates the turning curvature of the travel trajectory of the vehicle 1 using at least one of the camera image and map data as roadway data. By using the turning curvature of the driving trajectory, the determination function F33 of the subsequent functions can determine whether the vehicle will enter a curve in the near future, and the correction function F34 of the subsequent functions can appropriately determine the correction amount before entering or when starting to enter a curve, thereby preventing false alarms and an increase in the cumulative value. The curvature calculated here may be the curvature at the time of frequency component calculation or a statistical value (average, median, maximum, etc.) over a short-term fixed period of time or during recording of the driving trajectory of the vehicle 1. The curvature radius calculation function F32 calculates the curvature of the driving path (e.g., lane) relative to the vehicle 1 from the camera image, corrects this with the turning curvature of the vehicle 1 based on the operating data, and calculates the absolute driving path trajectory, and then calculates the radius of curvature.

[0028] The determination function F33 is a function for determining whether the radius of curvature of the road calculated by the radius of curvature calculation function F32 is less than a threshold value. The threshold value is set in advance for the radius of curvature and is stored, for example, in the memory of the ECU 11A (or another storage device). Although a single threshold value may be used to distinguish between straight roads and curved roads, in this embodiment, multiple threshold values ​​with different values ​​are set to determine the degree of curvature of a curved road (described later).

[0029] The correction function F34 corrects the frequency components that form the basis of the drowsiness evaluation value when the determination function F33 determines that the curvature radius of the road is less than a threshold. The evaluation value calculation function F35 calculates the drowsiness evaluation value based on the corrected frequency components when the frequency components are corrected by the correction function F34. When the frequency components are not corrected by the correction function F34, the evaluation value calculation function F35 calculates the drowsiness evaluation value based on the uncorrected frequency components. In the correction function F34, at least one computer constituting the drowsy driving warning device 100 corrects the frequency components by multiplying them by a coefficient of 1 or less. In this case, the at least one computer constituting the drowsy driving warning device 100 can change the degree of correction of the frequency components depending on the curvature radius of the road (when a coefficient is used, the magnitude of the coefficient can be changed depending on the curvature radius). Typically, the correction can be configured so that the smaller the curvature radius of the road, the greater the degree of reduction of the frequency components. A typical example is to set the coefficient α1 (0<α1<1) when the radius of curvature of the curved road is relatively large, and to set the coefficient α2 smaller (0<α2<α1) when the radius of curvature is relatively small.

[0030] - Drowsiness Detection - (1) When Driving on a Straight Road FIG. 3 is a schematic diagram showing an example of the driving trajectory of the vehicle 1 while driving on a straight road, and FIG. 4 is a diagram showing an example of the distribution of frequency component values ​​related to the driving trajectory. The frequency component calculation function F31 calculates in real time the lateral position of the vehicle 1 within the driving lane L while driving (e.g., the distance between the center line C of the driving lane L and the vehicle 1), and calculates the driving trajectory T of the vehicle 1. The lateral position of the vehicle 1 can be calculated, for example, by determining the distances from the vehicle 1 to the left and right dividing lines (white lines) of the driving lane L based on a camera image. The waveform represented by the driving trajectory T is decomposed into frequency components at predetermined time intervals t1, for example, by Fourier transform. In this manner, the frequency component calculation function F31 calculates the frequency components related to the driving trajectory T.

[0031] Furthermore, the frequency component values ​​(amplitude for each frequency) are accumulated for each predetermined time t1, resulting in a frequency distribution as shown in the histogram of FIG. 4. A specific frequency among the frequencies in this frequency distribution is the drowsiness frequency f (FIG. 4). The drowsiness frequency f is a characteristic value that appears when the driver is driving the vehicle 1 in a drowsy state. The evaluation value calculation function F35 described above accumulates the frequency components related to the driving trajectory T for each predetermined time t1, and calculates the ratio of the maximum value of the drowsiness frequency f (accumulated value) (the accumulated value of the drowsiness frequency on the left side in FIG. 4) to the average value Ave of each frequency component (accumulated value) as the drowsiness evaluation value.

[0032] The alarm issuing command function F36 compares the sleepiness evaluation value calculated successively with a preset value, and outputs an alarm issuing command to the alarm device when the sleepiness evaluation value exceeds the preset value.

[0033] As shown in FIG. 3, when the vehicle 1 is traveling on a straight road, the above algorithm can accurately detect the driver's drowsiness and output a warning.

[0034] (2) When traveling on a curved road Fig. 5 is a schematic diagram showing an example of a travel locus T of the vehicle 1 traveling on a curved road, and Fig. 6 is a diagram showing an example of the relationship between the lateral position of the vehicle 1 in the travel lane L while traveling and the turning curvature of the travel locus T of the vehicle 1. Fig. 6 is an example of data measured when the driver was not drowsy and had a high level of alertness.

[0035] When traveling on a curved road, the drowsy driving warning device 100 calculates the drowsiness evaluation value in the same way as when the vehicle 1 travels on a straight road, and outputs an alarm command to the warning device when the drowsiness evaluation value exceeds a predetermined value.

[0036] However, when traveling on a curved road, the driver's steering corrections made to align the vehicle 1's travel path T with the curve of the travel lane L cause the change in the lateral position of the vehicle 1 within the travel lane L to match the frequency components characteristic of drowsiness, which may increase the probability of false alarms compared to when traveling on a straight road. As shown in the example of Figure 6, when traveling on a straight section (time periods in the left and right regions of the figure), the driver's steering operation is low in frequency and amount, so the turning curvature (dotted line) of the vehicle 1's travel path T is stable at around 0. In contrast, when traveling on a curved section (time period in the center region of the figure), the driver is constantly operating the steering wheel and the amount of operation is adjusted successively, so the turning curvature (dotted line) of the vehicle 1's travel path T increases and the value changes oscillates.

[0037] As a result, as shown by the solid line in the figure, the lateral position of the vehicle 1 in the driving lane L changes more gradually when driving on a curved section than when driving on a straight section, and even though the driver is not feeling drowsy, the characteristics of driving on a straight road while feeling drowsy may appear. Therefore, in the above-mentioned configuration in which drowsiness is detected by simply calculating an evaluation value based on frequency components related to the driving trajectory TT, false alarms are more likely to occur when driving on a curved road than when driving on a straight road.

[0038] Therefore, in this embodiment, as described above, the radius of curvature of the road is calculated, and if the radius of curvature of the road is less than the threshold value and it is determined that the vehicle is traveling on a curved road, the frequency components calculated during that time are corrected to have smaller values.

[0039] 7 is a diagram showing an example of setting the degree of correction of frequency components with respect to the radius of curvature of the road. In FIG. 7, an example is shown in which the degree of correction of frequency components is changed in stages according to the degree of curvature of the curved road.

[0040] First, if the calculated radius of curvature R of the driving lane L is greater than the first threshold R1, which is set sufficiently large, and the lane L can be regarded as a straight road, the frequency components are not corrected. In other words, in this case, the correction amount of the frequency components is 0%, and if the correction coefficient is α0, the correction coefficient α0=1.

[0041] Next, if the radius of curvature R of the travel lane L is equal to or less than the first threshold R1 but equal to or greater than a second threshold R2 (R1<R2) that is set smaller than the first threshold R1, and the road is determined to be a gently curving road, the frequency components are corrected downward by a relatively small correction amount X%. In this case, the correction coefficient α1 (0<α1<1) is α1=(100-X) / 100.

[0042] Next, if the radius of curvature R of the driving lane L is equal to or less than the second threshold R2 and the road is determined to be a curved road with a curvature of a predetermined value or greater, the correction amount Y% for the frequency components is set to be larger than the correction amount X% for a gentle curve. In this case, the correction coefficient α2 (0<α2<α1) is α2=(100-Y) / 100.

[0043] By setting such a correction coefficient, when traveling on a curved road, the frequency components related to the traveling locus T are reduced in accordance with the radius of curvature of the curved road.

[0044] -Warning Procedure- Figure 8 is a flowchart showing an example of one cycle of the warning procedure by the drowsy driving warning device 100. While the vehicle 1 is being driven, the drowsy driving warning device 100 repeatedly executes the flow of Figure 8 at predetermined time intervals (for example, the predetermined time t1 shown in Figure 3).

[0045] When the flow of FIG. 8 starts, the drowsy driving warning device 100 first performs frequency analysis on the travel path T of the vehicle 1 for a predetermined time t1 up to the present time, and calculates component values ​​for each frequency during that time (step S11).

[0046] Next, the drowsy driving warning device 100 calculates the radius of curvature R of the lane L on which the vehicle 1 is currently traveling (step S12), and determines the amount of correction (e.g., a coefficient) for the frequency component (step S13) in accordance with the radius of curvature R. Here, as described in FIG. 7 , if the traveling road is a straight road, the correction coefficient is set to 1, and if the traveling road is a curved road, the coefficient is set to less than 1 (>0).

[0047] After determining the correction amount according to the curvature radius R of the road in step S13, the drowsy driving warning device 100 corrects the frequency component calculated for the predetermined time t1 calculated in step S11 by the determined correction amount to lower the value (step S14). Note that setting the correction amount to 0 (coefficient to 1) in step S13 is synonymous with not correcting the frequency.

[0048] After the frequency components have been corrected as necessary in step S14, the drowsy driving warning device 100 calculates the drowsiness evaluation value as described above (step S15) and determines whether the evaluation value is equal to or greater than a predetermined value (step S16). If the evaluation value is less than the predetermined value, the drowsy driving warning device 100 completes the procedure in Fig. 8 without outputting an alarm activation command to the alarm device, and then repeats the flow in Fig. 8. If the evaluation value is equal to or greater than the predetermined value, the drowsy driving warning device 100 outputs an alarm activation command to the alarm device to activate the alarm (step S17), and then repeats the flow in Fig. 8.

[0049] Effect (1) According to this embodiment, the curvature radius R of the road is calculated and a determination is made as to whether the curvature radius R is equal to or less than the first threshold value R1. If the curvature radius R is less than the first threshold value R1 and the road is determined to be a curved road, the frequency components related to the travel path T of the vehicle 1 are corrected and subtracted. This ensures that the steering corrections made by the driver to trace the road while traveling on a curved road coincide with specific frequency components characteristic of drowsiness, thereby preventing an unintended increase in the drowsiness evaluation value. This makes it more difficult to determine drowsiness when traveling on a curved road, and the level of determination of drowsiness can be consistent between traveling on a curved road and traveling on a straight road. This reduces the occurrence of false alarms due to the trajectory of the road (whether it is a straight road or a curved road), thereby reducing driver stress caused by false alarms. Furthermore, according to this embodiment, the occurrence of false alarms can be reduced and the warning output by the warning device can be accurately controlled regardless of the direction of the curve of the road, and not limited to situations where curves of different directions are repeated.

[0050] (2) The travel path of the vehicle 1 can be recognized not only from camera images captured by the onboard camera 21 but also from map data acquired by the MPU 11B. For example, when recognizing a travel path based on camera images, particularly when a stereo camera is used as the onboard camera 21, road shapes and the like can also be recognized from the images. Therefore, even on roads without clear dividing lines (white lines), it is possible to detect roadside bumps and the like and recognize the trajectory of the travel path. On the other hand, when map data is used, the computational load on the computer is reduced because no image processing is required. Furthermore, as the road shape data included in the map data is updated and enriched, correction accuracy can be improved. Of course, these methods can also be combined. For example, by reconciling travel path recognition based on camera images and travel path recognition based on road shapes acquired from map data, it is possible to recognize the travel path from the map data when, for example, the travel path cannot be recognized from camera images. Furthermore, the results of both recognitions can be compared to recognize the travel path with higher accuracy.

[0051] (3) By correcting the frequency components related to the travel locus T of the vehicle 1 by multiplying them by a coefficient of 1 or less, the frequency components can be corrected simply and rationally.

[0052] (4) By changing the degree of correction of frequency components according to the radius of curvature of the road (for example, the smaller the radius of curvature of the road, the greater the degree of reduction in frequency components), it is possible to accurately correct frequency components according to the situation, compared to when frequency components are uniformly corrected when driving on a curved road, and ultimately to calculate a drowsiness evaluation value that is highly valid.

[0053] -Modifications- Note that the present invention is not limited to the above-described embodiment, and includes various modifications. For example, the above-described embodiment has been specifically described in order to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to an embodiment having all of the described configurations.

[0054] Furthermore, a part of the configuration of one embodiment can be replaced with a part of the configuration of another embodiment. Also, a part of the configuration of another embodiment can be added to the configuration of one embodiment. Also, a part of the configuration of each embodiment can be deleted, and a part of another configuration can be added or replaced with a part of another configuration.

[0055] The correction coefficient does not have to be a fixed value; for example, it can be set in advance as a function with the radius of curvature of the road as a variable. In this case, the degree of frequency correction can be controlled more precisely according to the radius of curvature of the road, which is expected to improve the effectiveness of suppressing false alarms. Also, a configuration can be adopted in which the relationship between the radius of curvature and the appropriate degree of correction is learned by machine learning, and the degree of correction is determined according to the radius of curvature based on the results of the machine learning.

[0056] 1...vehicle (own vehicle), 7...output device (warning device), 11...computer, 100...drowsy driving warning device, L...lane (driving path), R...radius of curvature, R1...first threshold, R2...second threshold, T...driving trajectory, X, Y...correction amount (degree of reduction)

Claims

1. A drowsy driving warning device having a computer that calculates frequency components of a driving trajectory of a vehicle relative to a road and outputs an alarm command to a warning device when a drowsiness evaluation value based on the frequency components exceeds a predetermined value, wherein the computer calculates a radius of curvature of the road based on the curvature of the road and a turning curvature of the vehicle calculated from operating data of the vehicle, determines whether the radius of curvature is less than a threshold value, and corrects the frequency components if the radius of curvature is less than the threshold value.

2. A drowsy driving warning device according to claim 1, wherein the computer recognizes the driving route based on at least one of camera images and map data.

3. A drowsy driving warning device according to claim 1, wherein the computer calculates the turning curvature of the vehicle using at least one of the steering angle and the yaw rate as the operational data of the vehicle.

4. A drowsy driving warning device according to claim 1, wherein the computer corrects the frequency components by multiplying the frequency components by a coefficient of 1 or less.

5. A drowsy driving warning device according to claim 1, wherein the computer changes the degree of correction of the frequency components in accordance with the radius of curvature of the road.

6. A drowsy driving warning device according to claim 5, wherein the computer corrects the frequency components so that the degree of reduction increases as the radius of curvature of the road becomes smaller.

Images