Awakening level estimation system, awakening level estimation device, awakening level estimation method, and program

Abstract

To highly accurately estimate an arousal level of a subject with a small amount of calculation by using biological data.SOLUTION: An arousal level estimation system includes a biological sensor for continuously measuring prescribed biological data relating to an arousal level of a subject, an acquisition part for acquiring the biological data from the biological sensor, a normalization process part for executing a prescribed normalization process on the biological data, a feature amount calculation part for calculating a prescribed feature amount based on biological data in an individual difference adjustment calibration section out of the biological data after the normalization process, a threshold calculation part for calculating a threshold based on a predetermined regression equation and the feature amount, and an estimation part for estimating an arousal level of the subject based on the threshold and the biological data after the calibration section out of the biological data after the normalization process.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present invention relates to an arousal level estimation system, an arousal level estimation device, an arousal level estimation method, and a program.

Background Art

[0002] In recent years, in a vehicle driving support system or the like, a technique for acquiring biometric data of an occupant has been used. And, by applying various methods to the acquired biometric data, for example, the arousal level (degree of functioning of consciousness and sensation) of the occupant can be estimated.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the prior art, in order to obtain a highly accurate estimation result regarding the arousal level using biometric data, for example, it is necessary to use a large amount of data or perform frequency analysis, resulting in a problem of increased computational complexity.

[0005] Therefore, an embodiment of the present invention aims to provide an arousal level estimation system, an arousal level estimation device, an arousal level estimation method, and a program that can accurately estimate the arousal level of a target person with a small amount of computation using biometric data.

Means for Solving the Problems

[0006] An embodiment of the present invention provides an alertness estimation system comprising: a biosensor that continuously measures predetermined biological data related to the alertness level of a subject; an acquisition unit that acquires the biological data from the biosensor; a normalization processing unit that performs a predetermined normalization process on the biological data; a feature calculation unit that calculates a predetermined feature quantity based on the biological data in a calibration interval for adjusting for individual differences from the biological data after the normalization process; a threshold calculation unit that calculates a threshold based on a predetermined regression equation and the feature quantity; and an estimation unit that estimates the alertness level of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process. This configuration allows for the highly accurate estimation of a subject's level of arousal using biometric data, with minimal computational effort required for normalization, feature extraction, and threshold calculation.

[0007] Furthermore, the threshold calculation unit calculates a hyperarousal threshold based on the hyperarousal regression equation as the regression equation and the feature quantities, and calculates a low-arousal threshold based on the low-arousal regression equation as the regression equation and the feature quantities. The estimation unit estimates whether the subject's level of arousal is hyperarousal or not based on the hyperarousal threshold and the biological data after the calibration interval, and estimates whether the subject's level of arousal is low-arousal or not based on the low-arousal threshold and the biological data after the calibration interval. This configuration allows for the separate estimation of whether a subject's level of arousal is hyperarousal or hypoarousal.

[0008] Furthermore, the estimation unit sets a window interval of a predetermined time for the biological data after the calibration interval, calculates the average value of the numerical gradient of the approximation curve fitted to the biological data in the window interval, and repeats this series of processes while shifting the window interval. When the length of time during which the average value is 0 or less exceeds a predetermined threshold, the estimation unit estimates that the subject is drowsy. This configuration allows for highly accurate estimation of whether or not a subject is drowsy by using the average value of the numerical gradient of an approximation curve fitted to biological data.

[0009] Furthermore, the predetermined biological data is the skin electrical activity data of the subject. This configuration allows for the use of skin electrical activity data specifically as biological data.

[0010] Furthermore, the alertness estimation device of the embodiment of the present invention includes: an acquisition unit that acquires biological data from a biological sensor that continuously measures predetermined biological data related to the alertness level of a subject; a normalization processing unit that performs a predetermined normalization process on the biological data; a feature calculation unit that calculates a predetermined feature quantity based on the biological data in a calibration interval for adjusting for individual differences from the biological data after the normalization process; a threshold calculation unit that calculates a threshold based on a predetermined regression equation and the feature quantity; and an estimation unit that estimates the alertness level of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

[0011] Furthermore, the method for estimating the level of alertness according to an embodiment of the present invention includes an acquisition step of acquiring biological data from a biosensor that continuously measures predetermined biological data related to the level of alertness of a subject; a normalization step of performing a predetermined normalization process on the biological data; a feature calculation step of calculating a predetermined feature quantity based on the biological data in a calibration interval for adjusting for individual differences from the biological data after the normalization process; a threshold calculation step of calculating a threshold based on a predetermined regression equation and the feature quantity; and an estimation step of estimating the level of alertness of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

[0012] Furthermore, the program of the embodiment of the present invention causes the computer to function as an acquisition unit that acquires biological data from a biological sensor that continuously measures predetermined biological data related to the level of alertness of a subject; a normalization processing unit that performs predetermined normalization processing on the biological data; a feature calculation unit that calculates predetermined feature quantities based on the biological data in a calibration interval for adjusting for individual differences from the biological data after the normalization processing; a threshold calculation unit that calculates a threshold based on a predetermined regression equation and the feature quantities; and an estimation unit that estimates the level of alertness of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization processing. [Brief explanation of the drawing]

[0013] [Figure 1] Figure 1 is a schematic diagram showing the configuration of the vehicle system of the embodiment. [Figure 2] Figure 2 is a block diagram outlining the functional configuration of the vehicle system according to the embodiment. [Figure 3] Figure 3 shows an example of an EDA signal graph in an embodiment. [Figure 4] Figure 4 is a flowchart showing the hyperarousal estimation process by the information processing device of the embodiment. [Figure 5] Figure 5 is a flowchart showing the low-arousal estimation process by the information processing device of the embodiment. [Figure 6] Figure 6 shows an example of a graph of the EDA signal and numerical gradient in an embodiment. [Figure 7] Figure 7 shows an example of the EDA signal, approximation curve, and numerical gradient graphs during drowsiness estimation in the embodiment. [Figure 8] Figure 8 is a flowchart showing the drowsiness estimation process by the information processing device of the embodiment. [Modes for carrying out the invention]

[0014] Exemplary embodiments of the present invention will be disclosed below. The configurations of the embodiments shown below, as well as the actions, results, and effects brought about by such configurations, are examples. The present invention can be realized by configurations other than those disclosed in the following embodiments, and it is possible to obtain at least one of various effects and derivative effects based on the basic configuration.

[0015] FIG. 1 is a diagram schematically showing the configuration of a vehicle system 1 according to an embodiment. FIG. 2 is a block diagram showing an overview of the functional configuration of the vehicle system 1 according to the embodiment. In the vehicle system 1, as an example, the arousal level of the occupant 2 (target person) sitting on the seat 21 is estimated. The vehicle system 1 includes each configuration shown in FIGS. 1 and 2. Note that there are also configurations shown in FIG. 2 but omitted from the illustration in FIG. 1.

[0016] In the present embodiment, as predetermined biological data related to the arousal level of the occupant 2, electrodermal activity (EDA) data (hereinafter also referred to as an "EDA signal") will be described as an example.

[0017] The wearable device 12 continuously measures the EDA data of the occupant 2. The wearable device 12 is, for example, a smartwatch with an EDA data acquisition function. In the present embodiment, the wearable device 12 is used as a means for acquiring EDA data, but it is not limited thereto. EDA data may be acquired by a sensor other than the wearable device 12. For example, the EDA data of the occupant 2 may be acquired by a sensor disposed on the steering wheel.

[0018] The authentication device 13 is a device that acquires authentication information from an authentication tag or an ID (Identifier) card.

[0019] The vehicle information sensor 14 is a sensor that acquires various vehicle information. Examples of the vehicle information include steering information, accelerator operation information, brake operation information, vehicle speed information, vehicle interior temperature information, vehicle surrounding information, vehicle position information, and the like.

[0020] The display device 61 is a means for displaying various types of information, such as a liquid crystal display.

[0021] The audio equipment 62 is a means of generating various sounds, such as a speaker.

[0022] The fragrance device 63 is a device that diffuses a predetermined fragrance component inside the vehicle. By absorbing this fragrance component through the nose, lungs, skin, etc., occupants can regulate their autonomic nervous system, hormone balance, and the function of various organs.

[0023] The seat control mechanism 64 is a mechanism that changes the posture of the seat in which the occupant 2 sits and performs a massage function.

[0024] The drive mechanism 65 is a mechanism that drives the vehicle's power source (engine, motor, etc.) through the driver's accelerator operation or automatic drive.

[0025] The braking mechanism 66 is a mechanism that performs braking to decelerate and stop the vehicle through the driver's brake operation or automatic braking.

[0026] The steering mechanism 67 is a mechanism that changes the direction of travel of the vehicle through the driver's steering input or automatic steering.

[0027] The air conditioning mechanism 68 is a mechanism that performs air conditioning functions such as temperature control, airflow control, and ventilation.

[0028] Communication device 69 is a device that performs various communications, such as emergency notifications, to external devices.

[0029] The information processing device 5 is, for example, an ECU (Electronic Control Unit). The information processing device 5 may be implemented using an ECU for vehicle control, or it may be implemented using a different ECU. The information processing device 5 comprises a processing unit 51 and a storage unit 52. The processing unit 51 has the following functional configuration: an acquisition unit 511, a normalization processing unit 512, a calculation unit 513, an estimation unit 514, and a control unit 515.

[0030] The acquisition unit 511 acquires various information from other components. For example, the acquisition unit 511 acquires EDA data from the wearable device 12. The acquisition unit 511 also acquires authentication information from authentication tags and ID cards from the authentication device 13. The acquisition unit 511 also acquires various vehicle information from the vehicle information sensor 14. The acquisition unit 511 also acquires various information (such as data related to the occupants 2) from the cloud computer 7.

[0031] The normalization processing unit 512 performs a predetermined normalization process on the EDA data. Here, Figure 3 is an example of a graph of the EDA signal in the embodiment. For example, the normalization processing unit 512 performs the normalization process by subtracting the median value of the EDA data in the calibration interval for adjusting for individual differences from the EDA data. The calibration interval is the interval in the time-series EDA data that corresponds to a predetermined time (e.g., 5 minutes) during which the occupant 2 was instructed to rest.

[0032] Returning to Figure 2, the calculation unit 513 calculates various types of information. For example, the calculation unit 513 (feature calculation unit) calculates predetermined features based on the EDA data of the calibration interval from the EDA data after normalization processing.

[0033] Furthermore, the calculation unit 513 (threshold calculation unit) calculates the hyperarousal threshold (the value corresponding to the upper horizontal line in Figure 3) based on the hyperarousal regression equation (example of regression equation) and the features.

[0034] Furthermore, the calculation unit 513 (threshold calculation unit) calculates the low arousal threshold (the value corresponding to the lower horizontal line in Figure 3) based on the low arousal regression equation (example of a regression equation) and the features.

[0035] For example, let's represent skin electrical activity data as EDA, and the EDA within the calibration interval as EDAc. Then, from the EDAc within the calibration interval, we can calculate features such as the following. (Feature 1) Maximum value of EDAc (x1) (Feature 2) Minimum value of EDAc (x2) (Feature 3) Median of EDAc (x3) (Feature 4) Standard deviation of EDAc (x4)

[0036] Then, for example, the hyperarousal regression equation and the hypoarousal regression equation would be as follows. Hyperarousal regression formula: y high =-a1×x1-b1×x2+c1×x3+d1×x4+e1 Low arousal regression formula: y low =-a2×x1-b2×x2-c2×x3+d2×x4+e2

[0037] Here, the coefficients (a1, b1, c1, d1, e1, a2, b2, c2, d2, e2) are determined by learning from pre-acquired subject group data, and therefore are not fixed values.

[0038] These regression equations are examples only and are not limiting. For example, you do not need to use more than one of the above-mentioned features 1-4 as features. You may also use other features.

[0039] The estimation unit 514 estimates the level of alertness of crew member 2 based on a threshold and the EDA data after the calibration interval, which is part of the normalized EDA data.

[0040] For example, the estimation unit 514 estimates whether or not occupant 2 is in a hyperaroused state based on the hyperarousal threshold and the EDA data after the calibration interval.

[0041] Furthermore, the estimation unit 514 estimates whether or not occupant 2 is in a low-alert state based on the low-alertness threshold and the EDA data after the calibration interval.

[0042] Here, Figure 4 is a flowchart showing the hyperarousal estimation process by the information processing device 5 of the embodiment. In step S11, the acquisition unit 511 acquires EDA data from the wearable device 12. The normalization processing unit 512 performs a predetermined normalization process on the EDA data.

[0043] Next, in step S12, the calculation unit 513 calculates predetermined feature quantities based on the EDA data of the calibration interval from the EDA data after normalization processing.

[0044] Next, in step S13, the calculation unit 513 calculates the hyperarousal threshold based on the coefficient-learned hyperarousal regression equation and the features calculated in step S12.

[0045] Next, in step S14, the estimation unit 514 determines whether the EDA signal is above the hyperarousal threshold. If yes, the unit proceeds to step S15; otherwise, it proceeds to step S16.

[0046] In step S15, the estimation unit 514 estimates that crew member 2 is in a hyperaroused state, and the process returns to step S14.

[0047] In step S16, the estimation unit 514 estimates that crew member 2 is in a neutral state, and the process returns to step S14.

[0048] Next, Figure 5 is a flowchart showing the low-arousal estimation process by the information processing device 5 of the embodiment. In step S21, the acquisition unit 511 acquires EDA data from the wearable device 12. The normalization processing unit 512 also performs a predetermined normalization process on the EDA data.

[0049] Next, in step S22, the calculation unit 513 calculates predetermined feature quantities based on the EDA data of the calibration interval from the EDA data after normalization processing.

[0050] Next, in step S23, the calculation unit 513 calculates the low arousal threshold based on the coefficient-learned low arousal regression equation and the features calculated in step S22.

[0051] Next, in step S24, the estimation unit 514 determines whether the EDA signal is below the low arousal threshold. If yes, the unit proceeds to step S25; otherwise, it proceeds to step S26.

[0052] In step S25, the estimation unit 514 estimates that crew member 2 is in a low-awakening state, and the process returns to step S24.

[0053] In step S26, the estimation unit 514 estimates that crew member 2 is in a neutral state, and the process returns to step S24.

[0054] In this way, the estimation unit 514 can estimate whether or not the occupant 2 is in a hyperarousal or hypoarousal state. Note that the processes in Figures 4 and 5 are performed in parallel, for example.

[0055] Returning to Figure 2, the estimation unit 514 further sets a window interval of a predetermined time for the EDA data after the calibration interval, and calculates the average value of the numerical gradient of the approximation curve (for example, a quadratic polynomial approximation curve) fitted to the EDA data in the window interval, repeating this series of processes while shifting the window interval. Then, the estimation unit 514 estimates that the occupant 2 is drowsy when the length of time for which the average value is 0 or less exceeds a predetermined threshold.

[0056] Here, we use a sleepiness detection logic that determines a state of drowsiness when the EDA data value shows a continuous downward trend over a certain period of time. This will be explained using Figures 6 to 8.

[0057] Figure 6 shows an example of an EDA signal and numerical gradient graph in an embodiment. When there is an EDA signal as shown in Figure 6(a), the numerical gradient (difference) for every two adjacent points is as shown in Figure 6(b).

[0058] Figure 7 also shows an example of the EDA signal, approximation curve, and numerical gradient graphs during drowsiness estimation in the embodiment. Figure 7(a) shows the EDA data (code L1) for the entire interval. The vertical axis represents the EDA value, and the horizontal axis represents time. Drowsiness estimation is performed using this EDA data.

[0059] Figure 7(b) shows two minutes of EDA data (code L1), an approximation curve (code L2), and a numerical gradient (code L3). Here, the numerical gradient remains below 0 throughout.

[0060] Next, Figure 7(c) shows the EDA data (code L1), approximation curve (code L2), and numerical gradient (code L3) for 2 minutes, 5 seconds after Figure 7(b). Here, the numerical gradient shows an increasing trend, but remains below 0 throughout.

[0061] Next, Figure 7(d) shows the EDA data (code L1), approximation curve (code L2), and numerical gradient (code L3) for 2 minutes, 5 seconds after Figure 7(c). Here, the numerical gradient shows an increasing trend and changes from negative to positive along the way.

[0062] In this way, the estimation unit 514 sets a window interval of a predetermined time (for example, 120 seconds), calculates the average value of the numerical gradient of the approximation curve (for example, a quadratic polynomial approximation curve) fitted to the EDA data in the window interval, and repeats this series of processes while shifting the window interval by 5 seconds at a time. The estimation unit 514 then estimates that crew member 2 is drowsy when the length of time during which the average value is 0 or less exceeds a predetermined threshold (for example, 300 seconds).

[0063] Here, Figure 8 is a flowchart showing the drowsiness estimation process by the information processing device 5 of the embodiment. In step S31, the acquisition unit 511 acquires EDA data from the wearable device 12. The normalization processing unit 512 performs a predetermined normalization process on the EDA data.

[0064] Next, in step S32, the estimation unit 514 sets a window interval of 120 seconds and calculates a quadratic polynomial approximation curve from the EDA signals within the window interval. In this case, for example, the least squares method can be used.

[0065] Next, the estimation unit 514 calculates the average value of the numerical gradient derived from the difference between each of two neighboring points of the quadratic polynomial approximation curve.

[0066] Next, in step S34, the estimation unit 514 determines whether the average value is 0 or less. If yes, the process proceeds to step S35; otherwise, it proceeds to step S37.

[0067] In step S35, the estimation unit 514 determines whether the average value has been 0 or less for 300 seconds or more consecutively. If the answer is yes, the process proceeds to step S36; otherwise, the process proceeds to step S37.

[0068] In step S36, the estimation unit 514 estimates that crew member 2 is drowsy, and proceeds to step S38.

[0069] In step S37, the estimation unit 514 estimates that crew member 2 is not drowsy, and proceeds to step S38.

[0070] In step S38, the estimation unit 514 shifts the window interval by 5 seconds in the time series direction and returns to step S32.

[0071] In this way, the estimation unit 514 can estimate whether or not the crew member 2 is drowsy. Note that the process in Figure 8 is performed in parallel with, for example, Figures 4 and 5.

[0072] Returning to Figure 2, the control unit 515 performs various controls. For example, the control unit 515 controls the display device 61, sound equipment 62, fragrance equipment 63, seat control mechanism 64, drive mechanism 65, braking mechanism 66, steering mechanism 67, air conditioning mechanism 68, and communication equipment 69 according to the alertness level estimation result.

[0073] The control unit 515 displays warnings, status information, and driving suitability levels via the display device 61 according to the alertness level estimation results.

[0074] Furthermore, the control unit 515 generates warning sounds and the like using the sound equipment 62 according to the alertness level estimation result. For example, when the estimation unit 514 estimates that the occupant 2 is drowsy, the control unit 515 uses the sound equipment 62 to generate a warning message to the occupant 2 that there is a risk of falling asleep at the wheel.

[0075] Furthermore, the control unit 515 controls the seat control mechanism 64 according to the alertness level estimation result to change the seat's posture or perform a massage.

[0076] Furthermore, the control unit 515 controls the drive mechanism 65, braking mechanism 66, and steering mechanism 67 according to the alertness level estimation result, thereby generating a sensory warning, avoiding an accident, or stopping the vehicle on the shoulder of the road.

[0077] Furthermore, the control unit 515 controls the air conditioning system 68 to adjust the temperature and airflow, or to perform ventilation, according to the alertness level estimation result.

[0078] Furthermore, the control unit 515, for example, if it detects an abnormality occurring inside the vehicle, controls the communication device 69 according to the alertness level estimation result to send an emergency notification or registered user notification to a designated external organization.

[0079] Thus, according to the vehicle system 1 of this embodiment, the level of alertness of the occupant 2 can be estimated with high accuracy using skin electrical activity data, with minimal computational effort for normalization, feature calculation, and threshold calculation. For example, it does not require the use of large amounts of data or frequency analysis, thus reducing the computational effort.

[0080] Furthermore, whether crew member 2 is in a hyperarousal or hypoarousal state can be estimated separately by using different regression equations or thresholds.

[0081] Furthermore, by using the average value of the numerical gradient of the approximation curve fitted to the EDA data, it is possible to estimate with high accuracy whether or not occupant 2 is drowsy. For example, since driver drowsiness can lead to falling asleep at the wheel, it is possible to independently estimate the presence or absence of drowsiness, regardless of the estimation of hyperarousal or hypoarousal. If it is estimated that occupant 2 is drowsy, a warning message sound can be generated to alert occupant 2, for example, indicating the risk of falling asleep at the wheel.

[0082] Furthermore, it is possible to estimate a wide range of non-stationary emotions other than drowsiness on the arousal axis (e.g., excitement, surprise, etc.). Furthermore, environmental conditions such as the wearers' clothing have a relatively small impact on estimation accuracy.

[0083] Furthermore, since the threshold is set based on individual features within the calibration interval, individual differences in waveform characteristics are taken into account, eliminating the need to acquire prior data to determine the correct values.

[0084] Furthermore, it is possible to estimate the level of alertness with high accuracy and a relatively fast response rate based on biological data obtained solely from a standalone biological information acquisition device.

[0085] The program for realizing the functions of the above embodiment may be provided as a file in a format installable or executable format on the information processing device 5, recorded on a computer-readable recording medium such as a CD-ROM, flexible disk (FD), CD-R, or DVD (Digital Versatile Disk). Alternatively, the program may be stored on a computer connected to a network such as the Internet and provided by downloading it via the network. Furthermore, the program may be provided or distributed via a network such as the Internet.

[0086] Although embodiments of the present invention have been described above, these embodiments are presented as examples only and are not intended to limit the scope of the invention. This novel embodiment can be implemented in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. This embodiment and its variations are included in the scope and spirit of the invention, as well as in the claims of the invention and its equivalents.

[0087] For example, the biometric data used to estimate a subject's level of alertness is not limited to skin electrical activity data, but may also include other biometric data such as electroencephalography (EEG).

[0088] Furthermore, while the above example calculated and used separate hyperarousal and hypoarousal thresholds, this is not the only approach. For example, one could calculate and use only one threshold capable of distinguishing between hyperarousal and hypoarousal.

[0089] In addition, in parallel with estimating the level of arousal, it is also possible to estimate the subject's emotions using information such as heart rate data and respiratory data. [Explanation of Symbols]

[0090] 1...Vehicle system, 2...Occupant, 5...Information processing device, 7...Cloud computer, 12...Wearable device, 13...Authentication device, 14...Vehicle information sensor, 21...Seat, 51...Processing unit, 61...Display device, 62...Audio equipment, 63...Fragrance equipment, 64...Seat control mechanism, 65...Drive mechanism, 66...Braking mechanism, 67...Steering mechanism, 68...Air conditioning mechanism, 69...Communication equipment, 511...Acquisition unit, 512...Normalization processing setting unit, 513...Calculation unit, 514...Estimation unit, 515...Control unit

Claims

1. A biosensor that continuously measures predetermined biometric data related to the level of alertness of the subject, An acquisition unit that acquires the biological data from the biological sensor, A normalization processing unit that performs a predetermined normalization process on the aforementioned biological data, A feature calculation unit calculates predetermined features based on the biological data in the calibration interval for adjusting for individual differences, from the biological data after the normalization process, A threshold calculation unit calculates a threshold based on a predetermined regression equation whose coefficients have been learned through learning from subject group data, and the aforementioned features. An alertness estimation system comprising: an estimation unit that estimates the alertness level of a subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

2. The threshold calculation unit, Based on the hyperarousal regression equation as the regression equation and the feature quantities, the hyperarousal threshold is calculated. Based on the low arousal regression equation and the features, the low arousal threshold is calculated. The estimation unit, Based on the hyperarousal threshold and the biological data after the calibration interval, it is estimated whether the subject's level of arousal is hyperarousal or not. The arousal level estimation system according to claim 1, which estimates whether the arousal level of the subject is low or low based on the low arousal threshold and the biological data after the calibration interval.

3. The estimation unit further, The alertness estimation system according to claim 1, wherein, with respect to the biological data after the calibration interval, a window interval of a predetermined time is set, and the average value of the numerical gradient of the approximation curve fitted to the biological data in the window interval is calculated, and this series of processes is repeated while shifting the window interval, and when the length of time during which the average value is less than 0 becomes greater than or equal to a predetermined threshold, the system estimates that the subject is sleepy.

4. The arousal level estimation system according to claim 1, wherein the predetermined biological data is the skin electrical activity data of the subject.

5. An acquisition unit that acquires biological data from a biological sensor that continuously measures predetermined biological data related to the level of alertness of a subject, A normalization processing unit that performs a predetermined normalization process on the aforementioned biological data, A feature calculation unit calculates predetermined features based on the biological data in the calibration interval for adjusting for individual differences, from the biological data after the normalization process, A threshold calculation unit calculates a threshold based on a predetermined regression equation whose coefficients have been learned through learning from subject group data, and the aforementioned features. An alertness estimation device comprising: an estimation unit that estimates the alertness level of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

6. An acquisition step of acquiring biometric data from a biosensor that continuously measures predetermined biometric data related to the level of alertness of a subject, A normalization processing step which involves performing a predetermined normalization process on the aforementioned biological data, A feature calculation step in which predetermined features are calculated based on the biological data in the calibration interval for adjusting for individual differences from the biological data after the normalization process, A threshold calculation step in which a threshold is calculated based on a predetermined regression equation whose coefficients have been learned by learning from the subject group data, and the aforementioned features, A method for estimating the level of alertness of a subject, comprising: an estimation step of estimating the level of alertness of a subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

7. Computers, An acquisition unit that acquires biological data from a biological sensor that continuously measures predetermined biological data related to the level of alertness of a subject, A normalization processing unit that performs a predetermined normalization process on the aforementioned biological data, A feature calculation unit calculates predetermined features based on the biological data in the calibration interval for adjusting for individual differences, from the biological data after the normalization process, A threshold calculation unit calculates a threshold based on a predetermined regression equation whose coefficients have been learned through learning from subject group data, and the aforementioned features. A program for functioning as an estimation unit that estimates the level of alertness of the subject based on the threshold and the biological data after the calibration interval from the biological data after the normalization process.

Images