JP6209396B2 - Driving support device and computer program - Google Patents

Description

  The present invention relates to a driving support apparatus and a computer program that, based on information about a driver's biological state, induces a driver to wake up and outputs a stimulus that maintains the wakeful state.

  In Patent Document 1, there is an intermediate process such as a quasi-awakening state and a wakefulness-decreasing state in the process of transitioning from an awakening state to a sleeping state. No.) is disclosed, and if the awakening rhythm is expressed, a technique is disclosed in which a stimulus such as a sound is applied to induce the awakening state.

  On the other hand, the present applicant, in Patent Documents 2 and 3, etc., obtains a time series waveform of a frequency from a time series waveform of a biological signal that is mainly a wave of the cardiovascular system collected from a human upper body, and further, a frequency slope An apparatus having means for determining a human state by obtaining a time series waveform of the above and a time series waveform of frequency fluctuations and analyzing the frequency of these time series waveforms is disclosed.

Japanese Patent No. 2500378 WO2010 / 123125 publication WO2011 / 007886 Publication

  According to the technique disclosed in Patent Document 1, once a sound stimulus is applied after a decrease in arousal, a sound stimulus in the range of 5 seconds to 3 minutes is presented at a predetermined interval, preferably 1 minute. It is given intermittently, such as pausing for minutes, to induce wakefulness. However, in Patent Document 1, as described above, a predetermined pattern of sound stimulation is applied, and when it continues for several minutes to several tens of minutes, a person is used to it and does not necessarily promote arousal. However, it may be a factor that causes a decrease in arousal.

  Further, the technique of Patent Document 1 is to detect the above-mentioned wakefulness rhythm by measuring an electroencephalogram, and although it can be put into practical use as a device for preventing dozing during work in a factory or the like, a driver of an automobile or the like. It is not suitable as a device for maintaining the awakening state. On the other hand, the means shown in Patent Documents 2 and 3 by the applicant of the present application is to eliminate a pulse wave (Aortic Pulse Wave (APW)) that is a vibration generated on the body surface of the driver's back. It can be detected by restraint and is an excellent means for obtaining biological information during driving. However, the techniques disclosed in Patent Documents 2 and 3 are intended to detect a driver's sleep symptom phenomenon, an imminent sleep phenomenon, and the like. Among these, the impending sleep phenomenon is a phenomenon that appears at a timing several minutes before the onset of sleep, and in many cases is a sign that can be recognized as strong sleepiness, but Patent Documents 2 and 3 are more accurate than these phenomena. Therefore, the detection timing is set so that the phenomenon appears remarkably. Therefore, even when the warning is activated when the imminent sleep phenomenon is detected, the driver already feels strong sleepiness at that time, and depending on the warning method, it may be difficult to induce wakefulness.

  The present invention has been made in view of the above points, and induces wakefulness by giving a stimulus at a predetermined timing before recognizing strong sleepiness such as during an imminent sleep phenomenon, thereby further improving the biological state of the driver. It is an object of the present invention to provide a driving support device and a computer program that can be maintained in a state suitable for driving and that can measure a driver's biological state without restriction.

In order to solve the above problems, the driving support apparatus of the present invention analyzes a back body surface pulse wave, which is a driver's biological signal measured by a biological signal measuring device, and determines the biological state of the driver. A driving support device comprising means,
Having a stimulus applying means for applying a stimulus including at least one of an auditory, visual or olfactory stimulus to the driver;
The biological state determination means includes
From the index obtained by analysis of the back body surface pulse wave, it is determined a biological state that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness, and depending on the determination result, the stimulus applying means Determination output means for outputting a command to drive,
The determination output means detects the biological state determined to be a warning level that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness in order to induce awakening of the driver, and this It is characterized in that a command for driving the stimulus applying means is output at both timings at which relevant signs related to a decrease in arousal level appearing before the appearance of a warning level biological state are detected.

  The determination output means provides the stimulus for stimulating the driver to take a break when the time difference between the detection timing of the biological state at the warning level and the detection timing of the related sign is determined to be within a predetermined time. It is preferable that it is the structure which commands to a means. It is preferable that the stimulus applying unit includes both a sound output device that applies an auditory stimulus and an image display device that applies a visual stimulus. The image display device can display a change in each biological state in a normal state, and when applying a stimulus for prompting arousal or resting according to a command from the determination output unit, a plurality of images are displayed. It is preferable that the display is set alternately so as to give a visual stimulus. Three to ten types of images for applying visual stimuli set in the image display device are set corresponding to each biological state, and two to four images are alternately displayed for each type. It is preferable that the setting is made as described above.

The determination output means includes
Zero-cross detection means for obtaining a time-series waveform of the frequency using a zero-cross point in the time-series waveform of the back body surface pulse wave obtained from the biological signal measuring device;
Peak detection means for obtaining a time series waveform of frequency using a peak point in a time series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
Zero-cross slope time-series waveform calculating means for calculating the time-series waveform of the frequency by sliding calculation of the time-series waveform of the frequency obtained by the zero-cross detection means;
A peak slope time series waveform calculating means for obtaining a slope time series waveform of the frequency by sliding calculation of the time series waveform of the frequency obtained by the peak detection means;
Based on at least one index of the slope time series waveform obtained from the zero cross slope time series waveform computing means and the slope time series waveform computing means obtained from the peak slope time series waveform computing means, various types appearing with a decrease in the arousal level It is preferable that it is the structure which determines a biological condition.
The determination output means includes
Function analysis in the range of 0.001 to 0.0027 Hz is performed by frequency analysis of the slope time series waveform obtained from the zero cross slope time series waveform computing means or the slope time series waveform obtained from the peak slope time series waveform computing means. A signal, a fatigue acceptance signal in the range of 0.002 to 0.0052 Hz, and a frequency component corresponding to an activity adjustment signal of 0.004 to 0.007 Hz are extracted, and a distribution for obtaining a variation in the distribution rate of each of these frequency components It is preferable to have a configuration that includes rate calculation means and determines various biological states that appear as the arousal level decreases from the distribution rate of each frequency component obtained by the distribution rate calculation means.

The determination output means is configured such that, in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation means, the distribution ratio of the frequency component corresponding to the activity adjustment signal is the function adjustment. When the distribution ratio of the frequency component corresponding to the signal is exceeded, it is preferable that the related indication is determined as a related sign related to the decrease in the arousal level.
The determination output means includes a distribution ratio of each frequency component obtained by using the slope time series waveform obtained from the zero cross slope time series waveform computing means, and each frequency component obtained by using the peak slope time series waveform computing means. When the distribution ratios of the corresponding frequency components are compared with each other, and the two tend to deviate from each other, the awakening that appears due to the divergence between the activity of the sympathetic nerve function and the activity of the parasympathetic nerve function The configuration is preferably determined to be a related sign related to a decrease in degree.
The determination output means is a frequency at which the distribution ratio of the function adjustment signal corresponds to the activity adjustment signal in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation means. When the state of exceeding the component distribution rate continues for a predetermined time or more, or when the state of distribution of the activity adjustment signal exceeding the frequency component distribution rate corresponding to the function adjustment signal continues for a predetermined time or more It is preferable that it is the structure determined as the related symptom relevant to the fall of the arousal level.
When the amplitude of at least one of the slope time series waveform obtained from the zero cross slope time series waveform computing means and the slope time series waveform obtained from the peak slope time series waveform computing means is a predetermined value or more, the determination output means Means for determining whether the driver is in an uplift state;
When it is determined that the uplifting state has ended after it has been determined that the uplifted state, it is preferable that a command for prompting the driver to take a break is output to the stimulus applying means.
The determination output means further includes means for detecting a sleep onset sign phenomenon or an imminent sleep phenomenon,
It is preferable that when the sleep onset sign phenomenon or the impending sleep phenomenon is detected after being determined to be in the uplift state, a command for prompting the driver to take a break is output to the stimulus applying unit.
Furthermore, it has a subjective fatigue identification means for identifying the timing when the driver felt subjective fatigue,
The determination output means is means for determining whether or not a time difference between a timing at which subjective fatigue specified by the subjective fatigue specification means is felt and a stimulus output timing by the stimulus applying means is within a predetermined range. Have
It is preferable that when the time difference is determined to be within a predetermined range, a command for prompting the driver to take a break is output to the stimulus applying unit.
Preferably, the subjective fatigue identification means is an input means that is operated when the driver subjectively feels fatigue, a decrease in attention, or a decrease in arousal level including drowsiness.
A generalized probability table of index fluctuations obtained by analysis of the back body surface pulse wave and biological states appearing corresponding to the index fluctuations is stored in advance in the storage unit,
The determination output means is configured to determine each biological state in light of the probability table read by accessing the storage unit and reading the change in the driver's index obtained by analyzing the back body surface pulse wave It is preferable that In the probability table, it is preferable to determine whether or not the index variation corresponds to an appearance condition of the highest probability value regarding a predetermined biological state, and to determine that the biological state corresponds to the condition.

The computer program of the present invention is stored in a computer as a driving support device.
A computer program that analyzes a back body surface pulse wave, which is a driver's biological signal measured by a biological signal measuring device, and executes a biological state determination procedure for determining the biological state of the driver,
As the biological state determination procedure,
From the index obtained by the analysis of the back body surface pulse wave, it is determined a biological state that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness, and depending on the determination result, auditory, visual or It is configured to execute a determination output procedure for outputting a command for driving a stimulus applying unit that applies a stimulus including at least one of olfactory stimuli to the driver,
The timing at which the determination output procedure detects a biological state that is determined to be a warning level that appears in association with fatigue, a decrease in alertness or a decrease in arousal level including drowsiness in order to induce arousal to the driver, and this A command for driving the stimulus applying means is output at both timings at which related signs related to a decrease in arousal level appearing before the appearance of a warning level biological state are detected.

In the determination output procedure, when the time difference between the detection timing of the biological state at the warning level and the detection timing of the related sign is determined to be within a predetermined time, the stimulation is applied to the stimulus for prompting the driver to take a break. It is preferable to command the means.
The determination output procedure includes:
Zero-cross detection procedure for obtaining a time-series waveform of a frequency using a zero-cross point in a time-series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
A peak detection procedure for obtaining a time series waveform of a frequency using a peak point in a time series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
A zero-cross slope time-series waveform calculation procedure for calculating a time-series waveform of the frequency by sliding the time-series waveform of the frequency obtained by the zero-cross detection procedure;
Performing a peak slope time series waveform calculation procedure to obtain a slope time series waveform of the frequency by sliding calculation of the time series waveform of the frequency obtained by the peak detection procedure;
Based on at least one index of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure and the slope time series waveform computation procedure obtained from the peak slope time series waveform computation procedure, various types appearing with a decrease in the arousal level It is preferable to determine the biological state.
The determination output procedure includes:
Functional adjustment in the range of 0.001 to 0.0027 Hz by frequency analysis of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure or the slope time series waveform obtained from the peak slope time series waveform computation procedure. A signal, a fatigue acceptance signal in the range of 0.002 to 0.0052 Hz, and a frequency component corresponding to an activity adjustment signal of 0.004 to 0.007 Hz are extracted, and a distribution for obtaining a variation in the distribution rate of each of these frequency components It is preferable to execute a rate calculation procedure and determine various biological states that appear as the arousal level decreases from the distribution rate of each frequency component obtained by the distribution rate calculation procedure.
In the determination output procedure, in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation procedure, the distribution ratio of the frequency component corresponding to the activity adjustment signal is the function adjustment. When the distribution ratio of the frequency component corresponding to the signal is exceeded, it is preferable to determine the related sign related to the decrease in the arousal level.
The determination output procedure includes a distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform computation procedure and each frequency component obtained using the peak slope time series waveform computation procedure. When the distribution ratios of the corresponding frequency components are compared with each other, and the two tend to deviate from each other, the awakening that appears due to the divergence between the activity of the sympathetic nerve function and the activity of the parasympathetic nerve function Preferably, it is determined as a related symptom associated with a decrease in degree.
In the determination output procedure, a distribution rate of each frequency component obtained using the slope time series waveform obtained from the zero-cross slope time series waveform calculation procedure is a frequency at which the distribution rate of the function adjustment signal corresponds to the activity adjustment signal. When the state of exceeding the component distribution rate continues for a predetermined time or more, or when the state of distribution of the activity adjustment signal exceeding the frequency component distribution rate corresponding to the function adjustment signal continues for a predetermined time or more It is preferable to determine the relevant sign related to the decrease in the degree of arousal.
The determination output procedure, when at least one of the amplitude of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure and the slope time series waveform computation procedure obtained from the peak slope time series waveform computation procedure is a predetermined value or more, The procedure for determining whether or not the driver is in an uplift state is performed. After the determination that the driver is in an uplifted state, if it is determined that the uplifted state has ended, the instruction to prompt the driver to take a break is applied. It is preferable to output to the means.
The determination output procedure further executes a procedure of detecting a sleep onset symptom phenomenon or an imminent sleep phenomenon,
It is preferable to output, to the stimulus applying means, a command for prompting the driver to take a break when the sleep onset symptom phenomenon or the impending sleep phenomenon is detected after it is determined to be in the uplift state.
The determination output procedure includes a timing at which subjective fatigue is identified by a subjective fatigue identification unit that identifies a timing at which the driver feels subjective fatigue, and a stimulus output timing by the stimulus applying unit. Execute a procedure to determine whether the time difference is within a predetermined range;
When it is determined that the time difference is within a predetermined range, it is preferable that a command for prompting the driver to take a break is output to the stimulus applying unit.
A generalized probability table of index fluctuations obtained by analysis of the back body surface pulse wave and biological states appearing corresponding to the index fluctuations is stored in advance in a storage unit of a driving support device that is a computer,
The determination output procedure is to determine each of the biological states in light of the probability table read by accessing the storage unit for changes in the driver's index obtained by analyzing the back body surface pulse wave. Is preferred. In the probability table, it is preferable to determine whether or not the index variation corresponds to an appearance condition of the highest probability value regarding a predetermined biological state, and to determine that the biological state corresponds to the condition.

  The present invention is based on the index obtained by analysis of the back body surface pulse wave, only the timing at which the biological state determined as the warning level appearing with the decrease in the arousal level including fatigue, decreased attention, or drowsiness is detected. Instead, it is configured to sequentially output a command for driving the stimulus applying means even at a timing at which a related sign related to the biological state that appears before the appearance of the biological state determined as the warning level is detected. That is, according to the present invention, when each biological state appearing with a decrease in the arousal level is a warning level, for example, the timing when the impending sleep phenomenon is detected while the arousal level is decreasing and drowsiness is increasing. In addition to giving a stimulus at a timing at which a warning has been issued conventionally, the timing at which a related sign related to a decrease in arousal level is detected before reaching the impending sleep phenomenon, for example, the driver still has an imminent sleep phenomenon In the presumed state before reaching the state, stimulation is given in the case where the index obtained by the analysis indicates the rising stage of the fatigue level or the continuous period of the same biological state. Therefore, in order to give a stimulus as a warning when the driver's arousal level is equal to or higher than a predetermined level, the driver can be guided to a higher level of arousal level, and the wakefulness can be maintained. In addition, the timing of detecting the related signs of the present invention, i.e., the timing of detecting the fatigue period, the continuation period of the same biological state, etc. does not appear in a fixed pattern, but appears at random timing, Stimulus applied accordingly is also output at random timing, and it is easy to create a state in which the sympathetic nerve function is relatively dominant and easy to concentrate, and the arousal state can be sustained for a longer time.

Moreover, it is preferable that the present invention is configured to determine whether or not the driver has reached an elevated state. The uplifted state here means that the state of being lifted by the action of β endorphins, etc., while fatigue accumulates due to operation for a predetermined time or more, and after such a state, There is a high possibility that sudden fatigue will be caused and a sleep onset symptom phenomenon, and an imminent sleep phenomenon will appear. Therefore, it is determined whether or not it is in an uplifting state, and the timing at which it is determined that the state has been completed is determined. and, in addition it is preferred that the impending sleep behavior is an extension of the Satoshihiku state also configured to alert, whereby the fatigue driver extent before (his driver feels a so-called overwork which reaches a predetermined or higher degree of fatigue Notice that encourages you to take a break.

  Moreover, it is preferable that the present invention includes a subjective fatigue identification unit that identifies timing when the driver feels fatigue subjectively. The stimulus output timing by the stimulus applying means described above is determined in consideration of the relationship between the activity levels of the sympathetic nerve function and the parasympathetic nerve function, and the health condition in which the driver does not feel fatigue at the start of driving. When is excellent, it is normal that it hardly coincides with the timing at which a subjective fatigue feeling that is consciously aware of the decrease in arousal level. If they match, the determination of the period of increase in fatigue and the continuation period of the same biological condition is actually a determination under a situation where the degree of arousal has decreased, and as a result, the probability of occurrence of sleepiness is high. It is in a state and coincides with the detection timing of the sleepiness phenomenon. Therefore, when the degree of coincidence of these timings becomes high, it can be determined that the driver is considerably fatigued (a state close to fatigue that the driver himself feels overworked). For this reason, it is preferable that a command for prompting the driver to take a break when the time difference between the two is within a predetermined range is output to the stimulus imparting unit, whereby the driver can be surely prompted to take a break. .

FIG. 1 is a perspective view showing an example of a biological signal measuring apparatus for measuring a back body surface pulse wave used in an embodiment of the present invention. 2 is an exploded perspective view of the biological signal measuring apparatus shown in FIG. 3 is a cross-sectional view of a main part of the biological signal measuring apparatus shown in FIG. FIG. 4 is a diagram schematically showing the configuration of the biological state estimation apparatus according to one embodiment of the present invention. FIGS. 5A to 5F are diagrams showing examples of images set in the image display device. FIG. 6 is a diagram showing an example of a generalized probability table created using Bayesian estimation. FIG. 7 is a diagram showing data of a male subject A in his twenties among subjects. FIG. 8 is a diagram showing data of a male subject B in his 20s among the subjects. FIG. 9 is a diagram showing a test result comparing the amplitudes of the zero-cross slope time series waveforms at the timing of the awake state and the onset timing of the imminent sleep phenomenon. FIG. 10 is a diagram showing the examination result as to whether or not there is a relationship between the biological state of the subject and the amplitude of the time-series waveform of the zero cross frequency slope. FIGS. 11A to 11D are diagrams showing the results of the running experiment of Experimental Example 1. FIG. 12A to 12D are diagrams showing the results of a running experiment of Experimental Example 2. FIG. 13A to 13D are diagrams showing the results of a running experiment of Experimental Example 3. FIG. 14A to 14D are diagrams showing a part of the results of the running experiment of Experimental Example 4. FIG. 15A to 15D are diagrams showing a part of the result of the running experiment of Experimental Example 4. FIG. 16A to 16D are diagrams showing a part of the result of the running experiment of Experimental Example 4. FIG. 17A to 17D are diagrams showing a part of the results of the running experiment of Experimental Example 5. FIG. 18A to 18D are diagrams showing a part of the result of the running experiment of Experimental Example 5. FIG. 19A to 19D are diagrams showing a part of the results of the running experiment of Experimental Example 5. FIG. 20 (a) to 20 (e) are diagrams showing a part of the result of the running experiment of Experimental Example 6. FIG. FIGS. 21A to 21E are diagrams showing a part of the result of the running experiment of Experimental Example 6. FIG. 22A to 22E are diagrams showing a part of the results of the running experiment of Experimental Example 6. FIG. 23A to 23G are diagrams showing a part of the result of the running experiment of Experimental Example 6. FIG. 24 (a) to 24 (g) are diagrams showing a part of the result of the running experiment of Experimental Example 6. FIG. 25 (a) to 25 (g) are diagrams showing a part of the result of the running experiment of Experimental Example 6. FIG. FIGS. 26A to 26D are diagrams showing the results of a running experiment of Experimental Example 7. FIG. FIGS. 27A to 27D are diagrams showing the results of a running experiment of Experimental Example 8. FIG. FIGS. 28A to 28D are diagrams showing the results of the running experiment of Experimental Example 9. FIG. 29A to 29D are diagrams showing the results of a running experiment of Experimental Example 10. FIG. FIGS. 30A to 30D are diagrams showing the results of a running experiment of Experimental Example 11. FIG. FIGS. 31A to 31D are diagrams showing the results of a running experiment of Experimental Example 12. FIG. FIG. 32 is a diagram showing an analysis result of data on February 14 of subject B on the data of 400 professional drivers. FIG. 33 is a diagram showing an analysis result of data on February 15 of subject B on the data of 400 professional drivers. FIG. 34 is a diagram showing an analysis result of data on February 16 of subject B on the data of 400 professional drivers. FIG. 35 is a diagram showing an analysis result of data of February 17 of subject B on the data of 400 professional drivers. FIG. 36 is a diagram showing the analysis result of the data of the subject H on the data of 400 professional drivers. FIG. 37 shows the running results of subject A in Experimental Example 14, where (a) is the APW zero-cross slope time-series waveform, (b) is the APW zero-cross distribution rate, and (c) is the θ wave power spectrum by a simple electroencephalograph. (D) is the figure which showed the alpha wave power spectrum by a simple electroencephalograph, (e) is the figure which showed the beta wave power spectrum by a simple electroencephalograph. FIG. 38 is a diagram comparing the APW distribution rate of 0.0017 Hz and the power spectrum of the θ wave by a simple electroencephalograph. FIG. 39A shows the FFT analysis result of the time series waveform of the zero cross slope when the drowsiness sign occurs, and FIG. 39B shows the FFT analysis result of the time series waveform of the zero cross slope in the uplift state. is there. FIG. 40 is a diagram showing the average value of the quantification score in the time zone in which the drowsiness sign phenomenon appears and the quantification score in the time zone in which the uplift state appears. FIG. 41A is a diagram showing the incidence of drowsiness using Bayesian estimation with respect to the amplitude of the zero-cross slope time series waveform, and FIG. 41B is the occurrence of sleepiness using Bayesian estimation with respect to frequency. 41 (c) is a diagram showing the probability of being in an uplifted state, and FIG. 41 (d) is a diagram showing the probability of being a sleepiness predictor phenomenon. 42A to 42E are diagrams showing various data of the subject A in Experimental Example 15. FIG. FIG. 43 is a diagram illustrating a representative example of an estimated state obtained by Bayesian estimation and an actual place of occurrence of sleepiness. FIG. 44 is a diagram illustrating an estimated state obtained by Bayesian estimation, an actual drowsiness occurrence location, and a stimulus presentation location in Experimental Example 16.

  Hereinafter, the present invention will be described in more detail based on the embodiments of the present invention shown in the drawings. The biological signal information collected in the present invention is the back body surface pulse wave (APW), and is measured by the biological signal measuring device 1 (hereinafter referred to as “back body surface pulse wave measuring device”) 1 shown in FIGS. The The back body surface wave (APW) is a pressure vibration generated from the motion of the heart and aorta detected from the upper back of a person, and information on ventricular systole and diastole and the blood vessel wall serving as an auxiliary pump for circulation Elasticity information and blood pressure elasticity information. The signal waveform associated with heart rate variability includes sympathetic and parasympathetic nervous system activity information (parasympathetic activity information including the compensation of sympathetic nerves), and the signal waveform associated with aortic oscillation is sympathetic. Contains information on neural activity.

  Here, a schematic configuration of the back body surface pulse wave measuring device 1 will be described with reference to FIGS. 2 and 3. As shown in these drawings, the back body surface pulse wave measuring apparatus 1 includes a core pad 11, a spacer pad 12, a sensor 13, a front film 14, and a rear film 15.

  The core pad 11 is formed in, for example, a plate shape, and two vertically long through holes 11a and 11a are formed at symmetrical positions across a portion corresponding to the spinal column. The core pad 11 is preferably composed of a polypropylene bead foam formed in a plate shape. When the core pad 11 is composed of a bead foam, the foaming ratio is preferably in the range of 25 to 50 times, and the thickness is preferably less than the average diameter of the beads. For example, when the average diameter of 30 times expanded beads is about 4 to 6 mm, the core pad 11 is sliced to have a thickness of about 3 to 5 mm.

  The spacer pad 12 is loaded into the through holes 11 a and 11 a of the core pad 11. The spacer pad 12 is preferably formed from a three-dimensional solid knitted fabric. The three-dimensional solid knitted fabric includes, for example, a pair of ground knitted fabrics spaced apart from each other and the pair of grounds as disclosed in JP 2002-331603 A, JP 2003-182427 A, and the like. The knitted fabric has a three-dimensional three-dimensional structure having a large number of connecting yarns that reciprocate between the knitted fabrics to couple them together. When the 3D solid knitted fabric is pressed by the person's back, the connecting yarn of the 3D solid knitted fabric is compressed, tension is generated in the connecting yarn, and the vibration of the body surface through the human muscle accompanying the biological signal is propagated. The In addition, it is preferable to use a spacer pad 12 made of a three-dimensional solid knitted fabric that is thicker than the core pad 11. Accordingly, when the peripheral portions of the front film 14 and the rear film 15 are attached to the peripheral portions of the through holes 11a and 11a, the spacer pad 12 made of a three-dimensional solid knitted fabric is pressed in the thickness direction. Tension due to the reaction force of the film 15 is generated, and solid vibration (membrane vibration) is likely to occur in the front film 14 and the rear film 15. On the other hand, the spacer pad 12 made of a three-dimensional solid knitted fabric is also pre-compressed, and the connecting yarn that holds the shape of the three-dimensional solid knitted fabric in the thickness direction is also subjected to tension due to a reaction force, and string vibration is likely to occur.

  The sensor 13 is fixedly disposed on one of the spacer pads 12 before the front film 14 and the rear film 15 are laminated. As described above, the three-dimensional solid knitted fabric constituting the spacer pad 12 is composed of a pair of ground knitted fabrics and connecting yarns. In order to be transmitted to the rear film 15, the sensor 13 is preferably fixed to the surface of the spacer pad 12 (the surface of the ground knitted fabric). As the sensor 13, it is preferable to use a microphone sensor, in particular, a condenser microphone sensor.

  Next, the configuration of the driving support device 60 of the present embodiment will be described with reference to FIG. The driving support device 60 includes a biological state determination unit 61 and a stimulus application unit 62. The biological state determination means 61 includes a zero cross detection means 611, a peak detection means 612, a zero cross slope time series waveform calculation means 613, a peak slope time series waveform calculation means 614, a distribution rate calculation means 615, a determination output means 616, and the like. Thus, the back body surface pulse wave (APW) obtained from the sensor 13 of the back body surface pulse wave measuring device 1 is analyzed to obtain various time-series waveforms that serve as indicators for determination. The driving support device 60 is configured by a computer, and causes the storage unit of the computer to execute a zero-cross detection procedure that functions as the zero-cross detection unit 611 and to execute a peak detection procedure that functions as the peak detection unit 612, and a zero-cross slope time series A distribution rate functioning as a distribution rate calculating means 615 is executed by executing a zero cross slope time series waveform calculating procedure that functions as the waveform calculating means 613 and executing a peak slope time series waveform calculating procedure that functions as the peak tilt time series waveform calculating means 614. A computer program for executing a calculation procedure and executing a determination output procedure functioning as the determination output means 616 is set. The computer program can be provided by being stored in a recording medium such as a flexible disk, a hard disk, a CD-ROM, an MO (magneto-optical disk), a DVD-ROM, or a memory card, or transmitted through a communication line. Is possible.

  In addition to the storage unit described above, the driving support device 60 includes a stimulus applying unit 62. The stimulus imparting means 62 of this embodiment includes a sound output device (mechanism portion for outputting sound including a speaker) 621 that imparts an auditory stimulus, and an image display device (including a display) that imparts a visual stimulus. And a mechanism portion 622 for displaying an image. The stimulus to be displayed on the image display device 622 is arbitrary such as a warning message or a picture. However, if a plurality of kinds of pictures (for example, animation images) are prepared and displayed alternately, the cranial nerves are activated through vision. It is preferable because it is easy to do. In addition, this stimulus is a warning issued at the time of monotonous work such as driving, and aims to escape from the monotonous work and induces awakening that adds a complex system to the work. Therefore, the stimulus to be applied is not particularly limited as long as it can achieve this purpose, and if it is too strong, it disturbs driving. Therefore, the image set in the image display device 622 corresponds to each biological state, for example, as shown in FIGS. 5A to 5F, the mood display, the fatigue display, and the fatigue prediction. It is set in a range of 3 to 10 types, such as a display of a warning, a display of alerting, a display of applying light stimulus / voice stimulus, a display of applying a stimulus at a warning level such as an imminent sleep phenomenon preferable. Moreover, it is preferable to set it so that it may have two to four different images for each type, and different images are displayed depending on the mood level, fatigue level, and the like. Further, in the case where it is necessary to give a stimulus, in the display when applying the light stimulus / voice stimulus in FIG. 5E and in the display giving the stimulus at the warning level in FIG. It is preferable to apply stimulation by setting so that two images are alternately displayed. Of course, the stimulus imparting means 62 is not limited to this, and may stimulate only one of hearing and vision, or in addition to hearing or vision, or Instead of them, an olfactory stimulus may be given.

  The zero cross detection means 611 is a time series waveform (hereinafter referred to as “original waveform”) of the back body surface pulse wave (APW) obtained from the sensor 13 of the back body surface pulse wave measuring device 1. In a waveform after filtering a component not used for analysis of body motion or the like, a time series waveform of a frequency is obtained by using a point where the signal is switched from positive to negative (hereinafter referred to as “zero cross point”). If the zero cross point is obtained, it is divided every 5 seconds, for example, and the reciprocal of the time interval between the zero cross points of the time series waveform included in the 5 second is obtained as the individual frequency f, and the individual frequency f of the 5 second is obtained. The average value is adopted as the value of the frequency F for 5 seconds. Then, by plotting the frequency F obtained every 5 seconds in time series, a time series waveform of frequency fluctuation is obtained.

  The peak detecting means 612 first obtains the original waveform of the back body surface pulse wave (APW) obtained from the sensor 13 of the back body surface pulse wave measuring device 1 by smoothing differentiation and using the maximum value (peak). For example, the maximum value is obtained by a smoothing differential method using Savitzky and Golay. Next, for example, the local maximum value is divided every 5 seconds, and the reciprocal of the time interval between the local maximum values of the time series waveform included in the 5 seconds (the peak on the peak side of the waveform) is obtained as the individual frequency f. The average value of f is adopted as the value of frequency F for 5 seconds. Then, by plotting the frequency F obtained every 5 seconds in time series, a time series waveform of frequency fluctuation is obtained.

  The zero-cross slope time-series waveform calculating means 613 sets a time window having a predetermined time width with a predetermined overlap time from the time-series waveform of the frequency fluctuation obtained by the zero-cross detecting means 611, and sets a minimum two times for each time window. A frequency gradient is obtained by multiplication, and a time-series waveform is output at the time of the gradient. This calculation (movement calculation) is sequentially repeated to output a time-series change in the APW frequency gradient as a frequency gradient time-series waveform.

  The peak inclination time series waveform calculating means 614 sets a time window having a predetermined time width with a predetermined overlap time from the time series waveform of the frequency fluctuation obtained by the peak detecting means 612, and minimum two times for each time window. A frequency gradient is obtained by multiplication, and a time-series waveform is output from the gradient. This calculation (movement calculation) is sequentially repeated to output a time-series change in the APW frequency gradient as a frequency gradient time-series waveform.

  The dorsal body surface wave (APW) is a biological signal mainly including the state of control of the heart, which is the central system, that is, the state of sympathetic innervation of the artery, and the appearance information of the sympathetic nervous system and the parasympathetic nervous system. A waveform obtained by performing absolute value processing on the tilt time-series waveform of the biological signal by the zero-cross tilt time-series waveform computing unit 613 is more related to the state of control of the heart and reflects the appearance of the sympathetic nerve. . The thing by the peak inclination time series waveform calculating means 614 is related by the heartbeat fluctuation, and captures the dynamics of the parasympathetic nervous system in consideration of the compensation effect by the sympathetic nerve. It should be noted that the absolute value processing of the slope time series waveform by the peak slope time series waveform calculation means 614 is the parasympathetic nerve dynamics by the wavelet analysis of the fingertip volume pulse wave (this parasympathetic nerve dynamics have a sympathetic compensation effect added). Are relatively close to each other. Therefore, it is considered that the one using the zero-cross detection means 611 and the zero-cross slope time series waveform calculation means 613 can be used as an index representing the physical condition resulting from the adaptation to the stress that is dealt with by the control of the autonomic nervous system. . The one using the zero-cross detection means 611 and the zero-cross slope time series waveform calculation means 613 has high relevance to the state of heart control, and therefore includes information on heartbeat fluctuation notches. The frequency component in the vicinity of 0.5 Hz, which is the 0.5th-order component of the heartbeat component, is also obtained as information. Therefore, it is preferable to use data obtained by using the zero-cross detection unit 611 and the zero-cross slope time series waveform calculation unit 613 as a basis when determining the biological state using the APW.

  The distribution rate calculation means 615 is an inclination time series waveform (zero cross inclination time series waveform) obtained from the zero cross inclination time series waveform calculation means 613 or an inclination time series waveform (at peak inclination) obtained from the peak inclination time series waveform calculation means 614. Frequency analysis of each of the series waveforms) into a function adjustment signal in the range of 0.001 to 0.0027 Hz, a fatigue acceptance signal in the range of 0.002 to 0.0052 Hz, and an activity adjustment signal in the range of 0.004 to 0.007 Hz. Each corresponding frequency component is extracted, and the temporal variation of the distribution rate of each of these frequency components is obtained. That is, the sum of power spectrum values of frequency components corresponding to the three signals of the function adjustment signal, fatigue acceptance signal, and activity adjustment signal belonging to the ULF band (very low frequency band) to the VLF band (very low frequency band) is 100. The ratio of each frequency component is determined in time series.

  This is based on the following knowledge by the present applicant. That is, as described in Japanese Patent Application Laid-Open No. 2011-167362 and Japanese Patent Application Laid-Open No. 2012-95779 by the present applicant, human constancy is maintained with fluctuations, and the frequency band is in the ULF band and the VLF band. It is said that. On the other hand, in atrial fibrillation, which is one of heart diseases, the place where the characteristics of fluctuations of the heart and circulatory system are switched is said to be 0.0033 Hz. By capturing fluctuations around 0.0033 Hz, Information on sex maintenance can be obtained. In addition, the frequency band near 0.0033 Hz or less and the band near 0.0053 Hz are mainly related to body temperature regulation, and the frequency band of 0.01 to 0.04 Hz is said to be related to autonomic nerve control. Yes. Then, a frequency-gradient time series waveform for calculating these low-frequency fluctuations inherent in the biological signal is actually obtained and subjected to frequency analysis. As a result, the frequencies near 0.0017 Hz and 0.0033 Hz are lower than 0.0033 Hz. It was confirmed that there were fluctuations in the frequency band centered on 0.0035 Hz and, in addition to these two, fluctuations in the frequency band centered on 0.0053 Hz.

  The 0.0035 Hz signal (fatigue acceptance signal) is a fluctuation for maintaining homeostasis in response to externally input stress, and is a signal indicating the progress of fatigue in a normal activity state. The .0053 Hz signal (activity adjustment signal) is a signal in which the degree of influence due to the control of endocrine hormones during activity appears, and the 0.0017 Hz signal (function adjustment signal) lower than 0.0033 Hz is These signals are used to control body modulation and functional degradation, and these three frequency band signals interact with each other and act as a body temperature regulation function. Therefore, it is possible to determine the state of a person by using the time series change and distribution rate of the power spectrum of these signals.

  In the present embodiment, 0.0017 Hz is a function adjustment signal, 0.0035 Hz is a fatigue acceptance signal, and 0.0053 Hz is an activity adjustment signal. The function adjustment signal may be adjusted in the range of 0.001 to 0.0027 Hz, the fatigue acceptance signal may be adjusted in the range of 0.002 to 0.0052 Hz, and the frequency of the activity adjustment signal may be adjusted in the range of 0.004 to 0.007 Hz. it can.

  The determination output means 616 uses each index obtained by the analysis of the back body surface pulse wave to determine a biological state that is determined as a warning level that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness A command for driving the stimulus imparting means 62 is output at both the timing when the detection is detected and the timing when the related sign related to the decrease in the arousal level that appears before the appearance of the biological state of the warning level is detected.

The timing at which each biological state appearing with a decrease in the arousal level is determined to be a warning level refers to, for example, a timing at which the arousal level is significantly decreased and an imminent sleep phenomenon or a sleep onset phenomenon is detected. Whether or not it is an imminent sleep phenomenon, as disclosed in Japanese Patent Application Laid-Open No. 2012-239480 by the present applicant, shows a change in which the distribution rate of 0.0017 Hz rapidly decreases and the distribution rate of 0.0053 Hz rapidly increases. At the time, or in the power time gradient time series waveform proposed in Japanese Patent Application No. 2012-274104 (in this embodiment, the zero-cross slope time series waveform or the peak slope time series) is greater than a preset amplitude. After that, it can be specified as a point in time when the amplitude becomes equal to or less than a predetermined value. In more detail, in the latter, in the frequency-gradient time-series waveform using the zero-crossing detection method, after the waveform indicating the sleep onset phenomenon appears, the waveform shows a tendency to converge, and then shows a fluctuation fluctuation with a longer period. It is a point. For example, the frequency gradient time-series waveform using the zero cross detection method is smoothed and differentiated, and the point at which the positive gradient is switched to the negative gradient is detected as the peak, and the point at which the negative gradient is switched to the positive gradient is detected as the bottom. The amplitude value between the two is calculated. Next, it is determined whether or not there are a plurality of such amplitude values. If there are a plurality of such amplitude values, the magnitude is compared with the amplitude value of the previous data, and the ratio of the magnitudes is less than a predetermined number. If the value of the smoothed differential waveform falls within a predetermined range continuously for a predetermined calculation point or more and shows a tendency to converge, and then is determined to be an imminent sleep phenomenon when there is a larger fluctuation fluctuation with a longer period To do.
As described above, the zero-cross slope time series waveform of the back body surface pulse wave (APW) reflects the appearance state of the sympathetic nerve, and the peak slope time series waveform captures the dynamics of the parasympathetic nervous system. The imminent sleep phenomenon is a phenomenon accompanied by sleepiness that appears after the onset of sleep phenomenon, and is in good agreement with subjective sleepiness. Therefore, in capturing the impending sleep phenomenon, it is more preferable to determine by taking into account both the zero-cross slope time series waveform and the peak slope time series waveform. Specifically, when the fluctuation in the zero-cross slope time series waveform is small and the fluctuation in the peak slope time series waveform shows an expansion tendency in the time zone when the fluctuation is small, it is determined that the sleep phenomenon is imminent Is preferred. The fluctuation of the zero-cross slope time-series waveform is reduced because the sympathetic nerve function is reduced, and the fluctuation in the peak slope time-series waveform shows an increasing tendency because the parasympathetic nerve function is activated, which causes sleepiness. It shows an increasing phenomenon. Note that whether or not the fluctuation of the time-series waveform of the zero cross slope has become constant is determined, for example, by whether the amplitude and period of the determination time zone are within a predetermined range in relation to the amplitude and cycle of the previous predetermined time zone. (The determination based on whether or not the above-described convergence tendency is present corresponds to an example of the case where the fluctuation becomes constant). Further, whether or not the peak slope time-series waveform shows an expanding tendency depends on whether or not the amplitude of the determination time zone is a predetermined multiple or more in relation to the amplitude of the predetermined time zone before that, for example. Can be judged.

  Since the sleep onset symptom phenomenon is a phenomenon accompanied by an increase in sympathetic nerve function, it is basically determined by the appearance of a plurality of continuous waves having a large amplitude in the frequency gradient time series waveform. For example, as proposed in the above-mentioned Japanese Patent Application No. 2012-274104, the frequency gradient time series waveform using the zero cross detection method is smoothed and differentiated, and the point at which the positive gradient is switched to the negative gradient is detected as a peak, The point at which the negative slope changes to the positive slope is detected as the bottom, and the amplitude value between the two is calculated. Next, it is determined whether or not there are a plurality of such amplitude values. If there are a plurality of such amplitude values, the magnitude is compared with the amplitude value of the previous data, and the ratio of the magnitudes is a predetermined multiple or more. When a predetermined time or more has elapsed since the time when the amplitude value of the previous data was obtained, it is determined as a sleep onset sign phenomenon.

  The timing at which it is determined as an imminent sleep phenomenon or a predictive sleep phenomenon described above is a phenomenon in which the state of life accompanied by a decrease in arousal level, such as increased fatigue, decreased attention, or increased sleepiness It is necessary to capture these phenomena and give the driver a stimulus as a warning. This is because it is very likely that the patient will fall asleep if left unattended, but if the stimulus is applied only when the warning level is determined, the driver is not necessarily guided to the awake state. It may not be possible. Therefore, the present invention is characterized in that, even at the timing before the warning level is determined, a stimulus is applied to induce the driver to wake up and promote the continuation of the wakeful state. That is, a command for driving the stimulus applying means 62 is output even at a timing when a related sign related to a decrease in arousal level that appears before the warning level is determined. A related sign related to a decrease in wakefulness is, for example, that the index varies in the direction of decrease in the wakefulness in the end-of-wake period between the wakefulness state and the wakefulness state, or in the wakefulness state before that. Specifically, in the presumed state that the driver is not yet in the state of imminent sleep or sleep onset, the fatigue level rises, the same biological state continues, or the sympathetic nerve temporarily rises, etc. When it is determined that, the command for driving the stimulus applying means 62 is sequentially output. Thereby, a stimulus is given to the driver at a necessary and random timing, and the awakening state can be maintained for a longer time.

  In the “rising period of fatigue”, negative feedback is applied to the living body, but when the fatigue level is high, it is difficult to deal with in the awakening center, and often shifts to the sleeping center. Therefore, in the present embodiment, to the drowsiness or tiredness that occurs in such monotonous work, so-called complexity is given to the monotonous work by stimulating the five senses, thereby making it constant in the awakening center. It promotes the sex maintenance function.

  The “continuation period of the same biological state” means, for example, that the state of enhanced sympathetic nerve function continues for a predetermined time or longer, or conversely, the state of increased parasympathetic nerve function continues for a predetermined time or longer, and biological fluctuation is small. This is the period during which the status has continued. If the same biological state continues for a predetermined time or more, for example, if it continues for more than 2 minutes, the decrease in arousal level is not so great, even if you are not aware of fatigue at that time. In the future, the degree of arousal may be greatly reduced. This point is also the same for the “rising period of fatigue”, and if no stimulus is applied, fatigue may appear early. Moreover, the temporary enhancement of the sympathetic nerve often appears immediately before the appearance of the sleep onset sign phenomenon.

  In the present embodiment, including the “rising period of fatigue” and the “continuation period of the same biological state”, the biological state of the driver is an inclination time series waveform (zero cross inclination time series) obtained from the zero cross inclination time series waveform calculation means 613. Waveform) or at least one of the slope time series waveform (peak slope time series waveform) obtained from the peak slope time series waveform calculation means 614. The case of using the zero-cross slope time series waveform or the peak slope time series waveform here is not only the case where these slope time series waveforms are directly determined, but also the frequency analysis of each slope time series waveform. In the case where the target is determined as a determination target, the case where the frequency analysis is performed on the distribution ratio time series waveform of each frequency component described above by the distribution ratio calculation means 615 is also included.

  In the present embodiment, whether the fatigue level is rising or not is determined by the frequency component (0.0053 Hz in this embodiment) corresponding to the activity adjustment signal in the distribution ratio of each frequency component obtained using the zero-cross slope time series waveform. ) Is more than a predetermined frequency component (0.0017 Hz in this embodiment) corresponding to the function adjustment signal, preferably the frequency component (this book) corresponding to the activity adjustment signal. In the embodiment, when the distribution rate of 0.0053 Hz exceeds the distribution rate of the frequency component (0.0017 Hz in the present embodiment) corresponding to the function adjustment signal, it is determined that the fatigue level is rising. This is because, as described above, the tilt time-series waveform by the zero-cross tilt time-series waveform calculating means 613 is related to the state of heart control and reflects the appearance state of the sympathetic nerve. In addition, the function adjustment signal is a signal that rises when the state changes, and the activity adjustment signal indicates the degree of activity of the body temperature adjustment, so that the distribution rate of the activity adjustment signal exceeds the distribution rate of the function adjustment signal. This is because endocrine secretion becomes active, its activity increases, and 1 / f biological fluctuations are disrupted.

  In addition, the distribution ratio of each frequency component obtained using the zero-cross slope time series waveform and the distribution ratio of each frequency component obtained using the peak slope time series waveform were compared between the corresponding frequency component distribution ratios. In this case, it is determined that the fatigue level is rising even when both tend to deviate. The zero-cross slope time series waveform is highly relevant to sympathetic nerve function, and the peak slope time series waveform is a highly relevant index to parasympathetic nerve function. This is because the activity of the parasympathetic nerve function deviates and the fatigue level increases. It can be determined by statistically processing the data of a large number of subjects to determine whether the degree of divergence is determined as an increased state of fatigue. Alternatively, it may be set for each individual.

  Whether or not it is “continuation period of the same biological state” is determined by whether or not the state of increased sympathetic nerve function or the state of increased parasympathetic nerve function continues for a predetermined time or more. Among these, the enhanced state of the sympathetic nerve function is, for example, the frequency component (0.0017 Hz in this embodiment) corresponding to the function adjustment signal in the distribution ratio of each frequency component obtained using the zero-cross slope time series waveform. The determination is made based on whether or not the state in which the distribution rate exceeds the distribution rate of the frequency component (0.0053 Hz in this embodiment) corresponding to the activity adjustment signal continues for a predetermined time or more. Whether the parasympathetic nerve function is in an enhanced state is determined by, for example, the frequency component corresponding to the activity adjustment signal (0.0053 Hz in this embodiment) being a frequency component corresponding to the function adjustment signal (0 in this embodiment). .0017 Hz) is determined based on whether or not the state exceeding the distribution rate of 0017 Hz continues for a predetermined time or more. If such a state continues for a predetermined time or more, the biological fluctuation gradually becomes smaller and 1 / f is lost. The setting of the duration time can be determined by statistically processing data of a large number of subjects as described above, or may be set for each individual. In general, when the enhanced state continues for more than 2 minutes, the divergence between the sympathetic nerve function and the parasympathetic nerve function tends to increase thereafter, so a warning is given by setting the duration in the range of 2 minutes to 5 minutes. However, it is preferable that the 1 / f fluctuation can be maintained by activation.

  As described above, the determination output means 616 includes the “time of increasing fatigue” and the “continuation period of the same biological state” as the inclination time series waveform (zero cross inclination time series waveform) obtained from the zero cross inclination time series waveform calculation means 613. Alternatively, determination is made using at least one of the slope time series waveforms (peak slope time series waveforms) obtained from the peak slope time series waveform calculation means 614, and a signal for driving the stimulus applying means 62 is output. At this time, in the determination of the rising period of the fatigue level, the frequency component (0.0053 Hz in the present embodiment) corresponding to the activity adjustment signal is the frequency component (0.003 Hz in the present embodiment) corresponding to the distribution ratio. In the determination of the continuation period of the same biological state, the distribution rate of either the function adjustment signal or the activity adjustment signal is higher than the other. And the degree of those is divided into a plurality of stages (for example, three stages), thereby changing the type of stimulus (for example, volume, sound content, type of image, etc.) output by the stimulus applying means 62 It is preferable to set to.

The determination output means 616 is further set with means for detecting an elevated state . As described above, the elevated state means that the state is elevated due to the action of substances in the brain such as β-endorphin while fatigue accumulates due to operation for a predetermined time or longer. This is presumed to be similar to the so-called “runner's high” that a person may experience in running for a long time, and despite the accumulation of fatigue, the pain caused by the substance in the brain is reduced, It feels like driving continuously and smoothly for a long time. That is, as described above, when the state of enhanced sympathetic nerve function continues for a predetermined time or more, the divergence between the sympathetic nerve function and the parasympathetic nerve function tends to increase. Therefore, a warning is given at such a point of time, but when a warning is given and 1 / f fluctuation is maintained, the degree of activity may become very large. This is particularly seen when the amplitude of at least one of the zero-cross slope time series waveform and the peak slope time series waveform changes to a predetermined value or more. Therefore, when the amplitude of any tilt time-series waveform, preferably the zero-cross tilt time-series waveform having a high relevance to the sympathetic nerve function, is greater than or equal to a predetermined value, it is determined as an uplift state. When the uplift state is detected, the determination output means 616 of the present embodiment continues the enhanced state of the sympathetic nerve function, so that the stimulus is applied and the wakefulness state is maintained as long as possible as described above. .

  On the other hand, while such an uplifting state continues, the driver is actually not aware of it, and in fact, the driver is experiencing considerable physical and mental fatigue. Therefore, when this uplift state ends, the fatigue level increases at a stretch. Therefore, it is preferable that the stimulus imparting means 62 is set to operate so as to encourage the driver to take a break when it is determined that the elevated state has ended after the elevated state has been detected. The end of the uplift state is determined when the amplitude of at least one of the zero-cross slope time series waveform and the peak slope time series waveform, which has been changing with a predetermined amplitude or more in the uplift state, is less than or equal to a predetermined value.

  As a mode in which the stimulus applying means 62 is operated so as to encourage the driver to take a break, for example, a break is prompted with a louder warning than a normal warning through the sound output device 621 constituting the stimulus applying means 62. Or a command to repeatedly play the sound of a break in the nearest service area, a message to that effect displayed on the image display device 622, or a plurality of images displayed during normal stimulation It can be set to notify that it is time to take a break by shortening the time interval for alternately displaying (animation images, etc.). It is also possible to display the nearest service area on the image of the navigation device and perform route guidance to the service area. Further, the operation manager may be notified through a communication network so that the operation manager can directly issue a message prompting the driver to take a break through the communication network.

  Moreover, it is preferable that the determination output means 616 of this embodiment has a means for detecting a sleep onset sign phenomenon or an imminent sleep phenomenon. After the uplift state is detected by the judgment output means 616, when this sleep onset sign phenomenon or imminent sleep phenomenon is further detected, the degree of fatigue has reached a predetermined level, which is a state immediately before overwork. Therefore, the determination output means 616 determines whether an elevated state and a sleep onset sign phenomenon or an imminent sleep phenomenon are detected within a predetermined time, and activates the stimulus applying means 62 to encourage the driver to take a break. It is preferable that they are set as follows. In this case, since it is necessary to urge a break more reliably than the time when the above-described uplift state has ended, the operation of the stimulus applying means 62 is more powerful than the above, for example, the volume is further increased, and the image switching frequency is increased. It is preferable to raise it further.

  In addition, the present embodiment includes subjective fatigue identification means 63 that identifies the timing when the driver feels subjective fatigue. The subjective fatigue identification means 63 can be constituted by, for example, an input means that is operated when the driver subjectively feels fatigue or drowsiness above a predetermined level. As the input means, for example, an input button provided on the handle of the driver's seat, the display screen itself of the image display device 622 constituting the stimulus applying means 62 (screen touch type input means), or the driver is “tired” ”,“ I want to sleep ”and voice input means for inputting the voice as information as it is.

  The above-mentioned stimulus output timing by the stimulus applying means 62 is determined in consideration of the relationship between the activities of the sympathetic nerve function and the parasympathetic nerve function, and this timing does not necessarily cause the driver to feel subjective fatigue. It does not coincide with the time. However, when the coincidence between the timing when the driver feels fatigue subjectively and the output timing of the stimulus applying means 62 becomes high, the driver is considerably fatigued (the driver himself is overworked). It can be judged that the condition is close to the feeling of fatigue). Therefore, the determination output unit 616 preferably has a function of determining whether or not the output timing of the stimulus by the stimulus applying unit 62 and the input timing of the subjective fatigue specifying unit 63 are within a predetermined range. Even when it is determined that the time difference between the two is within the predetermined range, it is preferable to operate the stimulus applying means 62 so as to encourage the driver to take a break, as described above.

  In other words, the output timing of the stimulus by the stimulus applying means 62 is as described above in the awake state or in the borderline period between the awake state and the low state of the awake state. This is the moment when the period of temporary enhancement is reached. On the other hand, the timing at which the driver feels subjective fatigue by operating the subjective fatigue identification means 63 is a situation in which the degree of arousal is often reduced. Therefore, it is normal that the timing of both does not match. However, the phenomenon of increased fatigue level or continuation of the same biological condition occurs even after the decrease in arousal level, in which case it coincides with subjective fatigue such as low driving state and subjective sleepiness. Detected as manifestation. Therefore, the subject is provided with the subjective fatigue specifying means 63 as in the present embodiment, and the timing determined as the fatigue level rising period and the continuation period of the same biological state (that is, the output timing of the stimulus applying means 62), and the subjective fatigue specifying means By comparing the time difference with the input timing of 63, it is possible to determine whether or not the determination of the rising period of the fatigue level and the continuation period of the same biological state is the awake state. The setting of the time difference between the two is not necessarily limited, but when the time difference is within one minute, it is preferable to determine that the degree of coincidence between the two becomes high and the degree of arousal has decreased. However, if it is only determined in a single shot that the degree of coincidence is high, it means that the degree of coincidence has been returned to the awake state. In such a case, it is preferable to issue a command that prompts the driver to take a break.

  On the other hand, a biological state corresponding to a warning level with strong sleepiness such as an imminent sleep phenomenon usually coincides with a timing when subjectively strong sleepiness is felt. Accordingly, the detection timing of the biological state corresponding to the warning level such as the impending sleep phenomenon and the detection timing of the related signs are usually not coincident. Therefore, in the present invention, when the time difference between the detection timing of the biological state corresponding to the warning level and the detection timing of the related signs is determined to be within a predetermined time regardless of the presence or absence of the subjective fatigue specifying means 63 described above. In the same manner as described above, it is preferable that a stimulus for prompting the driver to take a break is instructed to the stimulus applying means 62.

  As described above, the present invention detects a related sign related to a decrease in arousal level and gives a stimulus in a stage before reaching a biological state corresponding to a warning level with strong sleepiness such as an impending sleep phenomenon. Is. Therefore, it is preferable that the biological condition corresponding to the warning level (the biological condition that often coincides subjectively as described above) is set so as to be detected more reliably. The imminent sleep phenomenon or the like can be detected as described above, but it is preferable to use the following method instead of or in combination with the imminent sleep phenomenon.

  That is, a generalized probability table of index fluctuations obtained by analysis of the back body surface pulse wave and biological states appearing corresponding to the index fluctuations is stored in advance in a storage unit of a computer constituting the driving support device 60. Let me. And the determination output means 616 is set as the structure which determines each biological state in light of the fluctuation | variation of a driver | operator's parameter | index obtained by the analysis of a back body surface pulse wave in light of the probability table read by accessing the memory | storage part.

  For example, as an index obtained by analyzing the back body surface pulse wave, the above-described zero-cross slope time series waveform can be used. A plurality of data is acquired in advance and a generalized probability table is created for the relationship between the amplitude of this zero-cross slope time-series waveform and each biological state such as the state of wakefulness and the state of imminent sleep. The generalized probability table can be created using, for example, Bayesian estimation. FIG. 6 is an example of a generalized probability table created using Bayesian estimation.

  This probability table was created by comparing the amplitudes of the time series waveform of the zero cross slope at the timing when the wakefulness of 20 subjects and the imminent sleep phenomenon occurred. 7 and 8 show data obtained by picking up data of two male subjects A and B in their 20s as representative examples. In addition, in order to cause an imminent sleep phenomenon, the experiment was performed by setting the stimulus applying means 62 of the driving support device 60 not to be driven. The arousal state was set to the timing when the content of the electroencephalogram α wave or β wave was high and estimated to be the most awake from the comments of the subject. At the onset of the impending sleep phenomenon, it is the timing when the content rate of the θ wave of the electroencephalogram switches from an increase to a constant, or the content rate of the θ wave is increased by about 10% from the arousal state. The timing is estimated to be immediately before.

FIG. 9 is a test result comparing the amplitudes of the time series waveforms of the zero-cross slope in the timing of the awake state and the onset timing of the imminent sleep phenomenon. Is also significantly smaller. Therefore, FIG. 6 is a generalized probability table created by dividing this amplitude into four stages and using Bayesian estimation. From this result, in the amplitude D = 5 × 10 ⁻⁴ or less, the expression of the impending sleep phenomenon is the highest probability value. Therefore, when the amplitude of the zero-cross slope time-series waveform to be determined actually obtained is 5 × 10 ⁻⁴ or less, the determination output means 616 indicates the occurrence of the impending sleep phenomenon in light of this probability table. It is set to be judged as time. Therefore, in a time zone in which it is not determined that this imminent sleep phenomenon has occurred, if the above-mentioned related signs are detected, a command for driving the stimulus applying means 62 can be output each time. In the above-described example, the imminent sleep phenomenon is captured in relation to the arousal state. However, the present invention is not limited to this, and can naturally be used for determination of other biological states such as a sleep onset sign. In addition, the probability table shown in FIG. 6 is merely an example. For example, the amplitude range shown in the probability table is divided into finer categories, and the probability is obtained for each of them. It is possible to more accurately determine the timing at which the probability value occurs.

  Note that FIG. 10 is a study of whether or not there is a relationship between the biological state of the subject and the amplitude of the zero-cross frequency slope time series waveform from the 2 × 2 cross table. It can be seen that there is a relationship between

(Experimental example)
A running experiment was conducted with a passenger car in which the back body surface pulse wave measuring device 1 according to the above embodiment was incorporated in the seat back of the driver's seat. The driving support device 60 mounted on the passenger car used in the experiment is in a state in which the determination output unit 616 does not output a signal for driving the stimulus applying unit 62, that is, in a state where the fatigue is increased or the sympathetic function is increased. Refer to the zero-cross slope time series waveform and / or peak slope time series waveform for the normal state that is not the continuous state, and the most comfortable state from the distribution rate values of the function adjustment signal, fatigue acceptance signal, and activity adjustment signal there To the previous state where the stimulus needs to be applied is divided into 6 stages (determination levels 1 to 6 indicated by the numerical values on the vertical axis in the “determination result”). Levels 7 to 1 are determined when a state in which a stimulus needs to be applied in a period and sleepiness information (including a sleep onset phenomenon and an imminent sleep phenomenon) such as subjective sleepiness and awake running state are detected. It is set to determine a. This determination result is set so that the determination output means 616 can display it as a time-series waveform on a display attached to the dashboard (also used as the image display device 622 of the stimulus applying means 62 described below). Yes.

  The stimulus applying means 62 is a display with a speaker attached to the dashboard (that is, a device that serves as both the sound output device 621 and the image display device 622), and different sounds according to the above three-stage determination results, It is set to display images. Further, the vehicle image display devices 622 of Experimental Examples 1 and 3 to 5 also serve as input means for the touch panel, and subjectively felt that they were “tired”, “I want to sleep”, “I am dull”, etc. In this case, when the corresponding display button is pressed, the driver's subjectivity is input.

(Experimental example 1)
The test subject is a healthy man in his 50s. FIG. 11 is a diagram showing a part of the result of an actual vehicle running experiment conducted on February 16, 2013. (A) of FIG. 11 is a diagram showing a slope time series waveform (zero cross slope time series waveform) obtained from the zero cross slope time series waveform calculating means 613, and (b) is a smoothed differential waveform of (a). (C) is a diagram showing a time-series waveform of the distribution rate of the zero-cross slope time-series waveform analyzed by the distribution rate calculating means 615, and (d) is a determination displayed on the display. It is the figure which showed the time series waveform of the result. The black triangle mark “Drowsiness Check” shown in the determination result of (d) is the timing when the subject feels drowsy and presses a button on the touch panel.

  From (a), the zero-cross slope time-series waveform shows a convergence tendency in which the period in which the amplitude decreases near 120 to 133 minutes becomes longer, and then shows a diffusion tendency in which the amplitude gradually increases and the period becomes shorter. I understand that. This is shown in the determination result of (d), and it can be seen that the state of the determination level 6 is intermittent from 127 to 136 minutes and the arousal level is lowered. In the distribution rate of (c), the distribution rate of the activity adjustment signal 0.0053 Hz exceeds the distribution rate of the function adjustment signal 0.0017 Hz at around 127 minutes and around 133 minutes, and it is determined that the fatigue level is rising. As a result, the stimulus applying means 62 is driven and applied with a stimulus in the vicinity of 127 minutes and in the vicinity of 133 to 136 minutes. By applying this stimulus, arousal is induced, and the determination level is between 1 and 3 for 137 to 142 minutes. In the vicinity of 136 to 144 minutes, the distribution ratio of the function adjustment signal 0.0017 Hz and the distribution ratio of the activity adjustment signal 0.0053 Hz are substantially constant, so that the level drops to the determination level 4 in the vicinity of 142 to 143 minutes. Stimulation is applied at that time, thereby inducing arousal, and thereafter a high arousal state is maintained. After 140 minutes, the amplitude of the zero-cross slope time series waveform is larger than that in (a), and it is determined that the point is in an uplifted state at this point.

  According to the present experimental example, in the rising period of the fatigue level and the continuation period of the same biological state, the stimulus applying means 62 is operated to apply the stimulus, thereby arousal induction can be achieved. It can be seen that it can be induced.

(Experimental example 2)
FIG. 12 is an example in which a running experiment conducted on August 18, 2012 by the same subject as in Experimental Example 1 was analyzed. In this experiment, the stimulus imparting means 62 is not provided, and no stimulus is imparted during the fatigue level rising period or the continuation period of the same biological state.
From (a), the amplitude of the zero-cross slope time series waveform decreases from 8 to 13 minutes and shows a convergence tendency that the period becomes longer, and the amplitude increases after 27 minutes. It can be determined that this falls into a hypoxic state in the vicinity of 23 to 25 minutes, and then the sleep onset symptom has developed. From (c), the distribution rate of the function adjustment signal 0.0017 Hz suddenly decreases in the vicinity of 30 to 32 minutes, and the distribution rate of the activity adjustment signal 0.0053 Hz also increases. It is determined that it has developed.
From the above, in the determination result of (d), although the initial level is transitioned at the determination levels 1 to 3, it tends to gradually decrease from around 18 minutes, and the determination levels 4 to 6 are greatly increased around 22 to 33 minutes. Has increased. Then, the user presses the drowsiness check button in the vicinity of 33 minutes, and it can be seen that strong sleepiness has occurred.

  That is, in this case, as in the present invention, when the stimulus applying means 62 is operated and the stimulus is not applied in the rising period of fatigue, the continuation period of the same biological state, etc., the arousal level does not increase. It turns out that it shifts to a consciousness state, and also the sleep symptom phenomenon is expressed.

(Experimental example 3)
The test subject was a man in his thirties, but he was not feeling well during the experiment and felt considerable fatigue. FIG. 13 shows experimental data of this subject. However, since the input vehicle for subjectivity identification is not attached to the vehicle used in Experimental Example 2, the subjectivity (attention reduced state, hypoxia state, sleepiness state, etc.) shown in words in FIG. Is shown. This is due to the self-report of the subject after the experiment.

  Up to about 16 minutes, the distribution rate of the function adjustment signal 0.0017 Hz is higher than that in (c), and the zero-cross slope time series waveforms in (a) and (b) also have fluctuations with moderate amplitude. You can see that you are driving in a relatively relaxed state. In the vicinity of 16 minutes 30 seconds, a determination level 10 that predicts a hypoxia state is output in the determination result, and a stimulus corresponding to the determination level 10 is output. At this time, as shown in (c), the distribution rate of the activity adjustment signal 0.0053 Hz increases and the distribution rate of the function adjustment signal 0.0017 Hz rapidly decreases. Thereafter, in (c), the function adjustment signal The distribution rate of 0.0017 Hz becomes small, and the distribution rate of the activity adjustment signal 0.0053 Hz also decreases and remains at a substantially constant value up to around 20 minutes. The amplitudes of the zero-cross slope time series waveforms of (a) and (b) are also relatively small. From this, it is presumed that the stimulus in the vicinity of 16 minutes and 30 seconds described above was output in response to the determination (predictive caution forecast) for predicting a hypoxic state.

  13A and 13B, the amplitude of the zero-cross slope time-series waveform decreases after 19-20 minutes, and then increases rapidly from 21 minutes to 25 minutes. In FIG. There is a big change that the distribution rate of the signal 0.0053 Hz rapidly increases and the function adjustment signal 0.0017 Hz rapidly decreases. Correspondingly, as shown in (d), a stimulus corresponding to the judgment level 11 is output at around 21 minutes and around 22 minutes. This is a warning because the subject's arousal level has decreased and drowsiness has increased. Thereafter, the amplitude of the zero-cross slope time series waveform of (a) and (b) is increased, the change of the distribution rate is also increased, and the distribution rate of the activity adjustment signal 0.0053 Hz is relatively large, and the function adjustment signal 0 Since the distribution rate of .0017 Hz is relatively small, the state where it cannot be relaxed continues. That is, it is in a state of resisting considerable sleepiness, and stimulation is applied intermittently at around 24 minutes 30 seconds, around 25 minutes 30 seconds, and around 27 minutes.

  As shown in the subjectivity of (a), this test subject determined that drowsiness increased to around 28 minutes and it was difficult to continue driving himself, and the intermittent warning described above was given. Paused.

  From this result, the stimulus warning of 16 minutes and 30 seconds has a difference of 1 minute and 30 seconds from the time of 18 minutes when the subjective hypotension was felt, and it can be said that there is still a slight difference, but around 21 minutes, The timing of the stimulus warning around 22 minutes is almost coincident with the time zone when the user feels sleepy. Since then, the subject continues to feel drowsiness subjectively, so the stimuli at around 24 minutes 30 seconds, around 25 minutes 30 seconds, and around 27 minutes almost coincide with the subjectivity.

  Therefore, if the degree of coincidence between the subjective timing and the stimulus output timing is within a predetermined time, within one minute from this experimental example, It can be seen that it is preferable to stop driving by applying a simple stimulus.

(Experimental example 4)
Experimental example 4 is the data of a healthy male in his 50s, the same as experimental example 1. 14 to 16 are diagrams showing analysis results in the driving support device 60 when traveling from Higashihiroshima City to Fukui Prefecture on February 14, 2013. FIG. (A) of each figure is the figure which showed the inclination time series waveform (zero cross inclination time series waveform) obtained from the zero cross inclination time series waveform calculating means 613, (b) is the smoothed differential waveform of (a). (C) is a diagram showing a time-series waveform of the distribution rate of the zero-cross slope time-series waveform analyzed by the distribution rate calculating means 615, and (d) is a determination displayed on the display. It is the figure which showed the time series waveform of the result.

  First, from FIG. 14 (a), during the period of 5 to 75 minutes, both the zero-cross slope time-series waveform and the peak slope time-series waveform change with an appropriate amplitude and period, and are moderately active in the endocrine system. The biological fluctuation of / f is also maintained, so that it can be determined that the arousal state is maintained without feeling much drowsiness, but this is due to the stimulus of the stimulus applying means 62.

  Specifically, when the state of the time zone is considered in detail with reference to FIG. 14C, the activity adjustment signal 0.0053 Hz exceeds the distribution ratio of the function adjustment signal 0.0017 Hz in about 26 minutes, and the increased state of fatigue is observed. Has occurred. In the determination result, determination level 7 is determined in the vicinity of the time point, that is, from 22 to 27 minutes, and the stimulation corresponding to the determination level is applied three times.

  From 27 minutes to 35 minutes and from 36 minutes to 48 minutes, the state where the function adjustment signal 0.0017 Hz exceeds the distribution ratio of the activity adjustment signal 0.0053 Hz continues. Meanwhile, in the determination result, the state determined as the determination level 7 to 8 is intermittently output a plurality of times, and the stimulus corresponding thereto is intermittently applied. In the vicinity of 50 to 55 minutes, the activity adjustment signal 0.0053 Hz tends to exceed the function adjustment signal 0.0017 Hz, and the stimulation of the determination level 8 is intermittently applied. In the vicinity of 65 minutes to 73 minutes, the state in which the function adjustment signal 0.0017 Hz exceeds the distribution rate of the activity adjustment signal 0.0053 Hz continues, and determination level 8 is determined around 68 minutes and 71 minutes. The stimulus according to it is given. In the vicinity of 75 minutes, the activity adjustment signal 0.0053 Hz tends to exceed the function adjustment signal 0.0017 Hz, the determination level 8 is determined, and a stimulus corresponding thereto is applied.

  That is, when the activity adjustment signal 0.0053 Hz tends to exceed the function adjustment signal 0.0017 Hz, the fatigue level is increased. In a state where .0017 Hz exceeds the distribution rate of the activity adjustment signal 0.0053 Hz, stimulation is applied so that the state of enhanced sympathetic nerve function continues appropriately, that is, to maintain the arousal state so that they do not deviate from each other. ing. As a result, as shown in FIG. 14 (a), during the period of 5 to 75 minutes, both the zero-cross slope time series waveform and the peak slope time series waveform change with an appropriate amplitude and cycle, and the endocrine system It is moderately active and maintains 1 / f biological fluctuations, thereby maintaining an arousal state without feeling much drowsiness.

  From 90 to 105 minutes and from 110 to 120 minutes in FIG. 15A, the amplitude of the zero-cross slope time-series waveform increases rapidly, and then continues with the amplitude increasing from 120 to 145 minutes. Since this amplitude is greater than or equal to a predetermined value set in advance, the determination output means 616 determines that these timings are in the uplift state. When this is compared with the distribution rate of the zero-cross slope time-series waveform in FIG. 15C, the activity adjustment signal 0.0053 Hz exceeds the distribution rate of the function adjustment signal 0.0017 Hz in the vicinity of 85 minutes and 95 minutes, and the increase in fatigue. A state has occurred. In terms of the determination result, a stimulus of determination level 7 is applied in the vicinity of 85 minutes. On the other hand, between 85 and 94 minutes and between 96 and 145 minutes, the function adjustment signal 0.0017 Hz exceeds the distribution rate of the activity adjustment signal 0.0053 Hz. Stimulations corresponding to determination levels 7 to 8 are applied to minutes, 125 minutes, 132 minutes, 135 minutes, 137 minutes, and 143 minutes. By this stimulation, the state of enhanced sympathetic nerve function continues, that is, the arousal state is maintained for a relatively long time.

  In the vicinity of 150 to 160 minutes from FIG. 15 (a), the amplitude of the zero-cross slope time series waveform is small, and this is below a predetermined value set in advance, so it is determined that the uplift state has ended. Therefore, at 156 minutes and around 158, a stimulus of determination level 7 is applied, and the wakefulness is being led. However, as shown in FIG. 16A, the amplitude of the zero-crossing tilt time series waveform increases rapidly after 165 minutes. When compared with the amplitude before that, it increases at a larger ratio than when it is determined to be an uplifted state, and is determined to be a predictive sleep phenomenon. In FIG. 16C, after 175 minutes, there is a sudden rise in the function adjustment signal 0.0017 Hz and a sudden drop in the activity adjustment signal 0.0053 Hz, and then the distribution rate of the function adjustment signal 0.0017 Hz is This greatly exceeds the distribution rate of the adjustment signal 0.0053 Hz. From this point as well, it can be determined that the sleep onset sign phenomenon has occurred after 175 minutes.

  Looking at the determination results, as shown in FIG. 16 (d), the stimulation at the determination level 8 is applied continuously or intermittently around 165 to 175 minutes. This is because the stimulus applying means 62 is continuously driven in order to strongly encourage the driver to take a rest because the elevating state has ended and a sleep onset sign phenomenon has been detected.

  According to Experimental Example 4, when it is determined that the degree of fatigue is increased, when it is determined that the same biological state has been continued for a predetermined time or more, a stimulus is applied, and the stimulation is randomly performed at a necessary timing. It was found that the driver's activities can be rhythmized and the driver can maintain the awake state for a long time. In addition, even when an uplifted state is reached, it can be stimulated to maintain that state, and drowsiness is difficult to visit. On the other hand, drowsiness may still occur due to long-term activity, but at that time, that is, when a sleep onset sign phenomenon is detected, a break can be urged with a stronger warning.

(Experimental example 5)
17 to 19 are data on a return route from Fukui Prefecture to Higashihiroshima City on February 16, 2013. FIG. In addition, in the above experimental example 1, in this experimental example 4, data of 120 to 150 minutes is taken as a typical case in which arousal induction is performed by the stimulus applying means 62 and analyzed in detail.
From FIG. 17D, a stimulus corresponding to the judgment level 7 is output around 20 minutes and around 30 minutes. As shown in FIG. 17 (c), this is the timing when the distribution rate of the activity adjustment signal 0.0053 Hz indicating an increase in the degree of fatigue exceeds the distribution rate of the function adjustment signal 0.0017 Hz. It works to do.

  In FIG. 17 (d) and FIG. 18 (d), stimulations of determination levels 7 to 8 are given at 35 to 45 minutes, 48 to 59 minutes, and 73 to 90 minutes. As shown in FIG. 17 (c) and FIG. 18 (c), the distribution ratio of the function adjustment signal 0.0017 Hz continues to exceed the distribution ratio of the activity adjustment signal 0.0053 Hz. It functions as a stimulus for sustaining the arousal state that is in an enhanced state.

  In FIG. 18 (d), stimulations of judgment levels 7 to 8 are applied at 93 to 105 minutes. This is an activity adjustment signal indicating an increase in fatigue level as shown in FIG. 18 (c). This is the timing when the distribution rate of 0.0053 Hz exceeds the distribution rate of the function adjustment signal 0.0017 Hz, and functions to induce wakefulness. During the period of 110 to 118 minutes, the distribution rate of the function adjustment signal 0.0017 Hz exceeds the activity adjustment signal 0.0053 Hz, which is a stimulus for maintaining arousal. A stimulus for awakening induction due to an increase in the degree of fatigue is applied from 127 to 136 minutes, and a stimulus for maintaining arousal is applied from 142 to 148 minutes. In the zero cross slope time series waveform of FIG. 18 (a), the amplitude value changes over a predetermined value from 140 to 154 minutes, and at this time point, it is determined as an uplift state. Stimulation was applied between 155 and 160 minutes, which is accompanied by the end of the uplifted state, and as shown in FIG. 18 (c), activity showing an increase in fatigue at 154 to 156 minutes. An increase in the distribution rate of the adjustment signal 0.0053 Hz is observed.

  Thereafter, from FIGS. 19C and 19D, stimulation is applied at around 173 to 176 minutes and 185 to 192 minutes where the degree of fatigue increases, which is the duration of the enhanced state of sympathetic nerve function 180. Stimulation is applied at around 185 minutes and 200 minutes. Thereby, the awakening state is maintained. Therefore, after about 170 minutes, the endocrine system is moderately active and 1 / f biological fluctuation is maintained by a randomly applied stimulus at a predetermined timing. As a result, a zero-cross slope time series waveform is maintained. In addition, both the peak slope time-series waveform change with an appropriate amplitude and cycle, and the awake state is maintained without feeling much sleepiness.

(Experimental example 6)
Experimental example 6 is the result of an actual vehicle running test from Saijo, Higashihiroshima City to Shimane Truck Association on February 25, 2013, as shown in FIGS. In addition, the test subject attached | subjected the simple electroencephalograph, and also measured the electroencephalogram. 20 (a) to 22 (a) are diagrams showing a slope time series waveform (zero cross slope time series waveform) obtained from the zero cross slope time series waveform calculation means 613, and FIGS. 20 (b) to 22 (b) are views. FIG. 20 is a diagram showing a time-series waveform of the distribution ratio of the zero-cross slope time-series waveform analyzed by the distribution-rate calculating means 615, and FIGS. 20 to 22 (c) are time-series waveforms of determination results displayed on the display. FIGS. 20 to 22 (d) are diagrams illustrating α-wave time-series waveforms, and FIGS. 20 to 22 (e) are diagrams illustrating θ-wave time-series waveforms. It is.
(A) in FIGS. 23 to 25 show the data of FIGS. 20 to 22 together with the peak inclination time series waveform, and (f) to (g) in FIGS. 23 to 25. Is the same data as (d) to (e) of FIGS. (B) in FIG. 23 to FIG. 25 is a distribution rate of 0.0017 Hz obtained from the zero-cross slope time series waveform and the peak slope time-series waveform, and (c) in FIGS. 23 to 25 are zero-cross slope time series waveforms. And the distribution rate of 0.0035 Hz obtained from the peak inclination time series waveform, and (d) in FIGS. 23 to 25 are the distribution ratio of 0.0053 Hz obtained from the zero-cross inclination time series waveform and the peak inclination time series waveform. is there.

  20B and 20C, the stimulus in the vicinity of 70 minutes is a stimulus for performing the awakening induction at the timing of increasing the fatigue level because the activity adjustment signal 0.0053 Hz is rising. 20 (b) and 20 (c), stimulation at 5 to 68 minutes and 73 to 80 minutes, excluding the vicinity of 70 minutes, was applied to maintain an enhanced state of sympathetic nerve function, that is, to maintain an arousal state. It can be judged as a stimulus. However, referring to FIGS. 23B to 23D, the zero crossing of the function adjustment signal 0.0017 Hz, the fatigue acceptance signal 0.0035 Hz, and the activity adjustment signal 0.0053 Hz is around 27 minutes, 45 minutes, and 72 minutes. This is a time zone in which at least one of the distribution rates obtained from the slope time series waveform and the peak slope time series waveform tends to deviate. Therefore, in these time zones, it can be determined that the fatigue level is increased due to the difference between the activity level of the sympathetic nerve function and the activity level of the parasympathetic nerve function, and it can be considered that the stimulus is given to induce awakening accordingly. .

  21 (b) and 21 (c), an increase in the distribution rate of the activity adjustment signal 0.0053 Hz is observed in the vicinity of 80 to 83 minutes, and a corresponding stimulus is applied. On the other hand, as shown in FIGS. 24B and 24E, in the vicinity of 83 to 85 minutes, there is a discrepancy between the distribution rate of 0.0017 Hz obtained from the zero-cross slope time series waveform and the peak slope time series waveform, and the corresponding stimulus. Is granted. About 90 minutes, as shown in FIGS. 21 (b) and 21 (c), the stimulation by the continuous determination of the enhanced state of the sympathetic nerve function is given. In the vicinity of 97 minutes, stimulation due to an increase in the activity adjustment signal 0.0053 Hz is applied, in the vicinity of 98 to 103 minutes, stimulation due to continued determination of the enhanced state of the sympathetic nerve function is applied, and in the vicinity of 108 to 114 minutes, the activity adjustment signal 0.0053 Hz Stimulation due to an increase in the increase is applied, stimulation in the vicinity of 122 to 129 minutes is applied by the continuous determination of the enhanced state of the sympathetic nerve function, and stimulation in the vicinity of 130 to 132 minutes is applied due to the increase in the activity adjustment signal 0.0053 Hz. Yes. In the vicinity of 135 to 143 minutes, stimulation based on the continuation determination of the enhanced state of the sympathetic nerve function is given, and thereafter, as can be seen from FIG. It has become. This is considered to be triggered by the stimulation in the vicinity of 135 to 143, and the uplifted state continues until 155 minutes. Thereafter, from FIG. 22 (a), the amplitude suddenly decreases in the vicinity of 160 minutes, and it is determined that the uplifted state has ended, and a stimulus is applied. In the vicinity of 160 to 168 minutes and in the vicinity of 175 to 180, the amplitude of the zero-cross slope time series waveform increases rapidly, and those obtained from the zero-cross slope time series waveform and the peak slope time series waveform in the distribution ratio of each frequency component It is determined that the sleep symptom phenomenon has appeared. Therefore, in these time zones, stimulations of determination levels 7 to 8 are intermittently applied to encourage a break.

  From Experimental Example 6, it is possible to more appropriately determine the increase in fatigue level by determining the deviation between the zero cross slope time-series waveform and the peak slope time-series waveform in the distribution ratio of each frequency component. It can be identified that it is possible to more appropriately apply the stimulus at the required timing. As a result, similar to Experimental Examples 1 and 4, a random stimulus is applied at an appropriate timing, which becomes a rhythmic feeling and can continue driving in a state of high arousal.

(Experimental examples 7 to 12)
26 to 29 show analysis data (experimental examples 7 to 10) of a male subject B in his 50s, and FIGS. 30 to 31 show analysis data (experimental) of a male subject A in his 50s. Examples 11 to 12) are shown. In addition, (a) of each figure is the figure which showed the inclination time series waveform (zero cross inclination time series waveform) obtained from the zero cross inclination time series waveform calculating means 613, (b) is the smoothing differentiation of (a). It is the figure which showed the waveform, (c) is the figure which showed the time series waveform of the distribution rate of the zero cross inclination time series waveform analyzed by the distribution rate calculating means 615, (d) is displayed on a display. It is the figure which showed the time series waveform of the determination result. The black triangle mark “Drowsiness Check” shown in the determination result of (d) is the timing when the subject feels drowsy and presses a button on the touch panel.

  In Experimental Example 7 of FIG. 26, the zero-cross slope time-series waveform fluctuates with a medium amplitude until around 23 minutes, and the determination result of (d) also changes in the vicinity of levels 1 to 4 to some extent. It turns out that it is in a healthy state. However, since the distribution rate of the activity adjustment signal 0.0053 Hz exceeds the distribution rate of the function adjustment signal 0.0017 Hz around 14 to 16 minutes, a stimulus corresponding to level 7 is applied around 15 minutes 30 seconds. ing. Before this stimulus was applied, it was often determined that the state was near level 4. However, after this stimulus was applied, the determination near level 1 was increased, and the degree of arousal was increased by the stimulus. You can see that

  In the vicinity of 23 minutes, stimulation is given by the determination of level 11. This is because the state in which the distribution rate of the function adjustment signal 0.0017 Hz exceeds the distribution rate of the activity adjustment signal 0.0053 Hz is continued, and thereafter, the state of the determination level 1 is increased and the arousal is induced. However, the subject presses the “Drowsiness Check” button around 25 minutes and 27 minutes. However, there is a difference of 2 minutes or more from the timing of applying the stimulus, which is inconsistent with the subjectivity, so it can be said that the feeling of fatigue is still small.

  In the vicinity of 28 minutes, the distribution rate of the function adjustment signal 0.0017 Hz is greatly reduced and the distribution rate of the activity adjustment signal 0.0053 Hz is increased. Therefore, the determination of level 8 is made and stimulation is given, and the arousal level is increased. doing. In the vicinity of 33 to 40 minutes, the distribution ratio of the function adjustment signal 0.0017 Hz continues to exceed the distribution ratio of the activity adjustment signal 0.0053 Hz, and a stimulus corresponding to level 7 is intermittently applied. ing. Due to the intermittent application of the stimulus, the average level of the subject's arousal level is very high over a period of 35 to 40 minutes (see (d)).

  Like this test subject, when he is healthy and in good health, the number of times he felt subjective fatigue was extremely low, and the stimulation was used as an aid to comfortable driving, indicating that he was able to drive with almost no fatigue. .

  In Experimental Example 8 of FIG. 27, the determination level 6 is large and the fatigue level is relatively high, but the stimulations of the determination levels 7 to 8 are intermittently applied as described above. The subject presses the drowsiness check button in the vicinity of 33 to 34 minutes, which coincides with the stimulation timing in the vicinity of 34 minutes, and can be determined as having a strong sleepiness. After that, intermittent stimulation is given correspondingly over 35 to 42 minutes, and awakening is induced.

  In Experimental Example 9 of FIG. 28, the arousal level is high until around 17 minutes, but due to the subsequent decrease in the arousal level, intermittent stimulation is applied over 19 to 22 minutes to increase the arousal level. After that, since the stimulus is given at random timing, a high arousal level corresponding to the determination levels 1 to 2 is maintained.

  Similarly to Experimental Example 9, Experimental Example 10 in FIG. 29 has a relatively high arousal level in the first half, but the arousal level has decreased from around 27 minutes. Corresponding to this, stimulation is applied intermittently thereafter, and as a result, a high arousal level of judgment levels 1 to 2 is maintained.

  In Experimental Example 11 of FIG. 30, as can be seen from the determination result, around 16-18 minutes, around 20-24 minutes, around 28 minutes, around 31 minutes, around 38-46 minutes, around 52-58 minutes, around 63- After that, the stimulus level is increased by a stimulus of about 66 minutes, and arousal induction is performed. In the vicinity of 16 minutes, the sleepiness check button is pressed, and it is reported that micro sleep has occurred, but the degree of arousal has increased since then by stimulation before and after that.

  In Experimental Example 12 of FIG. 31, the determination level 6 is high and the fatigue level is considered to be high, but the stimulations of the determination levels 7 to 8 are intermittently applied accordingly. As a result, after 65 minutes, the degree of arousal has increased, and stimulation is no longer applied.

(Experimental example 13)
32 to 36 are plots of data (indicated by white marks) for a total of 400 professional drivers (truck drivers) obtained by the determination output means 616 of the driving support device 60 described above. (A) of each figure is the data which plotted the frequency | count that the increase degree of the sympathetic nerve became more than predetermined in a zero cross inclination time series waveform, a distribution rate, etc., and (b) is zero cross inclination. Data obtained by plotting the time series waveform by absolute value processing and integrating the area for each time as a fatigue level as a fatigue level, and (c) is a zero cross slope time series waveform, distribution rate In this case, the number of times that the arousal level becomes a predetermined value or less is plotted in relation to the measurement time. In addition, the curve drawn in the range where the white mark of (b) is plotted is the boundary position corresponding to the case where it is determined that the fatigue is present before the operation, starting from the curve on the left side of the figure, the operation start Boundary position corresponding to the case where fatigue increases after 3 hours, boundary position corresponding to the case where fatigue increases after 6 hours from the start of operation, boundary corresponding to the case where fatigue increases after 9 hours from the start of operation Indicates the boundary position of the case where operation can be continued with less position and feeling of fatigue.

  By collecting data of a large number of people in this way, depending on the position where the degree of sympathetic nerve enhancement, fatigue level, and degree of wakefulness is plotted in relation to the measurement time (driving time), It is possible to determine whether each state is high or low. Therefore, these data are stored as basic data in the storage unit of the driving support device 60, and the subject's newly measured data is plotted according to where the subject's data is plotted in relation to the measurement time. The biological state at the time of measuring data can be determined.

  In addition, FIG. 32 shows the analysis result when subject B drove from Higashihiroshima City to Fukui on February 14 in black circles, and FIG. 33 drove on February 15 the next day. The analysis results are indicated by black circles. February 14th is the data when the physical condition is excellent and the uplifted state occurs during the measurement time, and the following February 15th is the feeling of fatigue due to the recoil of the uplifted state of the previous day, but the physical condition is good It is data in a state where arousal level is high.

  FIG. 34 shows the analysis result when subject B returns from Fukui direction to Higashihiroshima city on February 16, with black circle marks, and FIG. 35 shows the analysis result on the following February 17 with black circle marks. It is shown. February 16th is a state in which fatigue due to a business trip remains, but it is data when an uplifted state occurs on the way. Compared with data of February 14th to 15th, the degree of decrease in arousal level There are many occurrences. The data on February 17 shows that the feeling of fatigue remains subjectively due to the recoil of the uplifting state of the previous day, but there is a recovery tendency, and the degree of increase in sympathetic nerves, degree of fatigue, and degree of arousal are all It is low.

  FIG. 36 is a plot of the data of subject H on the data of 400 professional drivers in the same manner as described above. From this figure, it can be determined that the degree of enhancement of the sympathetic nerve is relatively high, while the degree of decrease in fatigue and arousal level is low.

  However, when the data of the subject H is compared with the data of the subject B shown in FIGS. 32 to 35, the plot position of the black circle mark of the subject H in FIG. It is closest to the plot position of the black circle. Subject B's February 17: “Although there is still a feeling of fatigue due to the recoil of the previous day's uplift state, there is a recovery tendency, and the degree of increase in sympathetic nerves, fatigue, and degree of arousal are In other words, when such data appears, the “uplifted state occurred a few hours ago (that is, data on February 16). It can be defined as Therefore, although the measurement time of the data of FIG. 36 of the subject H is about 3 hours, it can be estimated that a biological state corresponding to the “uplifted state” had occurred before that. Of course, this is only an example. Actually, by preparing a lot of data for each state such as the subject B and storing it in the storage unit, it is possible to determine which pattern the data to be examined approximates. It is possible to determine the state more accurately by determination, and it is possible to more accurately estimate the state several hours before the state.

(Experimental example 14)
Next, an experiment was conducted to verify the relevance to the electroencephalogram and the distinction between the drowsiness sign phenomenon (including the sleep onset sign phenomenon and the imminent sleep phenomenon) and the elevated state. In the experiment, a nonsmoker who was a non-smoker who got an informal outlet, measured EEG, APW, and subjective evaluation while driving a car test course. Subjective evaluation was in advance instructed to make a self-report when changes occurred regarding “degree of awakening”, “presence of sleepiness”, and “degree of relaxation”. Experimental results are shown in FIGS.

  FIG. 37 shows the traveling result of the subject A, and the measurement time is 180 minutes. In the test run, the result was that sleepiness occurred within 60 to 65 minutes. In the same time zone shown in FIG. 37A showing the APW zero-cross slope time-series waveform, the amplitude decreases and the period becomes longer. In the time zone of 43.5 to 61.8 minutes, the amplitude and period are disturbed. On the other hand, in FIG. 37 (b), 0.0053 Hz was dominant immediately after this time zone, and 0.0017 Hz was greatly reduced. The [theta] wave in FIG. 37 (c) changes stably up to just over 40 minutes, and shows an increasing tendency at over 40-70 minutes. The α wave in FIG. 37 (d) shows an increasing trend up to a little over 70 minutes. Therefore, from the viewpoint of how the brain waves fluctuate, it is considered that the subjects in this time zone were drowsy while in a relaxed state, but they do not show how the α and θ waves fluctuate oppositely. Although there is an increase in the θ wave, it is considered that the patient has fallen asleep. It should be noted that the portions where the electroencephalogram data in FIGS. 37 (c) to 37 (e) are missing are artifacts. From the above, it can be seen that a decrease in amplitude and a longer period in the APW zero cross slope time-series waveform may indicate the occurrence of sleepiness.

  In addition, it is reported that in the time zone surrounded by the solid line of 131.7 to 160 in FIG. In FIG. 37A, the amplitude increases from A to B (A = 0.000289, B = 0.001291) in the time zone indicated by the solid line, and the period is increased from C to D (C = 4). From 5 minutes to D = 5.7 minutes). That is, it changed from 0.0037 Hz to 0.0029 Hz. On the other hand, in the zero cross distribution rate of APW in FIG. 37 (b), 0.0035 Hz tends to decrease, 0.0017 Hz tends to increase, and 0.0053 Hz is stable. The θ wave in FIG. 37 (c) tended to decrease during this time period and remained stable after 160 minutes. The α wave in FIG. 37 (d) changed in conjunction with the θ wave. Therefore, in the time zone of 131.7 to 150 minutes, although relaxed and slightly drowsy, the β wave was high as shown by the arrow in FIG. The possibility is inferred.

  FIG. 38 is a result of comparing the APW zero cross distribution rate (0.0017 Hz) and the θ wave power spectrum obtained by a simple electroencephalograph. The distribution rate of 0.0017 Hz increases to 40 to 70 minutes, 95 to 120 minutes, and 130 to 150 minutes. The θ wave also increases in the same time zone. From these, it can be seen that the distribution rate of 0.0017 Hz and the θ wave showed the same tendency. From the above, it is suggested that changes in amplitude and period can capture a sign of sleepiness and an elevated state.

  Here, in order to discriminate these, fast Fourier transform (hereinafter referred to as FFT) was performed on the zero-cross slope time series waveform. And it decided to discriminate | determine two states using this quantification point method with respect to this result. FIG. 39A shows the result of performing FFT (both axis logarithmic display) on the time zone in which the subject A's drowsiness sign is captured. The analysis interval is 43.5 to 61.8 minutes. Increased sympathetic activity on the low frequency side (FIG. 39 (a) range (1)), medium frequency band (FIG. 39 (a) range (2)), and high frequency side (FIG. 39 (a) range (3)). The waveform shown was seen. As a result of using the quantification point method of the human state, the result of FIG.

  FIG. 39B shows the result of performing FFT (both axis logarithmic display) for the time zone in which the subject A's uplifted state was captured. The analysis interval is 131.7 to 150 minutes. A waveform showing an increase in sympathetic nerve activity was observed on the high frequency side (range (3) in FIG. 39 (b)). As a result of using the quantification point method of the human state, the result of FIG. 39B was 7 points.

From the above, it is considered that the sign of drowsiness and the uplifting state in which amplitude fluctuation occurs and the amplitude waveform is mixed can be discriminated by using the quantification point method. Therefore, the same analysis was performed for other conditions. FIG. 40 shows the average value of the quantification score in the time zone in which the drowsiness sign phenomenon appeared and the quantification score in the time zone in which the uplift state appeared. On the other hand, as a result of the test of the difference between the average values of two samples assuming equal variance, p = 6.47 × 10 ⁻¹² and p <0.05, and the quantification score of the drowsiness sign phenomenon and the uplift state It was suggested that there was a significant difference in. From the above, the drowsiness predictor phenomenon and the uplift state can be captured by mixing the amplitudes in the zero-cross slope time-series waveform and by increasing the frequency period, and these two states can be discriminated by the quantification points. all right.

  Here, FIG. 41A shows the amplitude of the zero-cross slope time series waveform, and FIG. 41B shows the result of calculating the incidence of sleepiness using Bayesian estimation with respect to the frequency. From this, it was estimated that drowsiness occurred in the range 1 to 3 or the range 7 to 10 where the amplitude was small, and drowsiness occurred in the range 1 to 3 or the range 7 to 10 where the frequency was low. Similarly, Bayesian estimation was used for the quantification points. FIG. 41 (c) shows the result of the probability of being in an uplift state, and FIG. 41 (d) shows the result of the probability of being a sleepiness sign phenomenon. Comparing FIGS. 41 (c) and (d), it was found that if the quantification score is low, there is a high possibility of being a drowsiness sign, and if the quantification score is high, there is a high possibility of being in an elevated state.

(Experimental example 15)
In order to verify the usefulness of sleepiness estimation using Bayesian estimation, further experiments were conducted with different conditions from the static state and the actual vehicle running state.

(Experiment in static state)
The fingertip volume pulse wave was measured using a finger clip probe (SR-5C, Konica Minolta) in order to capture the information of the peripheral circulatory system at the same time that APW was measured by the back body surface pulse wave measuring apparatus 1. Biometric data is recorded at 200 Hz. The test subjects were 12 healthy men from 26 to 40 years old (average age: 28.6, SD: 4.5). The patient was seated on a car seat equipped with the back body surface pulse wave measuring device 1 for 60 minutes, and the degree of sleepiness was evaluated every 5 minutes using the Visual Analog Scale (VAS). For the evaluation, a recording sheet with “very awake” at the left end of the 100 mm straight line and “very sleepy” at the right end is used, and the subject is in a position that matches his / her own state on the straight line. Mandatory to draw a vertical line.

  As a representative example of the experimental results, FIG. 42A shows the sleepiness evaluation of the subject A, FIG. 42B shows the power value slope time series of the fingertip plethysmogram, and FIG. 42C shows the APW frequency slope time series. Shown in As shown by the line frame in FIG. 42 (a), drowsiness becomes strongest 10 minutes after the start of measurement. The power value inclination time series waveform of the finger plethysmogram has a large amplitude and a long period as shown by the line frame in FIG. On the other hand, in the APW frequency gradient time-series waveform, as shown by the line frame in FIG. 42 (c), a large long-period amplitude was observed from 10 minutes to 20 minutes after the start of measurement. 42 (d) and 42 (e) show the results of fast Fourier transform of the dotted frame portions in FIGS. 42 (b) and 42 (c). It can be seen that the dominant frequencies are in agreement. From these results, it can be seen that the fingertip volume pulse wave and the APW slope time-series waveform have a large amplitude and a long period when drowsiness occurs. From this, it is considered that the presence or absence of drowsiness can be estimated from the amplitude and period of the APW frequency gradient time series waveform.

Next, using the Bayesian estimation method for the results of 12 subjects, the probability of being in a sleepy state was obtained from the amplitude characteristics of the APW frequency gradient time series waveform. Here, in the Bayesian estimation method, when there are a plurality of causes of certain data, the probability of each cause is obtained from the following equation (1). When replaced with this experiment, the data D is the amplitude of the APW frequency gradient time series, the cause H1 is an arousal state, and the cause H2 is a sleepiness state. It is assumed that the arousal state and the sleepiness state occur with the same probability (P (H ₁ ) = P (H ₂ ) = 1/2). The determination of the state of the subject was performed by normalizing the VAS value with the maximum value being 1 for each subject, 0.75 or less as an awake state, and 0.76 or more as a sleepy state.

Table 1 shows the results obtained by Bayesian estimation. The first column of Table 1 is the amplitude range of the APW frequency gradient time series, the second column is the number of data in the amplitude range of the wakefulness state, the third column is the number of data in the amplitude range of the sleepy state, the fourth column The eye is the sleepiness probability for each amplitude obtained from the equation (1). It can be seen that the sleepiness probability is high in the range where the amplitude of the APW frequency gradient time series is large. In addition, the sleepiness probability is high even in a range where the amplitude is small. Therefore, the APW frequency gradient time series has a small amplitude range of 5.95 × 10 ^{−5 to} 2.42 × 10 ⁻⁴ and a large amplitude range of 1.88 × 10 ^{−3 to} 2.24 × 10 ⁻³ . Therefore, the probability of sleepiness is high, and it is considered useful for estimating sleepiness.

(Experiment in actual vehicle running condition)
Using the Bayesian estimation probability table in Table 1, the sleepiness estimation was verified in an actual vehicle running experiment. However, compared to a static experiment performed in a resting state, the heart rate tends to increase in the driving state, and the heart rate variability is considered to increase. Therefore, it is necessary to change the threshold value captured in the static experiment, and it is necessary to capture a change larger than the change in the amplitude of the static experiment. Further, the AP body pulse wave measuring device 1 described above was used for collecting APW in an actual vehicle, but the amplitude division width was set large for the convenience of mechanical determination of the amplitude. The newly created probability table is shown in Table 2. In Table 1, the amplitude is in the range of 5.95 × 10 ^{−5 to} 4.96 × 10 ⁻⁴ and 1.81 × 10 ^{−3 to} 2.24 × 10 ⁻³ including the condition estimated to have high sleepiness probability. The estimated state at that time was the sleepy state, and the other amplitude range of 4.96 × 10 ^{−4 to} 1.81 × 10 ⁻³ was the awake state. The estimated state was compared with the actual state of the subject.

The subject who participated in this experiment was a 58 year old healthy man. The test course was driven for 40 to 100 minutes in an automobile with the back body surface pulse wave measuring device 1 mounted on the seat. This was done 6 times. Each time the subject felt sleepy, he pressed the time recording button installed in the car. FIG. 43 shows a representative example of the estimated state obtained by Bayesian estimation and the actual place where sleepiness occurs. The left vertical axis represents the estimated state, and the right vertical axis represents the state of the subject. The subject feels sleepy when estimated to be sleepy. During all 6 runs, there were 33 places that were estimated to be sleepy, of which 21 subjects felt sleepy. Table 3 shows the results of examining the relationship between the estimated state and the actual subject state in a 2 × 2 cross table. As a result of the chi-square test, P = 1.5 × 10 ⁻⁴ <0.05, and it can be determined that drowsiness can be estimated from the amplitude of the APW frequency gradient time series.

  From these, it was found that by using the Bayesian estimation method, when the APW frequency gradient time-series waveform has a small amplitude and a large amplitude, the probability of feeling sleepy is high. Therefore, by using the estimated probability, it is possible to estimate the sleepiness of the driver when the vehicle travels.

(Experimental example 16)
In the stimulus applying means 62, the sound output device (mechanism portion for outputting sound including a speaker) 621 for applying auditory stimulus and the image display device (image including display) for applying visual stimulus. An experiment on the effectiveness of moving image stimulation by 622 was performed. The image display device 622 is installed at a position that is easy to see from the subject on the dashboard, and is set so that images as shown in FIGS. 5A to 5F are displayed corresponding to each biological state. The back body surface pulse wave measuring device 1 was set in the driver's seat, and a 58-year-old healthy male subject drove the test course for 120 minutes for an experiment. Each time the subject felt drowsy, the observer reported and the time was recorded by the observer.

  First, FIG. 44 shows an estimated state obtained by Bayesian estimation using Table 2 of Experimental Example 15, actual drowsiness occurrence locations, and stimulus presentation locations. The left vertical axis represents the estimated state obtained by Bayesian estimation, and the right vertical axis represents the state of the subject. A triangle mark represents a place where a stimulus is presented by the stimulus applying means 62. During this run, the test subject stated that “the degree of arousal has decreased” four times between 65 minutes and 90 minutes. The estimated state is frequently sleepy between 63 and 88 minutes, and sleepiness can be estimated by Bayesian estimation.

  Also, the subject did not report drowsiness after 90 minutes. On the other hand, the number of stimulus presentations has increased since 80 minutes. In order to investigate the relationship between the subject's sleepiness and the number of stimulus presentations, the wakefulness level decreased from 65 minutes to 65 minutes when the experiment started and the wakefulness level increased from 65 minutes to 120 minutes. The relationship with the number of stimulus presentations per 5 minutes is shown in Table 4 as a 2 × 2 cross table.

  As a result of Fisher's exact test, P = 0.033 <0.05, the number of stimulus presentations may be related to an increase in the arousal level, and sound stimulation and animation stimulation induced the subject to arousal state. Can be estimated.

  In each of the above-described examples, the analysis is performed using the determination output means 616 set in the driving support device 60 that is an in-vehicle computer and the data stored in the storage unit thereof. Of course, the data can be set on the operation manager's computer and analyzed in the same manner as described above. In this case, data to be analyzed may be transmitted from the in-vehicle driving support device 60 to the operation manager's computer via a communication line (wireless or the like) so that the operation manager's computer can also analyze the data in real time. In addition, after the driving operation is completed, the data stored in the driving support device 60 may be taken out and analyzed afterward in the operation manager's computer. By collecting such data, the operation manager can grasp the situation at the time of driving for each driver, and can also use it to give advice for more appropriate driving.

DESCRIPTION OF SYMBOLS 1 Back body surface pulse wave measuring device 11 Core pad 12 Spacer pad 13 Sensor 60 Driving support device 61 Living body state determination means 611 Zero cross detection means 612 Peak detection means 613 Zero cross inclination time series waveform detection means 614 Peak inclination time series waveform detection means 615 Distribution Rate calculation means 616 Determination output means 62 Stimulus imparting means 621 Sound output means 622 Image display means

Claims (21)

A driving support device including a biological state determination unit that analyzes a back body surface pulse wave that is a biological signal of a driver measured by a biological signal measuring device and determines a biological state of the driver,
Having a stimulus applying means for applying a stimulus including at least one of an auditory, visual or olfactory stimulus to the driver;
The biological state determination means includes
From the index obtained by analysis of the back body surface pulse wave, it is determined a biological state that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness, and depending on the determination result, the stimulus applying means Determination output means for outputting a command to drive,
The determination output means detects the biological state determined to be a warning level that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness in order to induce awakening of the driver, and this It is configured to output a command to drive the stimulus applying means at both timings of detecting related signs related to a decrease in arousal level appearing before the appearance of a warning level biological state , and the biological level at the warning level When it is determined that the time difference between the detection timing of the state and the detection timing of the related signs is within a predetermined time, the stimulation applying means is instructed to apply a stimulus that prompts the driver to take a break. A featured driving support device. The stimulus applying means, the image display apparatus and driving support device both formed by both of an according to claim 1 for imparting a sound output device for imparting auditory stimuli, visual stimuli. The image display device can display a change in each biological state in a normal state, and when applying a stimulus for prompting arousal or resting according to a command from the determination output unit, a plurality of images are displayed. The driving support device according to claim 2 , wherein the driving support device is set so as to be displayed alternately to give a visual stimulus. Three to ten types of images for applying visual stimuli set in the image display device are set corresponding to each biological state, and two to four images are alternately displayed for each type. The driving support device according to claim 3 , wherein the driving support device is set to be performed. The determination output means includes
Zero-cross detection means for obtaining a time-series waveform of the frequency using a zero-cross point in the time-series waveform of the back body surface pulse wave obtained from the biological signal measuring device;
Peak detection means for obtaining a time series waveform of frequency using a peak point in a time series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
Zero-cross slope time-series waveform calculating means for calculating the time-series waveform of the frequency by sliding calculation of the time-series waveform of the frequency obtained by the zero-cross detection means;
A peak slope time series waveform calculating means for obtaining a slope time series waveform of the frequency by sliding calculation of the time series waveform of the frequency obtained by the peak detection means;
Based on at least one index of the slope time series waveform obtained from the zero cross slope time series waveform computing means and the slope time series waveform computing means obtained from the peak slope time series waveform computing means, various types appearing with a decrease in the arousal level The driving support device according to any one of claims 1 to 4, wherein a biological state is determined. The determination output means includes
Function analysis in the range of 0.001 to 0.0027 Hz is performed by frequency analysis of the slope time series waveform obtained from the zero cross slope time series waveform computing means or the slope time series waveform obtained from the peak slope time series waveform computing means. A signal, a fatigue acceptance signal in the range of 0.002 to 0.0052 Hz, and a frequency component corresponding to an activity adjustment signal of 0.004 to 0.007 Hz are extracted, and a distribution for obtaining a variation in the distribution rate of each of these frequency components The driving support device according to claim 5 , further comprising: a rate calculating unit that determines various biological states appearing with a decrease in the arousal level from a distribution rate of each frequency component obtained by the distribution rate calculating unit. The determination output means is configured such that, in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation means, the distribution ratio of the frequency component corresponding to the activity adjustment signal is the function adjustment. The driving support device according to claim 6 , wherein when the distribution ratio of the frequency component corresponding to the signal is exceeded, it is determined as a related sign related to the decrease in the arousal level. The determination output means, a distribution ratio of each frequency component obtained by using a gradient time series waveform obtained from the zero-crossing slope time-series waveform calculation means, the inclination at the time-series waveform obtained from the peak slope time-series waveform computation means When the distribution ratio of each frequency component obtained by using the comparison is between the distribution ratios of the corresponding frequency components, the activity of the sympathetic nerve function and the activity of the parasympathetic nerve function The driving support device according to claim 6 , wherein the driving support device determines that the related sign is related to the decrease in the degree of arousal that appears due to the divergence. The determination output means is a frequency at which the distribution ratio of the function adjustment signal corresponds to the activity adjustment signal in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation means. When the state of exceeding the component distribution rate continues for a predetermined time or more, or when the state of distribution of the activity adjustment signal exceeding the frequency component distribution rate corresponding to the function adjustment signal continues for a predetermined time or more The driving support device according to claim 6 , wherein the driving support device is determined to be a related sign related to a decrease in the arousal level. When the amplitude of at least one of the slope time series waveform obtained from the zero cross slope time series waveform computing means and the slope time series waveform obtained from the peak slope time series waveform computing means is a predetermined value or more, the determination output means Means for determining whether the driver is in an uplift state;
The driving support device according to claim 5 , wherein after determining that the uplifting state has been determined, the driving support device outputs a command for prompting the driver to take a break when the uplifting state is determined to have ended. Furthermore, subjective fatigue identification means consisting of input means operated when the driver feels subjectively fatigue, a reduction in alertness or a decrease in arousal level including sleepiness, is attached ,
The determination output means includes means for determining whether or not a time difference between an input timing of the subjective fatigue identification means and a stimulus output timing by the stimulus applying means is within a predetermined range,
The driving support device according to any one of claims 1 to 10 , wherein when the time difference is determined to be within a predetermined range, a command for prompting the driver to take a break is output to the stimulus applying unit. A generalized probability table of index fluctuations obtained by analysis of the back body surface pulse wave and biological states appearing corresponding to the index fluctuations is stored in advance in the storage unit,
The determination output means determines each of the biological states by referring to the probability table read by accessing the storage unit for a change in the driver's index obtained by analyzing the back body surface pulse wave. The driving support device according to any one of Items 1 to 11 . In a computer as a driving support device,
A computer program that analyzes a back body surface pulse wave, which is a driver's biological signal measured by a biological signal measuring device, and executes a biological state determination procedure for determining the biological state of the driver,
As the biological state determination procedure,
From the index obtained by the analysis of the back body surface pulse wave, it is determined a biological state that appears with a decrease in arousal level including fatigue, reduced attention, or drowsiness, and depending on the determination result, auditory, visual or It is configured to execute a determination output procedure for outputting a command for driving a stimulus applying unit that applies a stimulus including at least one of olfactory stimuli to the driver,
The timing at which the determination output procedure detects a biological state that is determined to be a warning level that appears in association with fatigue, a decrease in alertness or a decrease in arousal level including drowsiness in order to induce arousal to the driver, and this A command for driving the stimulus applying means is output at both timings of detecting related signs related to a decrease in arousal level appearing before the appearance of a warning level biological state, and the detection of the warning level biological state When the time difference between the timing and the detection timing of the related signs is determined to be within a predetermined time , the computer program is configured to instruct the stimulus applying unit to apply a stimulus that prompts the driver to take a break . The determination output procedure includes:
Zero-cross detection procedure for obtaining a time-series waveform of a frequency using a zero-cross point in a time-series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
A peak detection procedure for obtaining a time series waveform of a frequency using a peak point in a time series waveform of a back body surface pulse wave obtained from the biological signal measuring device;
A zero-cross slope time-series waveform calculation procedure for calculating a time-series waveform of the frequency by sliding the time-series waveform of the frequency obtained by the zero-cross detection procedure;
Performing a peak slope time series waveform calculation procedure to obtain a slope time series waveform of the frequency by sliding calculation of the time series waveform of the frequency obtained by the peak detection procedure;
Based on at least one index of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure and the slope time series waveform computation procedure obtained from the peak slope time series waveform computation procedure, various types appearing with a decrease in the arousal level The computer program according to claim 13, which determines a biological state. The determination output procedure includes:
Functional adjustment in the range of 0.001 to 0.0027 Hz by frequency analysis of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure or the slope time series waveform obtained from the peak slope time series waveform computation procedure. A signal, a fatigue acceptance signal in the range of 0.002 to 0.0052 Hz, and a frequency component corresponding to an activity adjustment signal of 0.004 to 0.007 Hz are extracted, and a distribution for obtaining a variation in the distribution rate of each of these frequency components The computer program according to claim 14 , wherein a rate calculation procedure is executed, and various biological states appearing with a decrease in the arousal level are determined from a distribution rate of each frequency component obtained by the distribution rate calculation procedure. In the determination output procedure, in the distribution ratio of each frequency component obtained using the slope time series waveform obtained from the zero cross slope time series waveform calculation procedure, the distribution ratio of the frequency component corresponding to the activity adjustment signal is the function adjustment. The computer program according to claim 15 , wherein when the distribution ratio of the frequency component corresponding to the signal is exceeded, it is determined as a related sign related to the decrease in the arousal level. The determination output procedure, the distribution ratio of each frequency component obtained by using a gradient time series waveform obtained from the zero-crossing slope time-series waveform calculation procedure, the gradient time series waveform obtained from the peak slope time-series waveform calculation procedure When the distribution ratio of each frequency component obtained by using the comparison is between the distribution ratios of the corresponding frequency components, the activity of the sympathetic nerve function and the activity of the parasympathetic nerve function The computer program according to claim 15 , wherein the computer program is determined to be a related symptom associated with a decrease in the degree of arousal that appears due to the divergence of. In the determination output procedure, a distribution rate of each frequency component obtained using the slope time series waveform obtained from the zero-cross slope time series waveform calculation procedure is a frequency at which the distribution rate of the function adjustment signal corresponds to the activity adjustment signal. When the state of exceeding the component distribution rate continues for a predetermined time or more, or when the state of distribution of the activity adjustment signal exceeding the frequency component distribution rate corresponding to the function adjustment signal continues for a predetermined time or more The computer program according to claim 15 , wherein the computer program is determined to be a related sign related to a decrease in the degree of arousal. The determination output procedure, when at least one of the amplitude of the slope time series waveform obtained from the zero cross slope time series waveform computation procedure and the slope time series waveform computation procedure obtained from the peak slope time series waveform computation procedure is a predetermined value or more, The procedure for determining whether or not the driver is in an uplift state is performed. After the determination that the driver is in an uplifted state, if it is determined that the uplifted state has ended, the instruction to prompt the driver to take a break is applied. 15. The computer program according to claim 14, which is output to the means. The determination output procedure includes: an input timing of a subjective fatigue identification unit including an input unit that is operated when the driver subjectively feels fatigue, a decrease in attention, or a decrease in arousal level including sleepiness; and the stimulus Executing a procedure for determining whether or not the time difference with the output timing of the stimulus by the applying means is within a predetermined range;
The computer program according to any one of claims 13 to 19 , wherein when the time difference is determined to be within a predetermined range, a command for prompting the driver to take a break is output to the stimulus applying unit. A generalized probability table of index fluctuations obtained by analysis of the back body surface pulse wave and biological states appearing corresponding to the index fluctuations is stored in advance in a storage unit of a driving support device that is a computer,
The determination output procedure determines each biological state in light of the driver's index fluctuation obtained by analyzing the back body surface pulse wave in light of the probability table read by accessing the storage unit. Item 21. The computer program according to any one of Items 13 to 20 .