JPH04250171A - Human machine system - Google Patents

Abstract

PURPOSE:To artificially humanize a machine by detecting human sensitivity and, according to this state, adapting and non-adapting the mechanical function of the machine to the human sensitivity. CONSTITUTION:A vehicle control part 1 detects a road environment and, according to this, controls the vehicle characteristics to the optimum running state. A false human reaction producing part 5 consisting of a human state detecting part 9 for estimating the physiological and psychological states of driver at real time, a stimulation control part for determining the indicated pattern of human reaction to driver, and a sense stimulation generating part 10 for generating a stimulation to a driver 11 on the basis of the instruction from the stimulation control part 6 detects the state of the driver 11 and produces a massage according to this state. The stimulation control part 6 is formed of an adaptable control part 7 and a non-adaptable control part 8, and the adaptable control part 7 conducts adaptable control so as to keep the driver in the optimum physiological and psychological states at real time at the time of generating the stimulation.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a human-machine system, and more particularly, to a human-machine system that is a human-machine system that affects the sensibilities of a driver in the relationship between an automobile and a driver.

0002

2. Description of the Related Art When humans continue to perform monotonous tasks for a long time, they experience a decrease in arousal (decreased attention), and after a while they return to an alert state. Although we can be vaguely aware of this rhythm of arousal, which repeats arousal and low arousal states, no method has been found to quantitatively measure it.

[0003] Conventionally, the circadian rhythm, which repeats sleep and activity in approximately 24-hour cycles, has been well known as a long-term human rhythm; , it does not show the rhythm of arousal over a time scale of about 10 minutes to several hours, from a state of tension to a state of decreased arousal. In general, human attention span is psychologically estimated at 20%.
It is said that it lasts about 30 minutes, but it is not clear what happens to humans when we look at it as a phenomenon over several hours.

Conventionally, as disclosed in Japanese Patent Application Laid-Open No. 55-164529, a device for detecting drowsiness while driving has been used to detect drowsiness based on the fact that the steering condition is disturbed or the response speed is delayed. 2. Description of the Related Art A drowsy driving detection device that gives a warning to a driver is known. In man-machine systems, it is important to design a system that allows the user to feel joy in mastering the machine, as exemplified by the relationship between a car and a driver, or a computer and a user.

[0005] In recent years, in the automobile industry, systems have been devised that utilize electronic control technology to adaptively control machine functions according to the driver's preferences, skills, and feelings. For example, U.S. Patent No. 4,829,434 discloses that the driver's state such as personal qualities, mood, preferences, physiology, psychology, etc.
It is disclosed that vehicle characteristics are changed by adaptive control based on the driving state of the vehicle and the environment. As a specific example, this U.S. patent detects the following distance, vehicle speed, and amount of rain, compares these with pre-stored data, and uses the judgment results to determine whether automatic transmission is suitable for drivers who prefer nimble driving. The company mentions a mechanism that changes the shift pattern to power mode and switches to economy mode for drivers who prefer comfortable driving.

[0006]

[Problem to be Solved by the Invention] In recent years, there has been an increasing need for technology to quantitatively understand the temporal characteristics of such decline in arousal to prevent it. Test drivers are required to perform harsh monotonous driving. The state of reduced arousal that affects drivers under these conditions (even if they are not dozing off, their judgment is impaired) is thought to be the cause of accidents caused by drivers.

[0007] The drowsy driving detection device disclosed in the above-mentioned Japanese Patent Application Laid-open No. 55-164529 detects a decrease in alertness and issues an alarm only when the driver is in a dozing state or in a state immediately before it. Therefore, it may be too late. Furthermore, in a vehicle based on the above-mentioned conventional electronic control technology, when adaptive control matching the driver's sensibilities is achieved, the following problems occur from the driver's perspective.

(1) The intellectual ability of the machine is superior to the driver's ability, and the control is too adapted to the driver's characteristics, causing the machine to perform tasks that should normally be judged and processed by the driver, causing the driver to The driver loses the desire to independently control the vehicle, and as a result, the driver's driving skills deteriorate, making accidents more likely. In other words, in a vehicle that performs adaptive control, the adaptive control changes the vehicle characteristics to the extent that it affects the driver's driving operation. Forced to drive a vehicle.

(2) Despite the high level of intellectual adaptability of the vehicle, the driver feels cold and boring, and the vehicle seems inorganic to the driver, causing him to lose the joy of driving and his affinity for the vehicle. The present invention has been made in view of this point, and its purpose is to provide a machine that is enhanced to simulate humanization by adding human emotional functions at the same time to a machine that only emphasizes intelligence. On the other hand, the aim is to improve the driver's motivation to drive by allowing the driver to enjoy interacting with this pseudo-humanized machine.

0010

[Means for Solving the Problems] The present invention has been made for the purpose of solving the above-mentioned problems, and has the following configuration as a means for solving the above-mentioned problems. That is, the human-machine system is equipped with an environment detection means for detecting the external environment, and the human-machine system has a function that allows humans to feel that the human-machine system is pseudo-human in accordance with human sensibilities. and control means for controlling.

Preferably, the control means includes a first detection means for detecting the physiological or psychological state of the human being;
Alternatively, the first control means controls the functions of the human-machine system so that the psychological state becomes comfortable. Preferably, the control means includes a first detection means for detecting the human physiology or psychological state, and a second control means for controlling the function of the human machine system so that the physiological or psychological state becomes uncomfortable. and means.

[0012] More preferably, the control means includes a first detection means for detecting the physiological or psychological state of the human being, and controls the functions of the human-machine system so that the physiological or psychological state is comfortable or uncomfortable. and third control means. Also, preferably, the control means:
an alertness level detection means for detecting the alertness level of a human; and a fourth apparatus for controlling the functions of the human machine system according to the alertness level.
control means.

Preferably, the arousal level detection means includes means for detecting brain waves, means for calculating the power amount and power data of the detected brain waves, and means for presenting a stimulus and detecting a reaction time to the stimulus. The apparatus includes a means for calculating an arousal degree estimation parameter by analyzing the correlation between the amount of power of an electroencephalogram and a reaction time, and a means for calculating an arousal degree from the arousal degree estimation parameter and the power data.

Preferably, the wakefulness detection means includes means for detecting brain waves, means for performing frequency analysis on time-series data of the detected brain waves, and calculating the wakefulness rhythm cycle by detecting a level peak on the frequency. and means to do so. Preferably, the arousal level detection means includes means for presenting a stimulus;
The apparatus includes means for detecting a reaction time to a presented stimulus, means for subjecting time-series data of the reaction time to frequency analysis, and means for detecting a level peak on the frequency to calculate an arousal rhythm cycle.

[0015] More preferably, the arousal level detection means includes means for detecting blink frequency, means for presenting a stimulus and detecting reaction time to the stimulus, and analyzing the correlation between blink frequency and reaction time. The apparatus includes means for calculating a degree of wakefulness estimation parameter, and means for calculating a degree of wakefulness from the degree of wakefulness estimation parameter and the blink frequency. Also, preferably,
The alertness detection means includes a means for detecting the heart rate, a means for presenting a stimulus and detecting a reaction time to the stimulus, and a means for calculating an alertness estimation parameter by analyzing the correlation between the heart rate and the reaction time. and means for calculating the alertness level from the alertness level estimation parameter and the heart rate.

[0016] Preferably, the control means includes an alertness level detecting unit for detecting the level of alertness of the human being, a measuring unit for measuring activity information of the human body, and a control unit for applying an external enemy stimulus to the human body according to the activity information. The present invention includes a stimulus presentation means for emitting a stimulus, and an arousal maintaining means for controlling the stimulus presentation means according to the degree of arousal to maintain the human's wakefulness state.

[0017]

[Operation] With the above configuration, human sensibilities are detected, and the mechanical functions of the human-machine system are adapted or non-adapted to the human sensibilities in response to the detected state.

[0018]

[Example] [Summary description of the invention] <Basic functions of the invention> The human machine system of the present invention differs from conventional adaptive systems in that: (1) the control system It does not directly affect the driver's driving behavior. (2) In order to promote compatibility between the driver and the vehicle, efforts are being made to make the vehicle pseudo-human. This is based on the guidelines.

Therefore, the human-machine system of the present invention has the following basic functions as a human-machine system. (1) Adaptive function A function that detects the driver's sensibility and adapts the functions of the machine, such as physical elements, control characteristics, and environment, to the driver's sensibilities so that the driver is comfortable in that situation.

[0020] Sensitivity detection here means creating an index that can quantitatively measure the human condition from human physiological quantities, psychological quantities, and behavior, and estimating the human condition using the results. means. In addition, the measurement of comfort is based on human physiology,
This is done by creating an evaluation index that matches the psychological state from physiological quantities and behavioral quantities (quantities that represent behavior), and estimating the human physiology and psychological state based on this. In order to achieve comfort, a method is used in which a stimulus is presented to a human, the effect of the stimulus is evaluated using an index, and the amount of stimulation is controlled so that the human is in the most comfortable state. The idea of so-called alertness biofeedback is included in this realization method. (2) Non-adaptive function A function that detects the driver's sensibility and makes the functions of the machine such as physical elements, control characteristics, environment, etc. non-adaptive to the driver's sensibilities so that the driver becomes uncomfortable in that state. (3) Detect the sensibility of the pseudo-humanoid function driver, and use physical elements and
A function that gives the driver the illusion that the machine has will and emotions by adapting or non-adapting machine functions such as control characteristics and environment to the driver's sensibilities. <Controlled objects> The controlled objects of the human machine system of the present invention include human sensibilities such as vision, hearing, touch, vestibular sensations such as acceleration and gravity, moods such as alertness, excitement, and tension, as well as pleasure. Emotions such as anger, sadness, and joy, fatigue caused by driving, attachment to things, boredom, and higher-level sensibilities such as love, hatred, values, ethics, and aesthetics, and the driver's driving skills and habits. This behavior is a characteristic of .

[0021] The machine functions to be controlled do not control functions that directly affect driving, which is the work of the driver. In other words, the functions to be controlled are indirect machine functions with respect to the target work, and do not directly affect the work. For example, changing the steering characteristics of a vehicle based on the driver's skill can be said to be direct control since the change greatly affects the driver's driving. However, even if the steering characteristic is changed and the driver can sense the change, if the change does not affect the driver's driving, it is indirect control.

[0022] Machine functions that indirectly affect
There are environmental stimuli such as sound, vibration, heat from air conditioning, air flow, and illuminance that directly affect humans sensory or psychologically. <Basic configuration of the invention> The human machine system of the present invention (hereinafter referred to as human machine system)
Mimetic Machine (abbreviated as HMM)
) is designed to give the driver the feeling that the machine's reactions are the result of human will. Therefore, in order to express the interaction between humans and machines, we will realize the communication pattern that occurs between humans and machines between humans and machines.

[0023] A relationship similar to the relationship between close human beings who exchange messages with each other, that is, a psychological phenomenon that humans derive the most psychological satisfaction through interaction with close people, is created in the man-machine system. Realizing this will solve the above-mentioned problems. Figure 1 shows interaction patterns obtained by observing interactions between close people. In the figure, an attentive response means always paying maximum attention to the other person's condition and reaction, and a personality response means responding with one's own will and assertiveness while being attentive. means. Furthermore, an emotional reaction is a self-centered reaction that sometimes prioritizes emotions, and a submissive reaction is a reaction in which a person tries to obey the person they respect as much as possible.

Based on these observations, the HMM of the present invention allows the machine to respond to the driver's actions by responding to the driver's actions in a manner that makes them feel attentive, and at times also responds with emotional and personality responses. It gives a sense of uniqueness. FIG. 2 is a block diagram showing the configuration of the HMM of the present invention. The HMM shown in the figure includes a vehicle control unit 1 that detects the road environment and controls the vehicle characteristics to an optimal driving state accordingly, and a driver 11 that detects the state of the driver and generates a message according to the state. Pseudo human reaction generation unit 5
Consisting of

The vehicle control unit 1 realizes optimal vehicle behavior according to the driver's operations and the road environment, and controls the vehicle to adapt to the external environment, such as ABS and four-wheel drive (4WS). ) corresponds to this. The pseudo-human reaction generation unit 5 controls the vehicle environment in order to return human-like reactions to the driver. Here, in order to influence the driver's physiological and psychological state through the vehicle's reaction, to control the physiological state to the optimum state, and to make the driver feel psychologically satisfied, we only transmit messages to the driver regarding the driver's state. Let's do it. This pseudo-human reaction generation unit 5 includes a human condition detection unit 9 that estimates the driver's physiological and psychological state in real time, a stimulation control unit 6 that determines a presentation pattern of human reactions to the driver, and a command from the stimulation control unit 6. The sensory stimulation generating section 10 generates stimulation for the driver 11 based on the following.

The stimulus control section 6 consists of an adaptive control section 7 and a non-adaptive control section 8. Upon receiving a signal from the vehicle behavior section 3, the adaptive control section 7 controls the driver in real time to optimize the physiological and psychological conditions when generating the stimulus. A command is issued to the sensory stimulation generating section 10 for adaptive control to maintain the state. In addition, the non-adaptive control section 8 irregularly interrupts the command to the sensory stimulus generation section 10 from the adaptive control section 7 based on the accumulation of long-term vehicle behavior and motion history information. Then, a command is issued to the sensory stimulus generation unit 10 to select a stimulus based on a normatively defined stimulus generation rule and generate a stimulus that is non-adaptive for the driver. However, the stimulus does not directly control the steering characteristics of the vehicle, but is a stimulus that is not directly involved in vehicle running, such as sound, vibration, scent, and air-conditioned environment inside the vehicle interior.

The operation of the adaptive control section 7 expresses an attentive response to the driver 11, and the adaptive control section 7 acts as a master to the non-adaptive control section 8, thereby expressing a submissive response. The driver 11 of the non-adaptive control unit 8
Express personality reactions and emotional reactions by adjusting the rules of reaction. The reason why the non-adaptive control unit 8 accumulates long-term vehicle behavior and movement history information is because in human interactions, memories of past events due to long-term interactions are generated.
This is to express that it affects the psychological relationship in the current interaction with the other person. For example, when driving at the upper limit of performance for a long period of time, such as when driving on a highway, the vehicle environment may be controlled to give the driver a message urging him to stop driving.

In the HMM of the present invention, the pseudo-human reaction generating section 5 does not directly influence the vehicle control section 1, and the vehicle control section 1 concentrates on controlling the vehicle. Control of vehicle behavior is not directly controlled according to the driver's physiological or psychological state, but is actively left to the driver's will. This is based on the idea that the joy of driving a vehicle lies in the driver's own independent act of controlling the vehicle. In other words, this HMM works on the driver's physiology and psychology using only environmental stimuli that directly affect humans, teaches the driver driving operations that match the vehicle behavior, and at the same time creates an environment suitable for the teaching.

Next, preferred embodiments of the present invention will be described in detail. <First Embodiment> FIG. 3 shows the H according to the first embodiment of the present invention.
1 is a block diagram showing the overall configuration of an MM. In the figure, the vehicle control unit 2 controls the vehicle behavior unit 3 according to the operation of the driver 11 and the road environment from the environment 20 detected by the detection unit 4.
The system controls the vehicle to achieve optimal vehicle behavior. From this vehicle behavior unit 3, the central control unit (CPU) 15 of the stimulation control unit 6 receives a vehicle body strain R as the vehicle behavior.
is output.

The human condition detection section 9 estimates the driver's physiological and psychological conditions in real time.
A smoothing circuit 21 smoothes the pulse wave from the fingertip volume pulse wave detector 13 provided to detect the human condition, and based on the result, calculations are performed to obtain the human condition quantity described later. It consists of a calculation section 22. The stimulus control unit 6 consists of an adaptive control unit 7, a non-adaptive control unit 8, and a central control unit (CPU) 15 that controls them.The adaptive control unit 7 controls the driver in real time to optimal physiological and psychological conditions during stimulus generation. A command is issued to the sensory stimulation vehicle control section 12 for adaptive control to maintain the state. In addition, the non-adaptive control section 8 irregularly interrupts the command of the adaptive control section 7 to the sensory stimulation vehicle control section 12 based on the accumulation of long-term vehicle behavior and movement history information.

The sensory stimulation vehicle control section 12 receives a command from the stimulation control section 6 and generates stimulation for the driver 11 based on the presentation pattern of the human reaction to the driver determined by the stimulation control section 6. FIG. 4 is a block diagram showing a detailed configuration around the CPU 15 that constitutes the stimulation control section 6. As shown in FIG. In the same figure, the CPU 15 sends A from the vehicle behavior section 3 to
The vehicle body distortion R is received via the /D converter 18, and at the same time, the human state quantity H is received from the human state detector 9 via the A/D converter 19. Based on these inputs, the CPU 15 generates sensory stimulus generation control parameters I, which will be described later.

FIG. 5 is a diagram for explaining a method of calculating vehicle body strain R in the vehicle behavior section 3. In the figure, the strain gauge 32 installed at the bottom of the vehicle body 31 is
The strain Sx in the direction and the strain Sy in the y direction are detected. The strain calculation circuit 33 calculates the root mean square of Sx and Sy from the results of strain detection by the strain gauge 32, and outputs the value as the vehicle body strain R to the stimulation control section 6.

FIG. 6 is a diagram showing a method for detecting the human condition of the driver in the human condition detecting section 9. As shown in FIG. Here, by processing the pulse wave (FIG. 6(a)) as a voltage waveform obtained by the fingertip volume pulse wave detector 13, an index H of the degree of irritability or startling of the driver 11 is calculated. (b) in Figure 6
) shows a waveform obtained by smoothing the pulse wave shown in FIG. The part of the triangle that is

For this smoothed waveform, the following equation (1
), the human state quantity H is obtained. As a result of the calculation, the waveform shown in Figure 6(c) is obtained, and the larger the value of H, the more irritated or upset the person is, and conversely, the smaller the value, the more calm the person is. show.

[0035]

[Math 1]

FIG. 7 is a map showing the relationship between the human state quantity H and the sensory stimulation generation control parameter I.
a) shows the relationship between the human state quantity H and the sensory stimulus generation control parameter I calculated by the adaptive control unit 7;
It has a characteristic that the larger the value of the state quantity H, the smaller the value of the parameter I. Further, (b) in FIG. 7 shows the relationship between the human state quantity H and the sensory stimulus generation control parameter I' calculated by the non-adaptive control unit 8, which is opposite to (a) in FIG. It has a characteristic that the larger the value of the state quantity H, the larger the value of the parameter I'.

Next, a procedure for calculating the sensory stimulus generation control parameters I and I' in this embodiment will be explained. FIG. 8 is a flowchart showing the main routine for calculating the sensory stimulation generation control parameters I, I'. is input, and in step S1b, the corresponding sensory stimulation generation control parameter I is calculated with reference to the map shown in FIG. 7(a). In step S2, it is determined whether or not an interruption, which will be described later, has occurred. If there is no interruption, the process proceeds to step S3, where I is output to the sensory stimulation vehicle control section 12 to enable comfortable driving.

However, if it is determined in step S2 that an interruption has occurred, a sensory stimulation generation control parameter I' is calculated in step S5. The calculation of this I' is shown in Figure 7.
This is done with reference to the map shown in (b). After replacing the parameter I with I' in step S6, the process proceeds to step S3 in which I is output. In step S4, a waiting time TW is provided to see the effect of outputting the parameter I, and after TW has elapsed, the process returns to step S1.

The flowchart shown in FIG. 9 shows the interrupt generation routine in step S2 of FIG. In the figure, the CPU 15 controls the vehicle behavior section 3 in step S10.
The vehicle body distortion R is inputted from , and it is determined in the next step S11 whether or not the vehicle is traveling on a rough road. Here, Figure 10
As shown in , when the time during which the value of the vehicle body strain R calculated by the vehicle behavior section 3 is equal to or greater than the predetermined value Rr exceeds the predetermined time tr seconds, it is determined that the vehicle is traveling on a rough road.

If the determination at step S11 is YES, the process advances to step S12, where it is determined whether the parameter I' has been outputted consecutively n times or more. If the result of this determination is YES, it is assumed that the non-adaptive control has continued for a certain period of time or more, and it is determined that the driver is unable to take a rest for some reason, and the process proceeds to step S13, canceling the rough road judgment made in step S11. perform adaptive control. However, if the determination in step S12 is NO, it means that adaptive control is being performed even though the vehicle has been driving on rough roads that put a strain on the vehicle body for a certain period of time, so the process proceeds to step S14. generates an interrupt. Therefore, the judgment in step S2 of FIG. 8 described above is YES, and non-adaptive control is performed, so that the control is performed in a direction that increases the level of irritation and shock, and as a result, the driver 11 loses his motivation to drive and takes a break. You will be encouraged to do so.

FIG. 11 is a block diagram showing the internal structure of the sensory stimulation vehicle control section 12. As shown in FIG. In the figure, a controller 34 receives a sensory stimulation generation control parameter I from a stimulation control section 6, and calculates a filter control signal F for selecting a characteristic of a variable filter 26 in accordance with the value. The white noise W from the white noise generator 25 is always input to the variable filter 26, and outputs a signal W' filtered with filter characteristics according to the value of F. This signal W' is converted into a digital signal by the A/D converter 30, and then sent to the controller 34 as a signal W''.
is input.

The variable filter 26 has a 1/f characteristic as shown in FIG. 12, and as the value of the parameter I increases, the signal W' changes from white noise to 1/fn noise (
n is a real number of 1 or more, preferably a value between 1 and 2). The controller 34 further calculates an excitation fluctuation control signal V, which will be described later, and the excitation control section 29
is this excitation fluctuation control signal V and engine rotation signal RP
A signal is sent to the vibrator 28 provided at the bottom of the seat 27 on which the driver 11 sits in accordance with M. As a result, sheet 2
7 is vibrated at a predetermined frequency which will be described later.

FIG. 13 is a map showing the frequency characteristics of vibration of the seat 27 by the vibrator 28, and FIG. 13(a) shows the relationship between the engine rotation signal RPM and the basic vibration frequency ω. In addition, FIG. 5B shows the relationship between the excitation fluctuation control signal V and the excitation frequency fluctuation Δω. The seat 27 is vibrated at a frequency that changes according to the sum of ω and Δω, which respectively correspond to the vibration fluctuation control signal V and the engine rotation signal RPM output from the vibration control unit 29 as described above. Ru.

In the end, the sensory stimulation vehicle control section 12 drives the vibrator 28 that supports the driver's seat in accordance with the sensory stimulation generation control parameter I from the stimulation control section 6, so that vibration stimulation is not applied to the driver. You can control how irritated or upset you are. FIG. 14 is a flowchart showing a procedure for calculating the excitation fluctuation control signal V and the filter control signal F in the controller 34. In step S15 in the same figure, the controller 15 determines a filter characteristic from the 1/f characteristic map shown in FIG. A filter control signal F is calculated and the signal F is output in step S16.

The controller 34 calculates an excitation fluctuation control signal V based on the output signal W'' from the A/D converter 30 described above in step S17, and sends it to the excitation control unit 29 in step S18. Outputs signal V. In step S19,
Time T since the filter control signal F was calculated in the previous process
Determine whether W has elapsed. If it is determined that the time TW has not elapsed, the process returns to step S17 to calculate the excitation fluctuation control signal V, but if it is confirmed that the predetermined time has elapsed, the process returns to step S15 and calculates the excitation fluctuation control signal V. A filter control signal F is calculated.

FIG. 15 shows the relationship between the human condition and the vibration frequency of the seat in the HMM according to the embodiment. In the same figure, when the human condition detection unit 9 of the HMM performs adaptive control based on the driver's degree of irritation or surprise, it uses a 1/fn type (n is a real number of 1 or more, between 1 and 2). When the driver is in a quiet state, a white noise-type fluctuation is applied. In addition, in non-adaptive control, fluctuations in the seat vibration frequency are given that are opposite to those in adaptive control.

As explained above, according to this embodiment,
The driver's level of irritation and nervousness is detected using fingertip volume pulse waves, and vehicle behavior is detected by detecting vehicle body distortion.
Seat fluctuation vibration according to these values is used as a stimulus to perform adaptive control so that the driver can drive comfortably. This has the effect of being able to use this as a stimulus to perform non-adaptive control and encourage the driver to rest. <Second Example> A second example according to the present invention will be described. Note that the overall configuration of the HMM according to this embodiment is the same as the HMM according to the first embodiment shown in FIG. 3, and therefore the description thereof will be omitted.

FIG. 16 is a block diagram showing the detailed configuration around the CPU 15 that constitutes the stimulation control section 6 of the HMM. In the figure, the CPU 15 receives a vehicle speed V from the vehicle behavior section 3 via the A/D conversion section 18, and at the same time receives a human state quantity H from the human state detection section 9 via the A/D conversion section 19. The CPU 15 generates the sensory stimulation generation control parameter I based on these inputs.

In this embodiment, the method for detecting the human state of the driver in the human state detection section 9 and the human state quantity H
The relationship between and the sensory stimulus generation control parameter I is the same as the detection method in the first embodiment (FIG. 6) and the relationship between the state quantity H and the parameters I and I' (FIG. 7). Next, a procedure for calculating the sensory stimulus generation control parameters I and I' in this embodiment will be explained.

Since the main routine for parameter calculation in this embodiment is the same as the parameter calculation procedure in the first embodiment shown in FIG. 8, only the interrupt generation routine will be described here. FIG. 17 shows an interrupt generation routine according to this embodiment, in which, in step S21,
The CPU 15 determines whether the vehicle is stuck in a traffic jam. Here, as shown in FIG. 18, the vehicle behavior unit 3 determines that there is a traffic jam if the vehicle speed V remains below a certain constant vehicle speed VJ for a predetermined time tJ seconds or more.

If the determination at step S21 is YES, the process proceeds to step S22, where it is determined whether the parameter I' has been outputted consecutively n times or more. If the judgment result is YES, it is assumed that the non-adaptive control has continued for a certain period of time or more, and it is judged that the driver is unable to take a rest for some reason, and the process proceeds to step S23, canceling the traffic congestion judgment at step S21. Perform adaptive control. However, if the determination in step S22 is NO, it means that adaptive control is being performed even though the traffic jam has continued for a certain period of time, and therefore an interrupt is generated in step S24.

As a result of this process, the determination at step S2 in FIG.
In case 1, your motivation to drive will be diminished and you will be asked to take a break. FIG. 19 is a block diagram showing the internal configuration of the sensory stimulation vehicle control section 12 of the HMM of this embodiment. In the figure, a controller 34 receives a sensory stimulation generation control parameter I from a stimulation control section 6, and calculates a filter control signal F for selecting a characteristic of a variable filter 26 in accordance with the value. The variable filter 26 includes a white noise generator 2.
The white noise W from 5 is always input, and the signal W is filtered with filter characteristics according to the value of F.
' is output. This signal W' is converted into a digital signal by the A/D converter 30 and then inputted to the controller 34 as a signal W''.

The variable filter 26 is similar to the first embodiment.
It has a 1/f characteristic as shown in Fig. 12, and as the value of the parameter I increases, the signal W' changes from white noise to 1/fn noise (n is a real number of 1 or more, and a value between 1 and 2). ). The controller 34 calculates a fuel injection control signal E based on the signal W'' as described later, and sends the calculation result to the EGI control unit 35.
performs fuel injection according to control signals from the EGI control unit 35.

FIG. 20 shows the flow of the fuel injection control signal E from the EGI control unit 35 to the fuel injection device 36.
This is a map showing the relationship between the fuel control signal and the fuel control signal sent to the engine. 4A shows the relationship between the fuel injection control signal E and the fuel injection amount, and FIG. 2B shows the relationship between the fuel injection control signal E and the injection interval. The safe tolerance range in the figure was established to prevent fluctuations in the amount of fuel injection and injection interval from changing excessively from the appropriate values. Do not take large values.

The sensory stimulation vehicle control section 12 of the HMM controls the engine output by controlling the fuel injection amount or the fluctuation of the injection interval according to the sensory stimulation generation control parameter I from the stimulation control section 6. Since the driver is given the stimulus of not only engine output but also engine sound and vibration, it is possible to control the degree of irritation and startling.

FIG. 21 is a flowchart showing the procedure for calculating the fuel injection control signal E and filter control signal F in the controller 34. In step S25 in the same figure, the controller 15 uses the sensory stimulation generation control parameter I from the stimulation control section 6 as shown in FIG.
A filter characteristic is determined from the /f characteristic map, a filter control signal F corresponding to the characteristic is calculated, and step S2
6 outputs the signal F.

In step S27, the controller 34 calculates a fuel injection control signal E based on the output signal W'' from the A/D converter 30, and then in step S28.
A signal E is output to the vibration control section 29. Step S29
Now, it is determined whether the time TW has elapsed since the filter control signal F was calculated in the previous process. Here time TW
If it is determined that the period has not elapsed, step S2
The process returns to step S2 to calculate the fuel injection control signal E, but if it is confirmed that the predetermined time has elapsed, the process returns to step S2.
Returning to step 5, the filter control signal F is calculated again.

FIG. 22 shows the relationship between human condition and fuel injection fluctuation in the HMM according to this embodiment. In the same figure, when the human condition detection unit 9 of the HMM performs adaptive control based on the driver's degree of irritation or surprise, it uses a 1/fn type (n is a real number of 1 or more, between 1 and 2). When the driver is at rest, it provides a white noise-type fluctuation. In addition, in non-adaptive control, fluctuations in fuel injection are given that are opposite to those in adaptive control.

As explained above, according to this embodiment,
The driver's level of irritation and nervousness is detected using fingertip volume pulse waves, and vehicle behavior is determined based on vehicle speed, and engine output and engine control are determined by fluctuations in fuel injection according to these values. The system uses sounds and vibrations as stimuli to perform adaptive control to help the driver drive comfortably, and when it is determined that traffic congestion has continued for a certain period of time, it uses a stimulus based on predetermined fluctuations in fuel injection. This has the effect of performing non-adaptive control and encouraging the driver to rest. <Third Embodiment> A third embodiment of the present invention will be described below.

In this embodiment, the driver's level of alertness is detected and a stimulus corresponding to the level of alertness is presented to the driver to maintain the driver in an optimal alertness state. FIG. 23 is a schematic block diagram showing the overall configuration of an HMM according to the third embodiment. In the same figure, the vehicle control unit 2 controls the driver 1
1 and the road environment from the environment 20 detected by the detection unit 4, the vehicle behavior unit 3 is controlled to realize optimal vehicle behavior.

The human state detection section 9 detects the wakefulness state of the driver 11 based on the brain waves from the driver 11. The adaptive control unit 7 issues a command to the sensory stimulation vehicle control unit 12 in order to perform adaptive control to maintain the driver in an optimal physiological and psychological state in real time during stimulation generation. Further, the sensory stimulation vehicle control unit 12 receives a command from the adaptive control unit 7 and generates stimulation for the driver 11 based on a presentation pattern of human reactions to the driver.

FIG. 24 is a detailed block diagram of the HMM of this embodiment. Here, a method is used to measure the driver's brain waves by gluing the electrode 2a to the head of the driver 11 and measuring the potential difference. . The brain waves of the driver 11 are guided to the physiological measurement amplifier 43 and subjected to signal amplification, and the A/D
After being converted into a digital signal by the converter 44, the CPUb
Sent to 42. The CPUb42 calculates an estimated value X of the alertness level, which will be described later, and delivers the estimated value X at that moment to the CPUa41 in accordance with instructions from the CPUa41 for controlling the entire system and controlling the alertness level.

[0063] The CPUa 41 constantly monitors the estimated level of wakefulness X, and at the same time measures the engine sound of the engine 45 using a microphone 46 provided in the engine room 54. The measured engine sound is converted into a digital signal by the A/D converter 47, and then taken into the CPUa41. When the CPUa 41 determines that the driver is in a low alertness state based on the estimated alertness value Sends a gain control signal to control its amplification degree. And the amplifier 50 is
Engine sound amplified according to a control signal from the CPU 1a is presented to the driver 11 as a stimulus via the speaker 53. The amplification degree of this amplifier 50 is as described below.
The CPU a 41 makes the determination based on the monitoring result of the engine sound output level from the speaker 53 picked up by the ear microphone 52 installed near the ear of the driver 11 .

On the other hand, when the CPUa 41 determines that the driver 11 is in an overstressed state, the scent presentation device 48
to present a calming scent to the driver. Incidentally, when the driver 11 has been operating the system for a long time and has accumulated physical fatigue, he or she can stop the system by operating an end switch 49.

Next, the awake state sustaining processing procedure in the HMM of this embodiment will be explained. FIG. 25 is a flowchart illustrating the procedure of optimal wakefulness sustaining processing according to the embodiment. In the figure, in step S31, the following values, parameters, etc. are set as initial values. (1) Optimal arousal level estimation value Rbest (2) Tension level limit value R1 (3) Arousal reduction limit value R2 (4) Permissible control arousal level estimate value range Rerr1 (for tension level) (5) Permissible control arousal level estimate value range Rerr2 (for low arousal level) (6) Engine sound capture time T1 (7) Speaker generated volume E0 (8) Safe limit engine volume Esafe (9) Engine on reproduction sound rate Δ0 (10) Safety limit counter Nsafe-lim (11)
Engine sound playback time at safety limit Tsafe (12) Scent presentation limit number of times Ns-lim (13) Personal alertness estimation parameters a1, a2, a3, a4, a5 (14) HMM drive time counter Nl (initial value set to 1) (15) HMM drive time upper limit counter Nlset Here, the relationship between the wakefulness level and the wakefulness estimate value will be explained.

FIG. 26 illustrates the relationship between the alertness level and the estimated alertness level, and both are in a monotonically increasing relationship. When the estimated arousal level is Rbest, it is assumed that the arousal state is optimal, and if we define a range of Rerr1 on the overstress side and Rerr2 on the arousal side with Rbest as the center (shaded area in the figure), the estimated arousal level is When it is within this range, it is considered to be a normal wakeful state.

Further, when the estimated arousal level is greater than R1, it is assumed to be a hypertension state, and when it is smaller than R2, it is assumed to be a low arousal state. In the next step S32, the estimated arousal level X1 is taken in, and in the following step S33, it is determined whether or not X1 is greater than R1, that is, whether or not the person is in an overstressed state. If the determination here is YES, it is determined that the vehicle is in an overstressed state, and the process proceeds to step S36, where an overstress countermeasure process, which will be described later, is executed.

However, if the determination in step S33 is NO, it is determined in step S34 whether X1 is smaller than R2 or not.
In other words, it is determined whether or not the person is in a low arousal state. If it is determined that the state is not in a low arousal state, the process returns to step S32 and the estimated arousal level X1 is taken in. On the other hand, if it is determined in step S34 that the patient is in a low arousal state, the process advances to step S35, and a low arousal countermeasure process, which will be described later, is executed.

In step S37, the value of the HMM driving time counter Nl is equal to the HMM driving time upper limit counter Nlset.
Determine whether it is greater than the value of . The verdict here is NO.
If so, it means that the optimal alertness level sustaining process has not been executed the set number of times, and the counter Nl is set in step S38.
After incrementing the value by 1, the process returns to step S32. However, the value of counter Nl is
, the determination in step S37 is YES.
In the next step S39, pressing of the end switch 49 by the driver is detected. If this switch is pressed, it is assumed that the driver intends to stop driving the HMM, and the process ends.

Next, a method for estimating the degree of alertness as a numerical value from brain waves in the HMM of this embodiment will be explained. FIG. 27 is a block diagram of the alertness level estimating section within the CPUb42 that constitutes the HMM of this embodiment. Here, by absolutely quantifying the degree of human arousal and applying stimuli such as vibrations, sounds, and scents to the driver accordingly, we maintain the driver's arousal state at a certain level in terms of biofeedback. , to prevent falling asleep or becoming overly excited.

In FIG. 27, the brain wave detection section 51a is comprised of a portion that detects brain waves from the electrodes 2 attached to the head of the driver 11 who is the subject, and amplifies the brain waves with an amplifier (not shown). The brain wave processing unit 52a processes the brain wave signal from the brain wave detection unit 51a and processes the signal into a physical quantity that makes it easy to estimate arousal. That is, the brain wave signal amplified by the brain wave detection unit 51a is taken in via an A/D converter (described later), and subjected to digital filter processing to obtain the δ wave (1-3H) of the brain wave.
Z ), θ wave (3-6Hz), alpha wave (8-13Hz
), β waves (13Hz - 30Hz) and processed into physical quantities necessary for arousal evaluation.

[0072] The arousal degree estimating unit 53a calculates an estimated value closely related to arousal based on the arousal degree estimation parameter and brain wave band data, which will be described later. Further, the arousal level determining unit 54a determines how much stimulation to present based on the arousal level estimated by the arousal level estimating unit 53a. Then, the stimulus presentation section 50
a presents or stops appropriate stimulation to the driver in response to instructions from the alertness level determination unit 54a.

The alertness level estimating section of this embodiment estimates the alertness level based on brain waves and reaction time, and uses this information to provide optimal stimulation to the driver in a biofeedback manner. Therefore, with reference to FIG. 28, the concept of biofeedback applied to this example will be described. Figure 28 is a diagram often used in psychology and ergonomics, where the vertical axis represents work efficiency;
Or, it represents processing ability, and if limited to driving, it corresponds to the intended steering operation and judgment ability. Also,
The horizontal axis is the amount of stimulation, which includes visual stimulation, auditory stimulation, tactile stimulation,
One can think of olfactory stimulation, etc. When driving, these can be compared to the flow of scenery in front of you, the noise and vibrations inside the car, and the smell.

The hatched part B in FIG. 28 indicates that a driver who increases the amount of stimulation beyond a certain level will no longer be able to perform accurate operations, and will eventually fall into a panic state (hyperarousal). ing. As an example, when driving at high speed and increasing the speed of the vehicle, you may eventually become unable to keep up with changes in the road environment and become unable to operate the steering wheel, or you may be unable to avoid a spin. be able to.

[0075] The hatched area A in the same figure indicates that the subject falls into a monotonous state due to the lack of stimulation in spite of the demand for attention, which causes a decrease in arousal. An example of this is a state of decreased alertness during monotonous driving such as on a highway. Incidentally, the goal of conventional drowsiness warning devices is to provide a warning to the driver when this region is reached.

As can be seen from the above explanation, humans cannot respond to information or environmental changes that exceed a certain amount, and similarly, arousal decreases when the amount of stimulation is less than a certain amount. For this reason, the most desirable safety device is to clarify the optimal amount of stimulation for each person, although it differs from person to person, and then adjust and provide effective stimulation near that amount according to the physiological state. That's true. This situation is shown in the center of FIG. 28 as the optimal stimulation amount.

In this embodiment, it is possible in principle to maintain the optimal wakefulness state shown in FIG. 28 by detecting the human wakefulness state and applying stimulation accordingly. First, a description will be given of an arousal degree estimation parameter determination device that determines parameters for calculating the arousal degree in the arousal degree estimating section 53a. The alertness estimation parameter determination device simultaneously measures reaction time and brain waves, which are highly correlated with the degree of alertness, and determines the alertness estimation parameters by performing multiple regression analysis based on the obtained brain wave data and reaction time data. It is something to do.

FIG. 29 is a schematic block diagram of the arousal level estimation parameter determination device. The figure shows a situation in which this device simultaneously measures a person's reaction time and brain waves over a long period of time, and it is possible to obtain estimated parameters by analyzing the correlation between the obtained reaction time and brain waves. can. Figure 2
9, the electroencephalogram processing unit 70 processes the subject (driver) 1
The reaction time measurement section 71 measures the time it takes for the test subject 11 to react to the given stimulus. Based on these measurement results, the arousal level estimation parameter determination unit 72 estimates the subject's own arousal level.

FIG. 30 is a block diagram showing the configuration of a reaction time measurement section 71 that measures reaction time (selected reaction time). In the same figure, the reaction stimulus presentation unit 30A generates a stimulus for reaction measurement to the subject 11 under the control of the selection reaction time measuring unit 35A, and the reaction from the subject 11 is transmitted to the selected reaction time measurement unit 31A via the reaction input unit 31A. It is input to the time measuring section 35A.

The selection reaction time measurement section 35A includes a main control section 36A consisting of a CPU 36a that controls the entire device, a storage section that stores various data, and a main control section 36A that sets the time at which a reaction stimulus is presented and measures reaction time. Timer 37A for
Consisted of. The data storage section includes a stimulus presentation time data storage section 36a for storing data regarding the time at which a response stimulus is presented, a stimulus pattern data storage section 36c for storing a response stimulus pattern, and There is a reaction time data storage section 36d for storing reaction times. Note that a disk device or a semiconductor memory is used as a storage medium in these storage units.

The reaction stimulus presentation section 30A includes a CRT 30a that generates a visual stimulus to the subject 11, a speaker 30b that generates a stimulus sound, and a built-in vibrating body, and can generate a tactile stimulus that presses the human body. It consists of a seating section 30c on which the subject can sit to easily respond to stimulation. On the other hand, the reaction input section 31A includes a button 31a for the subject 11 to respond to the stimulus with the finger 1c, and a microphone 31b for responding by voice. The reaction from the reaction input section 31A is sent to the selected reaction time measurement section 35.
It is stored in the reaction time data storage section 36d as reaction time data via the CPU 36a of A.

FIG. 31 shows an example of a reaction time measurement situation in which a CRT is used as a reaction stimulus presentation section. In the figure, a subject 11 sits on a seat 30c and holds a bar 40A equipped with a plurality of reaction buttons (not shown). The reaction stimulus is displayed on the CRT 30a in front of the subject 11 and is set to disappear momentarily. When the stimulus was displayed on the screen of the CRT 30a, subject 11
The task is to select and press a button as quickly as possible according to established rules.

On the CRT 30a, a colored circle mark according to the stimulus presentation time and stimulus pattern data set in the selection reaction time measuring section 35A is displayed momentarily with a size and brightness that is sufficiently visible to the subject 11. presented. There are three types of colors, and the order in which these colors are presented is randomly set. Furthermore, the display time intervals are also set randomly. Note that the type of color at this time may be any number of colors.

Next, the procedure for measuring reaction time in this example will be explained according to the flowchart shown in FIG. In step S30a in the same figure, the stimulus presentation time t1, the measurement end time, and TC are set as the missed time when the response is delayed due to a decrease in alertness. Stimulus presentation time t
1 is obtained by generating random numbers, and the presentation time interval may be uniformly and randomly selected from about 5 seconds to 30 seconds, but the time interval and distribution may be set arbitrarily. Further, it is desirable to set the measurement time to 30 minutes to several hours.

Next, in step S31a, a stimulus presentation pattern for selecting a color as a stimulus is randomly set. Here, set a circle with a color (red, blue, yellow). However, this randomness of stimulus generation may be biased, and in extreme situations may be presented in a certain order. This depends on the level of attentiveness, and if you are trying to increase monotony, it is best to set the stimulus pattern to one that is easy to predict.

[0086] In step S32a, the timer 37A is set to C.
The current time is notified to the PU 36a, and in the following step S33a, it is checked whether the stimulus presentation time t1 set in step S30a and the current time ta match. If the result of the determination here is YES, that is, when the set stimulus presentation time has been reached,
In the next step S34a, a predetermined stimulation pattern is presented to the subject. By doing this, it is possible to apply weight until the stimulus occurs.

In step S35a, the timer 37A notifies the CPU 36a of the current time tb after the stimulation is presented. Then, in step S36a, a reaction signal from the subject 11 is detected, and if there is a reaction, tb-ta is calculated as the reaction time in step S37a. However, step S3
If the reaction signal from the subject 11 cannot be detected in step 6a, the process proceeds to step S38, where Tc and waiting time tb-ta, which are preset as a missed time in case the subject's alertness level decreases and the reaction is delayed, are compared. . If this waiting time does not exceed Tc, the process returns to step S35a, where the current time tb is input, and the process waits again for a response signal in the next step S36a. The result of the determination in step S38a is Y.
If it is ES, that is, if the waiting time exceeds Tc, it is determined that the response has been missed, and the reaction time is conveniently set to Tc in step S39a.

In step S40a, the reaction time data for the stimulus presentation time obtained is stored in the reaction time data storage section 3.
6d, and in the following step S41a, the CPU 36a determines whether the measurement end time has come, and the determination result is N.
If YES, the process returns to step S30a to repeat the above process until the set time comes. However, if the measurement end time can be detected in step S41a, this process ends.

FIG. 33 is a block diagram showing the configuration of the electroencephalogram processing section 70. In the electroencephalogram detection section 70a shown in the figure,
Electrodes for brain waves are connected to the human head with a special conductive adhesive, and the brain waves obtained are used, for example, in a head amplifier (equivalent to the physiological measurement amplifier 43 in Figure 24) to counter noise. and amplify the signal. Brain wave detection section 7
The brain wave signal from 0a is sent to the A/D converter 61 (A in FIG. 24).
/D converter 44) and is stored in the brain wave data storage unit 62 as brain wave data. This data is separated into a δ wave band, a θ wave band, an α wave band, and a β wave band using digital filters (63a to 63d), and each band data is stored in a filtering data storage unit (64a to 64d).

Next, the power amount calculation section (65a to 65d)
The average power amount is calculated from the brain wave data of each band. The average time TP (approximately 1 second) at this time is determined in a timely manner. In order to make the average power data obtained in this way an average power amount highly correlated with the degree of arousal, the smoothing processing unit (66a to 66
Smoothing is performed in step 6d). A method for setting the magnitude of the smoothing time at this time will be described later. The average power data is stored in the power data storage section (67a to 67d) for each band.
is stored in

Regarding the electroencephalogram smoothing process, in order to correspond to the reaction time, the power amount from the stimulus presentation time to before the electroencephalogram smoothing time is averaged and used as the electroencephalogram power amount at that time. Here, the smoothing time (Tδ
, Tθ, Tα, and Tβ). These parameters need to be set so that the correlation between the smoothed brain waves (four variables) and the reaction time is the highest. Therefore, using the reaction time as the objective variable, the amount of δ wave power,
Multiple regression analysis is performed in a multiple regression analysis 72a by using four variables, the θ wave power amount, the α power amount, and the β wave power amount, as explanatory variables and changing their smoothing times. Among the multiple correlation coefficients obtained for various combinations of smoothing parameters, the largest multiple correlation coefficient is determined, and the combination of smoothing parameters at that time is determined as the optimal smoothing time for brain waves. At the same time, parameter a1 obtained from multiple correlation analysis
, a2, a3, a4, and a multiple regression equation using these as an arousal level estimation equation. This is predicted by reaction time, but
This is because reaction time and alertness can be considered to have a high correlation and correspond to each other.

Next, although the degree of arousal can be calculated quantitatively, it is necessary to determine at what threshold the stimulus presentation should be started. The determination procedure at this time is shown in the flowchart of FIG. At step S80a in the figure, reaction time data is taken in, and at the next step S81a, the reaction time data is logarithmically transformed. In step S82a, an average value A and a standard value σ are calculated. This is because it is generally known that the reaction time distribution is a Gaussian distribution. In order to know that the arousal has significantly deviated from the average, step S8
In step 3a, the threshold reaction time Tth=A+σ is set to A+2σ, so that when the reaction time becomes slow, it can be determined that the patient is approaching a decline in arousal. Also,
Similarly, if the reaction time becomes smaller than A-2σ from reaction time Tth=A-σ, it is also possible to judge that the person is considerably awake. Based on this idea, Tth is used as the value of the threshold arousal level at which arousal stimulation starts to be applied.

Obtained Tth and Tδ, Tθ, Tα, Tβ
, a1, a2, a3, a4, and a multiple regression equation using these as the alertness estimation parameter. Next, each component of the arousal degree estimating section within the CPUb42 that constitutes the HMM of this embodiment will be explained in detail. <Brain wave detection unit> Electrodes for picking up brain waves are connected to the head of the subject (driver) using a special conductive adhesive, and the brain waves obtained are amplified by a head amplifier (not shown) to counter noise. After that, it is amplified with an electroencephalogram amplifier (not shown). <Brain Wave Processing Section> FIG. 35 shows a detailed block diagram of the brain wave processing section 52a. The brain wave processing unit shown in the same figure has the same functions as the brain wave processing unit shown in FIG. Explanation will be omitted. <Awakening degree estimating section> The arousal degree estimating section 53a uses the arousal degree estimation parameter obtained by the arousal degree estimation parameter determination device shown in FIG. The degree of alertness is calculated from the power data of each band. Then, the value is passed to the wakefulness level determining section 54a.

[0094] The alertness level is expressed by the following estimation formula. That is, arousal level=a1*Pδ+a2*Pθ+a3*Pα+a
4*Pβ+a5

...(2-1) Also, the estimated value of the alertness level is: Estimated alertness level = (-1) x alertness level
...Calculate using (2-2).

FIG. 36 shows an example of estimating the alertness level using the above method.
and shown in FIG. 37. FIG. 36 shows the maximum multiple correlation coefficient, reaction time, and brain wave smoothing time obtained for each subject and the vibration conditions of the vibration device. In the figure, the numerical values represent, from the left, the multiple correlation coefficient, the smoothing time (seconds) of the reaction time when the coefficient was obtained, and the smoothing time (seconds) of the electroencephalogram. Note that the multiple correlation coefficient was rounded to the third decimal place.

FIG. 37 shows the results of FIG. 36 as changes over time in the reaction time and the predicted value of the reaction time using the multiple regression equation. In the figure, the vertical axis represents reaction time, and the horizontal axis represents elapsed time. Furthermore, in the figure, dots indicate reaction times, and solid lines indicate estimated values from brain waves. Note that the breaks between data represent breaks. In this example, the multiple correlation coefficient between reaction time and electroencephalogram is 0.89, which means that the above method is effective for estimating alertness level and that detailed estimation of alertness level is possible. are doing. Although it varies depending on the subject, the magnitude of the multiple correlation coefficient ranges from 0.7 to 0.9.
It can be between 8 and 8.

<Awakening Level Determining Unit> The processing procedure in the alertness level determining unit 54a will be explained with reference to the flowchart shown in FIG. In step S81 of the same figure, the alertness level estimated by the alertness level estimating section 53a is taken in, and the process proceeds to step S82.
The reaction time Rt corresponding to the alertness level is estimated. here,
It is dangerous to immediately determine that a change has occurred in the wakefulness state because Rt has once exceeded Tth. This is because there may be variations due to some factor.

Therefore, in step S83, a threshold frequency NS for signal output is set. In the following step S84, R
Let the value of a frequency counter N representing the frequency at which t exceeds Tth be 0. In other words, when Rt exceeds Tth more than a certain frequency NS, it is determined that arousal has decreased. However, Rt
The sampling time can be determined in a timely manner. The Rt determined in this way is used as a criterion for determining decreased arousal.

After NS is determined in step S83 and the frequency counter value is set to 0, short-term brain wave data is acquired in step S85. Then, in step S86, an estimated reaction time value Reeg (alertness level) is calculated using the estimated parameters and a multiple regression equation. In step S87, Reeg and R
If the estimated reaction time value Reeg does not exceed Rt, the process returns to step S85, and the process of acquiring the brain waves again and calculating the estimated reaction time value Reeg is repeated. However, if the reaction time estimate Reeg exceeds Rt in step S87, 1 is added to the frequency counter N in step S88, and the comparison between the frequency counter value N and the set frequency Nt is performed in step S89. It is done. If the judgment here is NO, the process returns to step S85;
If S, that is, when the value N of the frequency counter exceeds the set frequency Ns, it is determined that there is an arousal abnormality, and the next step S
At 90, an arousal reduction signal indicating a reduction in arousal is output. This arousal reduction signal is sent to the stimulus presentation section.

[0100] In step S91, it is determined whether the process is finished. Note that if Reeg exceeds Rt and does not exceed Rt for a while, a mechanism may be installed to reset the counter N as a noise. <Stimulus Presentation Unit> The stimulus presentation unit 50a is an arousal degree determination unit 54a.
Driver 11, who is a test subject, receives an arousal reduction signal from
Arousing stimuli such as sounds, vibrations, and scents are output for a certain period of time until a stop signal is received. Note that at this time, the type of stimulus and the method of presentation may be those obtained by the arousal stimulus setting device described later.

FIG. 39 is a block diagram showing the configuration of the arousal stimulus setting device. The present arousal stimulus setting device determines in advance the most appropriate type of stimulus and method of stimulus presentation for the individual, since the stimulus that increases the arousal effect may vary depending on the individual. In this device, the electroencephalogram processing section 102, reaction time measurement section 103, and arousal level estimation parameter determination section 101, excluding the stimulus presentation section 104 and the stimulation parameter setting section 110, have the same functions as the arousal state determination device shown in FIG. Therefore, its explanation will be omitted.

In FIG. 39, subject 11 is made to perform a choice reaction task, and the brain waves at that time are simultaneously recorded. Then, the alertness level estimation parameter determination unit determines parameters from the obtained reaction time and brain waves, and estimates the alertness level. Note that if the subject 11 has undergone evaluation by the arousal level estimation parameter determination unit 72 of the arousal state determination apparatus shown in FIG. 29 before being evaluated by this device, the arousal level can be estimated immediately.

[0103] In the stimulation parameter setting section 110, the type of stimulation and presentation method for determining the optimal stimulation state are provisionally set in advance. Then, the stimulus presentation unit 104
If it is determined that the arousal is in a state of decreased arousal, the stimulation parameter setting unit 11
According to the data set to 0, a sound is emitted from the speaker 105a via the speaker drive unit 105, a seat 106a with a built-in vibrator is vibrated at high frequency via the vibration drive unit 106, or the vibration drive unit 107 is caused to vibrate at a high frequency. The vibrating device 107a is caused to vibrate at a low frequency to present a stimulus to the subject 11. Then, the subject's level of arousal is estimated from the brain waves when the stimulus is given, and the estimated arousal level before and after the stimulus is presented to see if it is effective.

[0104] By changing the stimulation state and type of stimulation, it is possible to select the most effective stimulation for arousal control.
Furthermore, based on the obtained results, it is possible to maintain individual arousal and increase the effectiveness of arousal control. Next, Figure 4
25, the low arousal coping process corresponding to step S35 in the flowchart of FIG. 25 will be described in detail with reference to the flowchart shown in FIG. In the figure, step S10
1 resets the safe limit sound level counter, that is, NΔ
Set to 1. In the next step S102, as will be described later, an effective value Ea of the speaker generated sound is calculated, which indicates the power of the noise amount (amplified engine sound).

[0105] In step S103, as a safety volume check, the effective value E of the speaker generated sound obtained in step S102 is checked.
a and the safety limit engine volume Esa set as the initial value
Compare the level with fe. If it is determined that Ea is larger than Esafe, the sound generated by the speaker exceeds the safety limit and is dangerous, so the process proceeds to step S104, executes the sound generation process at the safety limit, which will be described later, and returns to step S102. .

However, in step S103, Ea is
If it is determined that it is smaller than fe, step S10
In step S5, a speaker sound, which will be described later, is generated, and in the following step S106, an estimated alertness level X2 is taken in according to the alertness level estimation method described above. In step S107, it is determined whether the driver's wakefulness level has reached the optimum level of alertness due to the generation of the speaker sound, that is, the alertness level estimate X2 and Rbest-
A comparison is made with the value of Rerr2. Here X2 is Rbe
If it is smaller than st-Rerr2, it is determined that the driver has not returned to the normal wakefulness state, and the process returns to step S102. However, if it is determined that X2 is larger than Rbest-Rerr2, it is assumed that the driver has returned from the low arousal state to the normal arousal state and this process ends.

FIG. 41 is a flowchart showing the procedure for calculating the effective value of the sound generated by the speaker. In the figure, in step S111, the CPUa41 takes in engine sound ein(t) via the microphone 46. here,
Engine sound capture from ein(t0) to ein(
t0 + T1). In the following step S112, the engine sound emitted from the speaker 53 (this is referred to as the ear sound S(t)) is captured via the ear microphone 52 placed near the driver's ear. This ear sound S (
t) is taken from S(t0) to S(t0 +T1
).

[0108] In step S113, according to equation (3), the effective value E of the engine sound captured in step S111 is calculated.
0, and in step S114, according to equation (4), the effective value Se of the ear sound taken in step S112 is calculated.
Calculate ar.

[0109]

[Math 3]

[0110]

[Math 4]

[0111] In step S115, the engine sound reproduction increase rate Δe is calculated according to the following formula. Δe=Δ・(X-Rbest)・NΔ
(5) Then, in the next step S116, the effective value of the reproduced engine sound (effective value of the sound generated by the speaker) Ea is calculated according to equation (6). Ea=E0・Δe

(6) In step S117, in order to protect the driver's ears, the sound pressure Eb during reproduction, which is the effective value of the ear sound when the engine sound is reproduced, is calculated according to equation (7).

[0112]

[Math. 7]

FIG. 42 is a flowchart showing the speaker sound generation process. In step S121 of the same figure, the CPU
a41 is step S of the flowchart shown in FIG.
113, and the effective value E0 of the engine sound found in S116.
and reproduction engine sound effective value (speaker generated sound effective value) E
From a, the gain G of the amplifier 50 is determined by the following formula. G=Ea/E0

(8) In the next step S122, the gain control signal eout (t) is calculated. This signal is obtained using the following formula.

eout (t)=G・e(t)

(9) FIG. 43 is a flowchart showing the processing procedure for the sound generated at the safety limit. In the same figure, step S1
At step 25, the safety sound playback time counter is reset (N
safe=1). Next, as the reproduction of the safe limit sound, the safe limit engine volume Esafe is reproduced in step S126, and in the following step S127, the reproduction is continued for the engine sound reproduction time Tsafe seconds at the safe limit.

[0115] In step S128, the speaker generated sound Ea is calculated using the same procedure as in the flowchart shown in Fig. 41, and in the following step S129, as a safety sound check,
A level comparison is made between the sound pressure Eb during reproduction and the safety limit engine sound volume Esafe set as an initial value. If it is determined in this step S129 that Eb is smaller than Esafe, the process ends, but if it is determined that Eb is larger than Esafe, the process proceeds to step S130, and the safety sound reproduction time counter Nsafe is set to the safety limit counter. It is determined whether the value is larger than Nsafe-lim.

[0116] If the determination result in step S130 is NO, it is determined that the number of reproductions of the safe limit engine volume Esafe has not reached the predetermined number of times, and the process returns to step S126. However, if the determination in step S130 is YES,
Even if the safe limit engine volume Esafe is presented a predetermined number of times, it has no effect on improving the driver's alertness and the driver's tolerance is exceeded, so a warning is issued in step S131.

Next, the overstress countermeasure process (step S36 in FIG. 25) according to this embodiment will be explained. Figure 44 shows
It is a flowchart showing a procedure for dealing with excessive tension. In the figure, in step S141, the scent presentation counter Ns is reset (Ns=1), and in step S142,
A calming scent (here, a diasmine scent) is presented. Then, in step S143, the estimated wakefulness level X3 is taken in according to the wakefulness level estimation method described above.

[0118] In step S144, step S142
Determining whether the driver's arousal level has reached the optimal arousal level due to the presentation of the scent, that is, the estimated arousal level X3 and Rbe
A comparison is made with the value of st+Rerr2. If X3 is greater than Rbest+Rerr2, it is determined that the driver is still in an overstressed state and has not returned to a normal alert state, and the process proceeds to step S146. In step S146, it is determined whether the scent presentation counter Ns is larger than the limit number of scent presentations Ns-lim. If the result of the determination is NO, the value of the scent presentation counter Ns is incremented by 1 in step S147. After that, the process returns to step S142 again.

However, the determination result in step S146 is Y.
If it is ES, it is determined that the fragrance presentation limit has been exceeded, and a warning is issued in step S148, and the process ends. Step S1
If it is determined in step S144 that X3 is smaller than Rbest+Rerr2, it is assumed that the driver has returned from the hypertonic state to the normal alert state, and the presentation of the scent is stopped in the next step S145.

As described above, according to this embodiment, the degree of alertness can be detected quantitatively and accurately using brain waves, and by providing stimulation to the driver according to the degree of alertness, a decrease in alertness can be prevented. This has the effect of being able to maintain the driver's individual optimum state of alertness. In addition, the present invention has the effect that it is possible to bring back the arousal level of a driver who has fallen into an excessively nervous state to a normal state, and guide him or her to a safe driving state, not only when the driver's arousal is low.

[0121] In the third embodiment described above, the degree of wakefulness was determined quantitatively using brain waves, but the driver's wakefulness state can be maintained simply by the following method. FIG. 45 is a block diagram showing the overall configuration of an arousal state determination device for quantitatively detecting an arousal rhythm. The device shown in the figure measures brain waves from the subject 11, processes them into physical quantities closely related to the degree of wakefulness, and detects the wakefulness rhythm cycle based on frequency analysis of the processed brain wave data.

In FIG. 45, the wakefulness rhythm measuring section 20c
Then, brain wave data from the subject 11 is measured and stored, and the wakefulness rhythm detection section 21c detects the wakefulness rhythm cycle based on the obtained data. Then, in the stimulus presentation section 22c,
A stimulus such as sound, which will be described later, is generated based on the wakefulness rhythm cycle. The wakefulness rhythm measurement section 20c receives signals from an electrode 2 that detects the brain waves of the subject 11, a head amplifier section 3c that preamplifies the weak current from the electrode 2 in order to improve the S/N ratio, and a signal from the head amplifier section 3c. Next-stage A/D converter 5c
The main amplifier section 4c performs signal amplification to obtain a level sufficient for conversion into digital data, and the main control section 6c controls the entire device. The main control section 6c includes a CPU 6a functioning as a calculation section, and an electroencephalogram data storage section 6b for storing digital signals converted by the A/D conversion section 5c. Note that an anti-aliasing filter may be provided before the A/D conversion section, or the sampling frequency may be made sufficiently high.

[0123] The wakefulness rhythm detecting unit 21c takes in brain wave data from the brain wave data storage unit 6b, and converts the brain waves into δ waves (1 to 3 Hz) and θ waves using a band filter 7c, which is a digital filter.
waves (4-7Hz), alpha waves (8-13Hz), beta waves (18
~30Hz) into four bands. Then, the power amount calculation unit 8c calculates the power amount of each brain wave band data.

[0124] To further remove noise from the obtained power amount, the smoothing time setting section 10c sets an appropriate smoothing time TS, and the smoothing processing section 9c performs smoothing processing, making it easier to detect the brain wave rhythm. Convert to data. The smoothed data is stored in the smoothed data storage unit 11c, and the frequency analysis unit 12c performs frequency analysis to read the peak of the level on the frequency, thereby detecting the rhythm cycle of wakefulness.

[0125] The arousal rhythm cycle obtained as a result of the frequency analysis is stored in the arousal rhythm cycle storage section 13c of the stimulus presenting section 22c, and the timer 14c and the stimulus generating section 15c are linked to generate the alarm at a predetermined time for a predetermined period of time. Present stimuli at intervals. With reference to the flowchart shown in FIG. 46, the wake-up rhythm detection processing procedure in the wake-up rhythm detection section 21c will be described.

In step S1c of FIG. 46, the brain wave data is taken in from the brain wave data storage section 6b, and in step S2c,
As mentioned above, the digital filter converts the brain wave data into δ waves,
Divided into four bands: θ waves, α waves, and β waves. Then, in the following step S3c, the power amount of the brain wave data for each band is calculated. Here, the square mean of the time series data of the filtered power spectrum for each band is taken. The average time TP at this time is approximately 1 second, and the sampling period is approximately 30 seconds. This makes it possible to follow time-series changes in the power spectrum for each band and at the same time reduce the amount of data.

In step S4c, smoothing processing is performed on the power amount obtained as a result of the calculation in step S3c by setting an appropriate smoothing time TS in order to further remove noise and obtain a tendency of change in power amount over a long period of time. to convert brain wave rhythms into data that is easy to detect. Note that parameter research has revealed that it is easier to detect brain wave rhythms when the smoothing time TS is set to approximately 300 seconds. In step S5c, the smoothed data is stored in the smoothed data storage section 11 in preparation for the next frequency analysis.

In step S6c, frequency analysis is performed on the smoothed data for each band, and the cycle of the wakefulness rhythm is detected by reading the peak of the level on the frequency. For frequency analysis here, Fourier transform or maximum entropy method (MEM) is used. In particular, since Fourier transform is ineffective for data with strong randomness, rhythm cycle detection using MEM, which can easily detect peaks, is effective.

[0129] Figure 47 shows a comparison of the results of analyzing the frequency obtained from the time it takes for a human to respond to a given stimulus and the rhythm analysis of brain waves (in this case, alpha waves), where the solid line indicates the response. The wavy lines represent time and brain waves. In terms of frequency, if there is a peak, it can be interpreted that there is an arousal rhythm at that peak frequency. As can be seen from Figure 47, the reaction time and the peak of the electroencephalogram are in good agreement.
Furthermore, since brain waves, especially alpha waves and theta waves, appear in a state that corresponds well to a state of decreased arousal, the brain wave rhythm can be directly interpreted as an arousal rhythm. The existence of such a rhythm means that the arousal state regularly repeats an arousal state and a low arousal state in multiple complex cycles.

FIG. 48 shows how stimulation is applied to the driver of the vehicle in accordance with the wakefulness rhythm cycle obtained by the wakefulness state determination device. In the same figure, a certain type of sound is generated from the stimulation presentation unit 22c of the wakefulness state determining device 100c, and the sound from the speaker 25c stimulates the driver to maintain wakefulness. As a stimulus,
In addition to sound, things that increase alertness, such as vibration, scent, and heat, may be used. As will be described later, this device continues to provide stimulation to the driver by tracking the individual's unique rhythm of arousal, for example, the phenomenon in which arousal increases and decreases after about 30 minutes, at almost regular intervals.

FIG. 49 schematically shows the change in alertness level over time. Next, the wakefulness sustaining processing procedure will be explained according to the flowchart shown in FIG. At step S10c in FIG. 50, the individual-specific wakefulness rhythm period Tr detected by the wakefulness rhythm detection section 21c is set in the wakefulness rhythm period storage section 13c, and at the next step S11c, when the wakefulness decrease time arrives in the timer 14c, , a time interval Δt (stimulus presentation time width) is set to determine how much stimulation is to be applied. Then, in step S12c, t=0 is similarly set in the timer 14c as the operation start time.

[0132] Time measurement is started from the time indicated by t=0, and in step S13c, it is determined whether one half of the time of the awakening rhythm period Tr has elapsed. If the determination here is YES, in the following step S14c, the timer 14c is again set to t=0 as the driving start time, and in step S15c, the stimulation generator 15c applies stimulation for a certain period of time as the first stimulation to the driver. give. This is because the driver is in an awake state at the start of driving, and the first state of decreased arousal is thought to arrive in half the time of the awakening rhythm cycle Tr. Therefore, after that, the state of reduced arousal occurs every wakefulness rhythm cycle Tr, and stimulation is given to the driver in accordance with that cycle.

[0133] In step S16c, the awakening rhythm cycle Tr
The timer setting time t is calculated from the arithmetic difference between the stimulus presentation time width Δt and
It is determined whether the stimulus presentation time has been reached, and if the determination is YES, it is determined whether the stimulus presentation time has not been exceeded in the next step S17c. While the determination at step S17c is NO, the stimulus presentation at step S19c continues. However, if it is determined in step S17c that the stimulus presentation time has exceeded, the stimulation is stopped in step S18c.

In step S20c, an operation related to the end of driving, such as pulling out the ignition key, is detected. If the operation is not completed, proceed to step S.
At step 21c, t=0 is set in the timer 14c. but,
If it is determined that the operation has ended, this process ends. By doing this, it is possible to detect brain waves, quantitatively detect the wakefulness rhythm from frequency analysis of time-series data of the brainwaves, and maintain the driver in an alert state by providing stimulation according to that rhythm. It becomes possible to do so.

Next, a method for detecting wakefulness rhythm based on reaction time to stimulation will be explained. FIG. 51 is a block diagram showing the overall configuration of an arousal state determination device for detecting an arousal rhythm based on reaction time to a stimulus. In the same figure, the wakefulness state determination device of the third embodiment (Fig. 3
0), and the same components as those of the wakefulness state determination device (FIG. 45) for quantitatively detecting the wakefulness rhythm are given the same numbers, and detailed explanation thereof will be omitted here.

[0136] Furthermore, the reaction time measurement situation and reaction time measurement procedure here are also similar to the measurement situation in the third embodiment (
31) and the measurement procedure (FIG. 32), the explanation thereof will be omitted. The flowchart shown in FIG. 52 shows an awakening rhythm detection processing procedure based on the reaction time in the awakening rhythm detection section 21c of FIG. 51.

At step S50c in FIG. 52, reaction time data is fetched from the reaction time data storage section 36d, and at the next step S51c, the reaction time data is subjected to logarithmic conversion in the logarithmic conversion section 38c. Next, in step S52c, a smoothing time width TS is set, and in step S53c, smoothing for noise removal and averaging processing for frequency analysis are performed. This TS is set to approximately 300 seconds since the period of the wakefulness rhythm is 5 minutes or more. In addition, the average reaction time when the stimulus is presented during a time period TS ahead of the measurement start time is calculated. How to find this average is
It may be an additive average, a weighted average, or an exponential average. In this way, the sampling period T
Shift the time by samp and repeat the same operation to find the reaction time up to the final measurement time. The setting time of Tsamp should be determined based on the characteristics of the random number of the stimulus presentation time width. For example, if the stimulus is set to occur uniformly in about 20 seconds, it is preferable to set Tsamp to approximately 20 seconds.

[0138] In step S54c, the reaction time data subjected to the above processing is stored in the smoothed data storage section 11c,
In the next step S55c, frequency analysis is performed on the reaction time, and the period of the wakefulness rhythm is determined by detecting the peak of the level on the frequency. Frequency analysis methods include the Freeier transform method and the maximum entropy method (MEM).
).

FIG. 53 shows an example of frequency analysis performed by applying the maximum entropy method to reaction time data. In this example, there are peaks at 8 minutes and 3 minutes, and it can be seen that the subject subject to measurement has a characteristic of having an awakening rhythm of 8 minutes and 3 minutes. Based on the wakefulness rhythm cycle obtained in this way, the stimulus presentation section 22c provides a stimulus to the driver of the vehicle, thereby making it possible to maintain the driver's wakefulness state. The manner in which the stimulus is presented to the driver using the stimulus presenting section 22c is similar to that shown in FIG. 48, so illustration and description thereof will be omitted. Also,
The arousal sustaining processing procedure in the stimulus presenting unit 22c is also the same as the process in the stimulus presenting unit 22c described above (FIG. 50), so the explanation will be omitted.

[0140] In this way, since the rhythmic cycle of wakefulness can be quantitatively measured from the reaction time to a stimulus, the rhythmic cycle of wakefulness can be measured in accordance with the detected rhythmic cycle of wakefulness when driving on a monotonous highway for a long time. By applying a stimulus to a driver, it is possible to bring the driver back into an alert state before the driver falls into a state of alertness, thereby achieving sustained alertness. <Modification> In the HMM of the third embodiment, the alertness level is estimated as a numerical value from the brain waves, but the alertness level can be estimated based on heart rate or blink frequency in addition to the brain waves.

[0141] First, a method for estimating the degree of alertness based on heart rate will be explained. In FIG. 54, the cardiac potential detection unit 51
aa is composed of a portion that detects the cardiac potential from the electrodes 2 attached to the heart of the driver 11 who is the subject, and amplifies the cardiac potential signal by an amplifier (not shown). The cardiac potential processing section 52aa processes the cardiac potential signal from the cardiac potential detection section 51aa and processes the signal into a physical quantity that facilitates estimation of wakefulness. That is, the cardiac potential signal amplified by the cardiac potential detection unit 51aa is taken in via an A/D converter, which will be described later, and processed into a heart rate necessary for evaluating the alertness level by calculation.

[0142] The wakefulness estimating unit 53a calculates an estimated value closely related to wakefulness based on wakefulness estimation parameters and heart rate data, which will be described later. Further, the arousal level determining unit 54a determines how much stimulation to present based on the arousal level estimated by the arousal level estimating unit 53a. And the stimulus presentation section 50a
The controller presents or stops appropriate stimulation to the driver according to instructions from the alertness level determination unit 54a.

[0143] The alertness level estimating section of this modification estimates the alertness level based on the heart rate and reaction time, and uses this information to provide optimal stimulation to the driver in a biofeedback manner. Note that the concept of biofeedback applied to this modification is the same as that explained in FIG. 28 for the third embodiment, so its explanation is omitted here, but also in this modification, In principle, it is possible to maintain the optimal wakefulness state shown in FIG. 28 by detecting the human wakefulness state and applying stimulation accordingly.

[0144] First, a description will be given of an arousal degree estimation parameter determining device that determines parameters for calculating the arousal degree in the arousal degree estimating section 53a. The wakefulness estimation parameter determining device simultaneously measures a reaction time and a cardiac potential signal that are highly correlated with the degree of wakefulness, and calculates the heart rate from the cardiac potential. Then, the arousal level estimation parameter is determined by performing regression analysis based on the obtained heart rate data and reaction time data.

FIG. 55 is a schematic block diagram of the arousal level estimation parameter determination device. The figure shows a situation in which this device simultaneously measures a person's reaction time and heart rate over a long period of time, and calculates estimated parameters by analyzing the correlation between the obtained reaction time and heart rate. be able to. In FIG. 55, the cardiac potential processing unit 70aa controls the subject (
The cardiac potential from the electrode 2 attached to the driver (driver) 11 is measured, and the reaction time measuring section 71 measures the time it takes for the subject 11 to react to the stimulus given. Based on these measurement results, the arousal level estimation parameter determination unit 72 estimates the subject's own arousal level.

[0146] The method for measuring the reaction time (selected reaction time) in this modification is the same as that in the third embodiment, so its explanation will be omitted (see FIG. 30). FIG. 56 is a block diagram showing the configuration of the cardiac potential processing section 70aa. In the electrocardiogram detection unit 149 shown in the figure, electrocardiogram electrodes are connected to the parts that sandwich the human heart with a special conductive adhesive, and the electrocardiogram signal obtained thereby is transmitted, for example. A head amplifier (corresponding to the physiological measurement amplifier 43 in FIG. 24) takes noise countermeasures and amplifies the signal.

[0147] The cardiac potential signal from the cardiac potential detection section 149 is
A/D converter 61 (corresponding to A/D converter 44 in FIG. 24)
The electrocardiographic data storage unit 62a stores the electrocardiographic data as electrocardiographic data via the
It is stored in a. The heart rate is stored in the electrocardiogram data storage section 62a.
The original heart rate calculation unit 150 calculates the heart rate based on the electrocardiogram data stored in a. The obtained original heart rate data is smoothed by the averaging processing unit 152 in order to obtain average heart rate data that is highly correlated with the degree of alertness. Note that the time setting for this smoothing will be described later. Further, the heart rate data after the averaging process is stored in the heart rate storage section 153.

[0148] Regarding the smoothing process of heart rate data, in order to correspond to the reaction time, the heart rate from the stimulus presentation time to before the heart rate smoothing time is averaged, and this is taken as the heart rate at that time. . Here, it is necessary to determine the heart rate smoothing time. These parameters need to be set so that the correlation between the heart rate and reaction time after smoothing is the highest. Therefore, the regression analysis section 72aa performs a regression analysis by using the reaction time as an objective variable and the heart rate as an explanatory variable while changing these smoothing times. Among the correlation coefficients obtained for various combinations of smoothing parameters in this way,
The maximum correlation coefficient is determined, and the combination of smoothing parameters at that time is determined as the optimal smoothing time for heart rate. Furthermore, the parameters a1 and a2 obtained from the correlation analysis and a regression equation using these parameters are defined as an alertness estimation equation. This is because although reaction time is predicted, reaction time and alertness can be considered to have a high correlation and correspond to each other.

Next, although the degree of arousal can be calculated quantitatively, it is necessary to determine at what threshold the stimulus presentation should be started. The determination procedure at this time is the same as the flowchart in FIG. 34 (third embodiment). here,
The obtained Tth, a1, and a2 and the regression equation using them are used as alertness estimation parameters. Next, each component of the arousal degree estimating section of this modification will be explained in detail. <Cardiopotential detection unit> Electrodes for picking up the cardiac potential are connected to the part that pinches the heart of the subject (driver) using a special conductive adhesive, and the obtained cardiac potential signal is sent to a head amplifier (not shown). After being amplified and taking noise countermeasures, it is amplified by a sanitary amplifier (not shown). <Cardiopotential Processing Unit> FIG. 57 shows a detailed block diagram of the cardiac potential processing unit 52aa. The cardiac potential processing section in the same figure has the same function as that shown in FIG. 56, so the same components are denoted by the same reference numerals, and detailed explanation thereof will be omitted here. <Awakening level estimating unit> The alertness level estimating unit 53a calculates the heart rate processed by the electrocardiogram processing unit 52aa using the alertness level estimation parameter obtained by the alertness level estimation parameter determination device shown in FIG. 55 and the regression equation. Calculate alertness level from numerical data. Then, the value is passed to the wakefulness level determining section 54a.

[0150] The degree of alertness is expressed by the following estimation formula. That is, alertness level=a1*H+a2
…
(10) However, FIG. 58 shows an example in which H is estimated by the method using the heart rate or higher. FIG. 58 shows the maximum correlation coefficient obtained for each subject and the vibration conditions of the vibration device. Note that the correlation coefficient was rounded to the third decimal place. Further, the smoothing time is 300 seconds for both reaction time and heart rate.

[0151] From FIG. 58, there is a high correlation between heart rate and reaction time, which means that detailed estimation of the degree of alertness is possible. However, this method does not provide as good accuracy as the estimation of alertness level using brain waves shown in the third embodiment. <Awakening Level Determining Unit> The processing procedure in the alertness level determining unit 54a is the same as the flowchart shown in FIG. 38 (third embodiment), so a description thereof will be omitted. <Stimulus Presentation Unit> The stimulus presentation unit 50a is an arousal degree determination unit 54a.
Driver 11, who is a test subject, receives an arousal reduction signal from
This device outputs arousing stimuli such as sounds, vibrations, and scents for a certain period of time until a stop signal is received.

FIG. 59 is a block diagram showing the configuration of the arousal stimulus setting device. The present arousal stimulus setting device determines in advance the most appropriate type of stimulus and method of stimulus presentation for the individual, since the stimulus that increases the arousal effect may vary depending on the individual. In this device, the stimulation presenting section 104, the electrocardiogram processing section 161 excluding the stimulation parameter setting section 110, the reaction time measuring section 103, and the arousal level estimation parameter determining section 101 have the same functions as the arousal state determination device shown in FIG. Therefore, its explanation will be omitted.

In FIG. 59, subject 11 is made to perform a choice reaction task, and the electrocardiogram at that time is simultaneously recorded. and,
The alertness level estimation parameter determination unit determines parameters from the obtained reaction time and heart rate, and estimates the alertness level. Note that if the subject 11 has undergone evaluation by the arousal level estimation parameter determination unit 72 of the arousal state determination apparatus shown in FIG. 55 before being evaluated by this device, the arousal level can be estimated immediately.

[0154] In the stimulation parameter setting section 110, the type of stimulation and presentation method for determining the optimal stimulation state are provisionally set in advance. Then, the stimulus presentation unit 104
If it is determined that the arousal is in a state of decreased arousal, the stimulation parameter setting unit 11
According to the data set to 0, a sound is emitted from the speaker 105a via the speaker drive unit 105, a seat 106a with a built-in vibrator is vibrated at high frequency via the vibration drive unit 106, or the vibration drive unit 107 is caused to vibrate at a high frequency. The vibrating device 107a is caused to vibrate at a low frequency to present a stimulus to the subject 11. The arousal level of the subject is then estimated from the heart rate when the stimulus is applied, and the estimated arousal level before and after the stimulation is compared to see if it is effective.

[0155] By changing the stimulation state and type of stimulation, it is possible to select the most effective stimulation for arousal control.
Furthermore, based on the obtained results, it is possible to maintain individual arousal and increase the effectiveness of arousal control. As described above, according to this embodiment, the degree of alertness can be detected quantitatively and accurately based on the heart rate, and by providing stimulation to the driver according to the degree of alertness, a decrease in alertness can be prevented and the driver It has the effect of being able to maintain an individual's optimal state of arousal.

[0156] In addition, the present invention has the effect that it is possible to restore the alertness of a driver who has fallen into an excessively nervous state to a normal state and guide him to a safe driving state, not only when his alertness is low. Next, a method for estimating arousal level based on blink frequency will be explained. In FIG. 60, the blink detection section 51bb
is composed of a part that detects eye movement from electrodes 2 attached above and below or on the left and right sides of the eyeball of the driver 11 who is the subject, and amplifies the eye movement potential by an amplifier (not shown). The blink processing section 52bb processes the eye movement signal from the blink detection section 51bb to perform signal processing into a physical quantity that facilitates estimation of wakefulness. That is, the eye movement signal amplified by the eye blink detection section 51bb is taken in via an A/D converter, which will be described later, and processed by calculation into the eye blink frequency necessary for evaluating the degree of alertness.

[0157] The wakefulness estimation unit 53a calculates an estimated value closely related to wakefulness based on wakefulness estimation parameters and blink frequency data, which will be described later. Further, the arousal level determining unit 54a determines how much stimulation to present based on the arousal level estimated by the arousal level estimating unit 53a. Then, the stimulus presenting unit 50a presents or stops appropriate stimuli to the driver according to instructions from the arousal level determining unit 54a.

[0158] The arousal level estimating unit of this embodiment performs blinking,
The driver's level of alertness is estimated based on the driver's response time and response time, and the information is used to provide optimal stimulation to the driver in a biofeedback manner. Note that the concept of biofeedback applied to this modification is the same as that explained in FIG. 28 for the third embodiment, so its explanation is omitted here, but also in this modification, In principle, it is possible to maintain the optimal wakefulness state shown in FIG. 28 by detecting the human wakefulness state and applying stimulation accordingly.

Next, a description will be given of an arousal level estimation parameter determining device that determines parameters for calculating the arousal level in the arousal level estimating section 53a. The alertness estimation parameter determination device simultaneously measures reaction time and eye movement signals, which are highly correlated with the degree of alertness, and performs regression analysis based on the blink frequency data and reaction time data obtained from the eye movements. This determines the alertness estimation parameter.

FIG. 61 is a schematic block diagram of the arousal level estimation parameter determination device. The figure shows a situation in which this device simultaneously measures a person's reaction time and eye movements over a long period of time, and calculates estimated parameters by analyzing the correlation between the obtained reaction time and blink frequency. be able to. In the alertness estimation parameter determining device, a blink processing unit 70bb measures eye movements from an electrode 2 attached to a subject (driver) 11, and a reaction time measuring unit 71 measures the time it takes for the subject 11 to react to a given stimulus. are measured, and based on these measurement results, the arousal degree estimation parameter determination unit 72 estimates the subject's unique arousal degree.

[0161] The method for measuring the reaction time (selected reaction time) in this modification is the same as that in the third embodiment, so its explanation will be omitted (see FIG. 30). FIG. 62 is a block diagram showing the configuration of the blink processing section 70bb. In the eye movement detection unit 155 shown in the same figure, electrodes are used to connect the upper and lower or left and right sides of the human eyeballs with a special conductive adhesive, and the eye movement signals obtained thereby are sent to a head amplifier ( Noise countermeasures are taken using the physiological measurement amplifier 43 (corresponding to the physiological measurement amplifier 43 in FIG. 24), and the signal is amplified.

The eye movement signal from the eye movement detection section 155 is sent to the eye movement signal data storage section 62bb as eye movement signal data via the A/D converter 61 (corresponding to the A/D converter 44 in FIG. 24). Stored. The eye movement signal data is generated by the original blink frequency calculation unit 15 based on the eye movement signal data stored in the eye movement signal data storage unit 62bb.
It is calculated as 6.

[0163] The obtained original blink frequency data is smoothed by the averaging processing unit 158 in order to obtain average blink frequency data that is highly correlated with the degree of alertness. The average blink frequency data after the averaging process is stored in the blink frequency data storage section 159. The procedure for setting the time for smoothing and determining the threshold when starting stimulus presentation is the same as the method for estimating the arousal level based on the heart rate described above, so we will not explain it here. is omitted.

[0164] Next, each component of the alertness level estimating section of this modification will be explained in detail. <Eye movement detection unit> Electrodes for picking up eye movement signals are connected to the top and bottom or left and right sides of the subject's (driver's) eyeballs using a special conductive adhesive, and the obtained signals are sent to a head amplifier (not shown). After being amplified and taking noise countermeasures, it is amplified by a sanitary amplifier (not shown). <Eye movement signal processing section> FIG. 63 shows the eye movement signal processing section 52.
bb shows a detailed block diagram. The blinking processing section in this figure has the same function as that shown in FIG. 62, so the same components are denoted by the same reference numerals, and detailed explanation thereof will be omitted here. <Awakening degree estimating section> The arousal degree estimating section 53a uses the arousal degree estimation parameter and regression equation obtained by the arousal degree estimation parameter determination device shown in FIG.
The degree of alertness is calculated from the blink frequency data processed in . Then, the value is passed to the wakefulness level determining section 54a.

[0165] The degree of wakefulness is expressed by the following estimation formula. That is, alertness level=a1*E+a2
…
(11) However, E is the blink frequency. Examples of arousal level estimation using the above method are shown in FIGS. 64 and 65. Figure 6
4 indicates the maximum correlation coefficient obtained for each subject and the vibration conditions of the vibration device. Note that the correlation coefficient was rounded to the third decimal place. Further, the smoothing time is 300 seconds for both reaction time and heart rate.

FIG. 65 shows the results of FIG. 64 showing changes over time in predicted values based on reaction time and blink frequency. In the figure, the vertical axis represents logarithmic reaction time, and the horizontal axis represents elapsed time. In addition, the dots in the figure indicate the reaction time, and the solid line indicates the reaction time.
Shows predicted values based on blink frequency. Note that data breaks represent breaks between experimental sessions. From these figures, it can be seen that the above-described method is effective in estimating the degree of alertness, and allows for detailed estimation of the degree of alertness. however,
This method does not provide as good accuracy as the estimation of alertness level using brain waves shown in the third embodiment.

<Awakening Level Determining Unit> The processing procedure in the alertness level determining unit 54a is the same as the flowchart shown in FIG. 38 (third embodiment), so a description thereof will be omitted. <Stimulus Presentation Unit> The stimulus presentation unit 50a is an arousal degree determination unit 54a.
Driver 11, who is a test subject, receives an arousal reduction signal from
This device outputs arousing stimuli such as sounds, vibrations, and scents for a certain period of time until a stop signal is received.

FIG. 66 is a block diagram showing the configuration of the arousal stimulus setting device. The present arousal stimulus setting device determines in advance the most appropriate type of stimulus and method of stimulus presentation for the individual, since the stimulus that increases the arousal effect may vary depending on the individual. In addition, the reaction time measurement section 103 and the alertness level estimation parameter determination section 101 in this device are as follows:
Since the function is the same as that of the wakefulness determination device shown in FIG. 61, a description thereof will be omitted.

In FIG. 66, the subject 11 is made to perform a selection reaction task, and the eye movements at that time are simultaneously recorded. Then, the alertness level estimation parameter determination unit determines parameters based on the obtained reaction time and blink frequency, and estimates the alertness level. In addition, before receiving the evaluation using this device, subject 11
If the evaluation is performed by the arousal level estimation parameter determination unit 72 of the arousal state determination device shown in 1, the arousal level can be estimated immediately.

[0170] In the stimulation parameter setting section 110, the type of stimulation and presentation method for determining the optimal stimulation state are provisionally set in advance. Then, the stimulus presentation unit 104
If it is determined that the arousal is in a state of decreased arousal, the stimulation parameter setting unit 11
According to the data set to 0, a sound is emitted from the speaker 105a via the speaker drive unit 105, a seat 106a with a built-in vibrator is vibrated at high frequency via the vibration drive unit 106, or the vibration drive unit 107 is caused to vibrate at a high frequency. The vibrating device 107a is caused to vibrate at a low frequency to present a stimulus to the subject 11. Then, the subject's level of arousal is estimated from the blink frequency when the stimulus is applied, and the estimated arousal level before and after the stimulus is presented to see if it is effective.

[0171] It is possible to select the most effective stimulus for arousal control by changing the stimulus state and type of stimulus.
Furthermore, based on the obtained results, it is possible to maintain individual arousal and increase the effectiveness of arousal control. As described above, according to this embodiment, the degree of alertness can be detected quantitatively and accurately based on the blink frequency, and by providing stimulation to the driver according to the degree of alertness, a decrease in alertness can be prevented and the driver It has the effect of being able to maintain an individual's optimal state of arousal.

[0172] In addition, the present invention has the effect that it is possible to restore the level of alertness of a driver who has fallen into an excessively tense state to a normal state, and to guide him or her to a safe driving state, not only when the driver's alertness is low. In addition, although the target is limited to the driver in the above embodiment,
For example, users such as computers can also be applied. Furthermore, the scope of application is not limited to the automobile industry, but can also be applied to life-related industries such as home appliances and construction, where there is a human-machine relationship.

[0173]

[Effects of the Invention] As explained above, according to the present invention,
By detecting human sensibilities and adding human-like emotional functions to machines according to their state, machines can be made into pseudo-humans, and humans can gain psychological satisfaction from interactions between humans and machines. effective.

[Brief explanation of the drawing]

[Figure 1] ~

FIG. 2 is a diagram for explaining the basic configuration of the present invention,

[Figure 3
] ~

FIG. 15 is a diagram for explaining the first embodiment according to the present invention,

[Figure 16] ~

FIG. 22 is a diagram for explaining a second embodiment according to the present invention,

[Figure 23] ~

FIG. 53 is a diagram for explaining a third embodiment according to the present invention,

[Figure 54] ~

FIG. 66 is a diagram for explaining a modification of the third embodiment.

[Explanation of symbols]

5 Pseudo human reaction generation unit 6 Stimulation control section 7 Adaptive control section 8 Non-adaptive control section 9 Human condition detection section 10 Sensory stimulus generation part 11 Subject (driver)

Claims (11)

[Claims] 1. A human-machine system equipped with an environment detection means for detecting an external environment, the human-machine system configured to detect the human-machine system so that a human feels that the human-machine system is pseudo-human in accordance with human sensibilities. A human machine system characterized by comprising a control means for controlling the functions of the system. 2. The control means includes a first detection means for detecting the human physiology or psychological state, and a first control means for controlling the function of the human-machine system so that the physiological or psychological state is comfortable. 2. The human machine system according to claim 1, further comprising means. 3. The control means includes a first detection means for detecting the human physiology or psychological state, and a second control means for controlling the functions of the human-machine system so that the physiological or psychological state becomes uncomfortable. 2. The human machine system according to claim 1, further comprising means. 4. The control means includes a first detection means for detecting the physiological or psychological state of the human being, and a first detecting means for controlling the functions of the human-machine system so that the physiological or psychological state becomes comfortable or unpleasant. 2. The human machine system according to claim 1, further comprising a control means of 3. 5. A claim characterized in that the control means includes an alertness level detection unit that detects the alertness level of the human being, and a fourth control unit that controls the functions of the human machine system according to the alertness level. The human machine system according to item 1. 6. The arousal level detection means includes a means for detecting brain waves, a means for calculating the power amount and power data of the detected brain waves, a means for presenting a stimulus and detecting a reaction time to the stimulus, and a means for detecting a brain wave. Claim 1, further comprising means for calculating an arousal level estimation parameter by analyzing the correlation between the amount of power and the reaction time, and means for calculating an arousal level from the arousal level estimation parameter and the power data. The human machine system according to item 5. [Claim 7] The arousal level detection means includes means for detecting brain waves, means for performing frequency analysis on time series data of the detected brain waves, and detecting a peak level on the frequency to calculate an arousal rhythm cycle. 7. The human machine system according to claim 6, further comprising means. 8. The arousal level detection means includes means for presenting a stimulus, means for detecting a reaction time to the presented stimulus, and means for performing frequency analysis on time series data of the reaction time.
7. The human machine system according to claim 6, further comprising means for detecting a peak of a level on a frequency and calculating an arousal rhythm cycle. . 9. The alertness level detection means includes a means for detecting blink frequency, a means for presenting a stimulus and detecting reaction time to the stimulus, and analyzing a correlation between blink frequency and reaction time to estimate alertness level. 6. The human machine system according to claim 5, further comprising means for calculating a parameter and means for calculating an arousal level from an arousal level estimation parameter and a blink frequency. 10. The alertness level detection means includes a means for detecting heart rate, a means for presenting a stimulus and detecting reaction time to the stimulus, and analyzing a correlation between heart rate and reaction time to estimate alertness level. 6. The human machine system according to claim 5, further comprising: means for calculating a parameter; and means for calculating an alertness level from an alertness level estimation parameter and a heart rate. 11. The control means includes: an arousal degree detection means for detecting the arousal degree of the human being; a measuring means for measuring the activity information of the human body; and a stimulus presentation for emitting an external stimulus to the human body in accordance with the activity information. 2. The human machine system according to claim 1, further comprising: means for controlling said stimulus presenting means in accordance with said degree of arousal to maintain said human being in an awake state.