JPH03204343A - Doze-driving detecting device - Google Patents

Abstract

PURPOSE:To precisely detect dozing at the wheel by comparing a first integrated angle with a second integrated angle, and conducting dozing judgment when the angles are nearly the same and the ratio of first half period time to second half period time is a determined comparison value or more. CONSTITUTION:In an integrated angle comparing means 50, an integrator 52 integrates signals from a displacement calculator 44 in half-period unit, an integration comparator 56 compares the present integrated value (second integrated value) with the integrated value of the latest half period, and outputs a coincidence signal when these values are coincident to each other. This coincidence signal means the stable running confirmation of a vehicle. In a half-period time comparing means 60, a time computing element 62 receives the signal from the calculator 44 and calculates the period in which displacement angle is moved from one polarity to the other polarity, and a time comparator 66 compares the ratio of first and second half period time values with a preliminarily set comparison value 68 and outputs a dozing detection signal when the ratio is larger. A judging apparatus 70 outputs an alarm signal when it receives the coincidence signal and detection signal at a determined vehicle speed or more.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a drowsy driving detection device, and particularly to a drowsy driving detection device that detects drowsiness when a vehicle is running stably on a highway or the like.

[Prior Art] In general, it is said that drowsy driving is particularly likely to occur under conditions such as long-distance driving or monotonous driving.Sometimes, this drowsy driving causes the driver himself to become aware of his drowsiness and take measures such as stopping. However, if the driver himself is not aware of this, in other words, if he is about to fall asleep, his driving ability will gradually decline without him noticing, resulting in a delay in judgment. This can lead to dangerous conditions such as a decrease in steering ability.

Therefore, there is a strong social need for a device that can detect not only drowsy driving, but also sluggish driving that the driver is not aware of, and provide necessary and appropriate warnings to the driver.

Conventional methods for detecting drowsy driving can be roughly divided into two types: first, methods that attach electrodes directly to the driver and observe the driver's behavior and level of consciousness; and second, methods that use various sensors placed in various parts of the vehicle to provide steering control. Detects the amount of change, vehicle speed, etc.
There is a method that makes a judgment based on the driving behavior of the vehicle.

The first method is a method that involves direct contact with the driver, so it is a heavy burden and is used only in limited cases, while the second method is currently attracting attention.

As an example of a device that implements such a second method, that is, a method for determining driving behavior, there is a first example device shown in FIG. 8, for example.

In the figure, 10 is a control circuit to which sensor outputs of a steering angle sensor 12, a vehicle speed sensor 14, a shift operation detection sensor 16, and a brake pedal operation detection sensor 18 are connected, and it determines drowsy driving based on this information. It is something. When the control circuit 10 determines that the driver is falling asleep, it sends a signal to the alarm 20, which wakes the driver up.
This will prevent drowsy driving.

Here, for example, to describe the procedure for determining drowsiness regarding vehicle speed and steering angle, first, a standard change pattern is created by calculating the average of the vehicle speed and steering angle, which serve as criteria for judgment, and then using this standard change pattern and the measurement time. Compare the changes in vehicle speed and steering angle.

If the comparison results do not match, it is determined that the driver is drowsy while driving. In other words, if the vehicle deviates from the reference change pattern for each case, it is determined that the vehicle is falling asleep (for example, see Japanese Patent Laid-Open No. 15230-1983).

Further, as a second example of the conventional drowsy driving detection device, the sensor section of the first example device shown in FIG.
One example is a device that adds various sensors to detect driving conditions, etc., and determines whether the vehicle has fallen asleep if the conditions inside and outside the vehicle do not change within a certain period of time, and if no driving operation is performed within that period (for example, Japanese Patent Application Laid-Open No. (See 1984-135738).

Therefore, the conventional devices shown in the first and second examples above detect drowsy driving by detecting driving that deviates from the standard pattern or non-operational driving within a certain period of time under predetermined conditions. Accuracy largely depends on the condition setting values that serve as the criteria for judgment.

[Problems to be Solved by the Invention] However, considering that the occurrence of falling asleep is affected by factors such as individual differences and physical condition in addition to surrounding environmental conditions, the above conventional method does not directly specify the conditions for determining the condition. Furthermore, even if the judgment conditions are varied depending on external circumstances, taking various aspects into consideration as a premise of the judgment criteria can be expected to improve accuracy to a certain extent, but the driver's behavior at the time of judgment It is difficult to understand accurately.

In addition, in the first conventional example, when driving around curves or changing lanes frequently at a constant speed (without using turn signals), the average value of the amount of steering change is used as the reference change pattern, so this average value In addition, in the second conventional example, when driving in a straight line with little steering, the drowsiness determination condition is easily met, resulting in false detection. Ta.

Purpose of the Invention The present invention has been made in view of the above-mentioned conventional problems,
The purpose is to provide a drowsy driving detection device that can accurately detect drowsy driving by confirming that the vehicle is running stably and eliminating false detection factors based on the peculiar phenomenon of steering observed during stable, straight-line driving. It is in.

[Means for Solving the Problems] In order to achieve the above object, the drowsy driving detection device according to the present invention includes a steering angle sensor that detects a steering angle, and a steering angle sensor that has one polarity continuous around a reference angle steering angle. a first half-cycle time calculation means for calculating a first half-cycle time which is time; and a first half-cycle time calculation means for calculating a first integrated angle by integrating the steering angle within the first half-cycle over time.
an integrated angle calculation means, a second half-period time calculation means for calculating a second half-period time that is a time during which the other polarity continues following the first half-period, and a second integrated angle within the second half-period. a second integrated angle calculation means for calculating the first integrated angle and the second integrated angle;
and determining means for comparing the integrated angles and determining a doze when they are substantially the same and when the ratio of the first half-cycle time and the second half-cycle time is greater than or equal to a predetermined comparison value. , is characterized in that it determines whether someone is dozing off based on the periodic characteristics of the steering angle.

[Function] Characteristics of Periodic Steering Angle Variation First, the phenomenon of periodic minute fluctuations in the steering angle that is observed when the vehicle is running in a straight line, which is a prerequisite for the present invention, will be described.

Generally, it is known that when a vehicle is running in a straight line or on a gentle curve, that is, when it is running stably, the phenomenon shown in FIG. has also been confirmed. Note that FIG. 6 is a graph in which the vertical axis represents the steering angle (de g) and the horizontal axis represents time t, and shows temporal fluctuations in the steering angle while the vehicle is running.

As shown in Fig. 6 (A), the steering fluctuates almost periodically, and the driver unconsciously repeats minute steering to the left and right in order to stabilize the vehicle (driving in a straight line). It is presumed that there are.

In addition, in the investigation by the present inventors, the time of each adjacent half cycle of such periodic fluctuations during stable running (for example, T1
, T2) are approximately the same, and furthermore, it has been confirmed that the steering angle integrated values (Sl, S2) within each half cycle are also approximately the same. If the driver is different, there will be a difference in the above-mentioned cycle and steering width, but as long as the vehicle is running stably, the above-mentioned phenomenon will generally occur.

On the other hand, Fig. 6(B) shows periodic fluctuations in the steering angle just before falling asleep while driving, and shows that the driver's consciousness fades during period ω, and at time t he realizes and operates the steering wheel ω. It is shown.

As shown in the figure, the integrated angles S1 and Sj almost match, but the half cycle times T1 and Tj do not match. In other words, as the driver's awareness diminished and the rhythm of steering wheel operations became disrupted, one half cycle became longer. Since the cumulative angle is considered to depend on the traveling direction of the vehicle, if the vehicle's traveling direction is the same before and after one cycle, the cumulative angle is considered to be almost the same before and after drowsy driving (Si −3j ) . Note that in this case, the integrated angle sh and Si do not match.

FIG. 7 is a graph based on the results of experiments conducted by the present inventors.

Here, the horizontal axis is the measurement elapsed time, and the vertical axis is the ratio α(
(To-1/Tn) within a certain period of time. As is clear from the figure, a somewhat vague state P2
Also, the half-period ratio α is increasing. In addition, P
is normal, and P3 is a case of dozing off.

Effect of the means according to the present invention Based on the above-mentioned periodic fluctuation characteristic, if the means according to the present invention described above is used, the steering angle is first detected by the steering angle sensor, and the first half-cycle time calculating means and the first integrated angle are detected. The calculation means can calculate the first half-cycle time and the integrated value within the first half-cycle.

Next, similarly to the first half cycle, a second half cycle time and a second integrated angle calculation means are used to calculate the second half cycle time following the first half cycle and the integrated angle within that half cycle. I can do it.

Then, the determining means determines that the vehicle is dozing when the cumulative angles in the two successive half cycles determined above match and the ratio of the half cycle times is greater than a predetermined value, thereby ensuring stable running of the vehicle. It is possible to detect drowsy driving after checking.

In other words, if the integrated angles do not match, factors such as a lane change or a crosswind may be considered, so it is possible to eliminate false detections caused by this and perform detection in accordance with the driving situation.

[Embodiments] Hereinafter, preferred embodiments of the present invention will be described based on the drawings.

FIG. 1 shows the overall configuration of a drowsy driving detection device according to the present invention.

This device detects drowsy driving based on the fluctuation characteristics of the steering angle when the vehicle is running stably as described above, and calculates the integrated angle in each half cycle and compares the integrated angle in each adjacent half cycle. and half-cycle time comparison means 60 for determining the time of each half-cycle and comparing the times of adjacent half-cycles.

In addition, as mentioned above, the periodic fluctuation characteristic of the steering angle is a phenomenon that can be seen when driving stably in a straight line on expressways, etc., but the inventors' actual measurements have shown that the same characteristic appears even on gentle curves. This has been confirmed, and in order to deal with this, displacement calculation means 40 is provided in this embodiment.

This displacement calculation means 40 calculates the average value of the steering angle within a certain period of time and calculates the current steering angle displacement using this average value as a reference.These means and their operation will be described in detail below.

30 in the figure is a steering angle sensor provided in the steering system of the vehicle, which outputs a predetermined signal according to the steering angle to the displacement calculation means 40 of the next stage.

This displacement calculation means 40 includes an average steering angle calculator 42 that calculates the average steering angle within a predetermined time and a displacement calculator 44 that calculates the displacement (displacement angle) of the current steering angle based on the average steering angle. It is configured.

The average steering angle calculator 42 inputs the signal from the sensor 30, calculates the average value within a predetermined time, and outputs it sequentially. Even if the steering is turned to the right or left, the average value can be used as a reference for the steering cycle fluctuation.

That is, by using the average of the steering angles as a reference point in this way, even if the periodic fluctuations in the steering angles do not necessarily vary around 0 degrees, it is possible to detect cycles and, in turn, to detect dozing off. In addition, if I may add,
When the curve curvature is large, periodic fluctuations are less likely to appear, but at the same time it is assumed that the probability of falling asleep is also low.

The displacement calculator 44 receives the output of the average steering angle calculator 42 and the output of the sensor 30, calculates a displacement angle by subtracting the average steering angle from the current steering angle, and outputs a predetermined signal representing the displacement angle. It is outputting.

Therefore, as described above, according to this displacement calculation means 40,
[1] It is possible to extend the detection tolerance range from straight lines where dozing is likely to occur to gentle curves.

Next, the integrated angle comparing means 50 includes an integrator 52 that calculates the integrated steering angle in a half cycle, a storage unit 54 that temporarily stores the integrated value, and an integrated comparator 56 that compares the integrated values of successive half cycles. It consists of

The integrator 52 integrates the signal from the displacement calculator 44 in half-cycle units, and then sends the integrated value to the storage unit 54.

Here, the storage unit 54 temporarily stores the integrated value.

Then, in the next half cycle, this process is repeated in the same manner as described above, but in parallel with this, when the integration of the half cycle is completed, the stored 1
The integrated value of the previous half cycle (first integrated value) and the current integrated value (second integrated value) from the integrator 52 are sent to the integrated comparator 56.
is output to.

The integration comparator 56 compares the first and second integration values,
A match signal is output only when the values match. That is, what this coincidence signal means is to confirm that the vehicle is running stably, as described above.] 2 When the vehicle is driving around a sharp curve or changing lanes, the integrated values do not match and a coincidence signal cannot be obtained.

Therefore, by comparing this integrated value, it is possible to clearly distinguish the cause of false detection of falling asleep. Note that in this embodiment, since the integrated values in each half cycle are affected by various factors and do not always match, the range in which they are considered to be a match is slightly expanded to make an appropriate judgment.

Next, the half-cycle time comparing means 60 includes a time calculator 62 that calculates the time of each half-cycle, a storage unit 64 that temporarily stores the half-cycle time value, and a ratio of the times of successive half-cycles that calculates the ratio. It is comprised of a time comparator 66 that compares with a predetermined comparison value 68.

The time calculator 62 receives the output signal of the displacement calculator 44,
The period during which the displacement angle shifts from one polarity to the other, that is, the half cycle time, is calculated and the time value is output to the storage section 64. Here, the storage unit 64 temporarily stores the time value.

Then, in the next half cycle, the same process as above is repeated, but when the half cycle time is calculated in parallel with this operation, the stored 1
The time value of the previous half cycle (first half cycle time value) and the current time value (second half cycle time value) are sent from time calculator 62 to time comparator 66, respectively.

The time comparator 66 calculates the ratio α (first half-cycle time value/second half-cycle time value) of the first and second half-cycle time values,
Compared with a preset comparison value β68, if the α value is larger than the β value, that is, if sudden steering is performed or the steering rhythm changes, it is considered to be unnatural driving,
Outputs a dozing detection signal. Note that in this embodiment, the β value is set to 1 and 6, which were obtained through experiments and are capable of determining a somewhat vague state, and a potentially dangerous state of the driver can also be detected. Further, this β value may be modified as appropriate according to various conditions.

Therefore, as described above, the half-cycle time comparing means 60 can detect unnatural steering when the vehicle is running. Although this detection includes steering when changing lanes and steering due to a sudden crosswind, which is not necessarily drowsy driving, this false detection can be eliminated by the determination device 70 described below.

The match signal output from the integrated angle comparison means 50, the detection signal output from the half-cycle time comparison means 60, and the vehicle speed signal from the vehicle speed sensor 72 in this embodiment are input to the determiner 70, respectively.

Then, the determiner 70 outputs an alarm signal when the vehicle speed is equal to or higher than a predetermined speed and both the coincidence signal and the detection signal are obtained.

Here, in this example, the determination was made based on the assumption that the vehicle speed was equal to or higher than a predetermined vehicle speed, because firstly, it is difficult to obtain periodic fluctuation characteristics of the steering angle on roads such as urban areas, and secondly, the target vehicle speed was monotonous expressways. This is due to the fact that warnings are particularly requested in situations where drowsy driving is likely to occur.

However, with this determiner 70, it is possible to determine whether the driver is falling asleep while the vehicle is running stably, after excluding the factors of erroneous detection. Then, by inputting the output signal to the alarm device 80 and taking necessary warning measures, it is possible to make the driver aware not only when he or she is drowsy while driving, but also when the driver is just about to do so. This makes it possible to prevent dangerous situations from occurring.

Next, as a second embodiment, a processing procedure will be explained based on the flowchart shown in FIG. 2 in a case where a central processing unit including the necessary memory performs the same processing as in the above device.

First, this process is started when the vehicle starts operating or is started manually (Sl).

Next, in S2, it is determined whether the vehicle speed is equal to or higher than a predetermined speed, and if this condition is met, the process moves to the next step S3. In addition,
If the conditions do not match at this time, the process moves to S17, and the process from S1 is repeated again.

Next, in S3, the hand chain θ of the steering angle for the past X seconds is calculated. Then, in S4, the signal θ from the sensor is changed to θ
is subtracted to calculate dθ.

In S5, it is determined whether this dθ is 0 or not. In other words, d
The half-cycle zero-crossing point can be detected by θ-0.

If the conditions do not match here, the process moves to the next step 86. In S6, the dθ is integrated and substituted into Σ. In other words, by repeating the process of S6, d
The sum Σ (integrated angle) of θ increases.

Next, in S7, counting is performed. Then, each time the S7 process is repeated, the CN value increases by 1,
The half cycle time (-CN) is thereby calculated.

When the processes 82 to S7 up to this point are completed, the process moves to 817 and continues again from Sl.

If the conditions match in the judgment of S5, first, S8 to S
In ll, the values assigned to the variables are replaced with different variables, and as a result, the previous half-cycle time value is CN2, its integrated angle is Σ2, and the current half-cycle time value is C
NI and the integrated angle at this time are respectively substituted into Σ1. That is, the processing from 88 to Sll is performed for each zero point of a half cycle, and variable values are replaced for the following processing.

In S12, CN (half cycle time value) and Σ (integrated angle) are reset. S1. , 3, the sum of Σ1 and Σ2 is taken and substituted into Q. Note that if the integrated angles match, it is Q-0.

In S14, it is determined whether Q is 0 or not. And Q
If they match, that is, in the case of stable running, the following three
The process proceeds to S15, and if the answer is NO, the process moves to S17 and the process is repeated again from sl.

In S15, the value obtained by dividing CN2 from CNI is the predetermined value β.
It is determined whether the value is greater than or not. At this time, if the conditions match, that is, if dozing is detected, a process for determining dozing, for example, an alarm is performed in S16, then the process moves to 317, and the process is repeated from 81 again.

Therefore, as described above, according to the second embodiment, similarly to the first embodiment, it is possible to detect drowsy driving.

What is distinctive about this process is that the timer function is performed by a counter, and when the displacement angle reaches zero (dθ-0), the process is greatly branched and calculations are performed. The basic dozing detection principle is the same.

Next, a description will be given of a drowsy driving detection device equipped with a comparison value correction calculating means as a third embodiment.

As is well known, drowsy driving is greatly affected by various factors such as driver fatigue, driving time, and temperature inside the car. Similarly, periodic fluctuations in steering angle also depend on these factors. It is thought that

In other words, the α value changes slightly depending on these conditions.

Therefore, in order to perform a more appropriate dozing determination, it is necessary to slightly correct the comparison value β based on the above conditions. Therefore,
As such a correction means, there is a comparison value correction calculation means shown in FIG.

Reference numeral 90 in the figure is a correction coefficient calculating section, to which signals from a timer 92, a temperature sensor 94, and a clock 96 are input. The timer 92 is activated at the start of operation and outputs a driving time signal indicating the degree of driving fatigue. Further, the temperature sensor 94 monitors the temperature inside the vehicle and outputs a temperature signal indicating the comfort level. Furthermore, the clock 96 outputs a signal indicating the current time, ie, the degree of sleepiness during the day or night.

9 As a method for correlating each of these pieces of information with drowsy driving, this embodiment uses a non-Neumann type so-called fuzzy inference, and the correction coefficient calculation unit 90 performs calculations based on this.

As is well known, this fuzzy inference has the advantage of being able to process ambiguous quantities using membership functions (degree of belonging functions) and inference rules, and is particularly useful for the apparatus according to the present invention. It is a processing method. For example, it can handle nonlinear functions and perform multivariate analysis.

The membership function and inference rule are stored in the function storage section 98.

Figure 4 is an example of such a membership function,
The driver's suitability for each condition is shown.

Here, FIG. 4 (A) is the degree of conformity to the driver's driving time, (B) is the degree of conformity to the temperature inside the vehicle, (C) is the degree of conformity to the driving time, and (D) is a function for weighting the inference results. are shown respectively.

Further, the items (labels) of each rule are given as A0 1 to A3, B1 to B3, C1, C2, and D1 to D3, respectively, and a predetermined fuzzy control law (rule) is determined thereby.

Next, the specific processing procedure will be explained based on FIG.

First, in 5101, each piece of information is taken in. In 5102, each piece of information is converted into each membership function (A), (B).
, (C), and then weights the membership functions shown in (D) based on the inference rules. Then, the centroid method is used to find the centroid of this membership function, that is, the correction coefficient ε. Note that to obtain the correction coefficient ε, the method is not limited to the center of gravity method, but may be appropriately selected and used such as the height method.

At 8103, the correction coefficient determined at 5102 is output to the multiplier 99. By repeating the above process, it is possible to successively obtain an appropriate correction coefficient ε that matches each piece of information. Although the membership functions in FIG. 4 are set based on the knowledge base obtained through experiments in this embodiment, they are not limited to this and may include nonlinear quantities. Further, requirements such as weather and humidity may be added to the conditions.

The correction coefficient ε obtained by the above process is then multiplied by ε by the preset value β in the multiplier 99 to obtain the corrected comparison value γ, and the time comparison shown in FIG. 1 is obtained. 66.

Therefore, according to the dozing detection device equipped with this comparator correction calculation means, it is possible to use the appropriate comparison value γ according to the inside and outside conditions of the vehicle as the comparison value that is the standard for dozing off, thereby further improving the dozing off judgment accuracy. can be improved. This has the advantageous effect that it is possible to accurately detect even a somewhat vague state that the driver is not aware of, and to provide a more appropriate warning.

[Effects of the Invention] As explained above, according to the drowsy driving detection device according to the present invention, drowsiness while the vehicle is running stably can be detected accurately and accurately.

In particular, since false detections can be removed by comparing the integrated angle, reliability is high.Also, by setting the comparison value of the half-cycle time ratio to an appropriate value, it is possible to detect even a state where the driver is slightly drowsy. It is possible to accurately inform the user of his or her state before falling asleep.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the configuration of the drowsy driving detection device according to the present invention, FIG. 2 is a drowsiness determination flowchart according to the second embodiment, and FIG. 3 is the configuration of the comparison value correction calculation means according to the third embodiment. Fig. 4 is an explanatory diagram of the membership function, Fig. 5 is a flowchart of the correction coefficient calculation process, Fig. 6 is an explanatory diagram showing periodic fluctuations in the steering angle, and Fig. 7 is an explanatory diagram of the membership function within a certain period of time. FIG. 8, a graph showing changes in the maximum value of α, is a configuration diagram of a conventional drowsy driving detection device. 30 ... Steering angle sensor 40 ... Displacement calculation means 50 ... Integration comparison means 60 ... Half cycle time comparison means 3 70 ... Judgment device 72 ... Vehicle speed sensor 90 ... Correction coefficient Arithmetic unit 4

Claims (1)

[Scope of Claims] A steering angle sensor that detects a steering angle; a first half-cycle time calculation means that calculates a first half-cycle time that is a time during which one polarity continues around a reference steering angle; and the first half-cycle time. a first integrated angle calculation means that calculates a first integrated angle by integrating the steering angle within a half cycle over time; and a second integrated angle calculation means that calculates a second half cycle time that is a time during which the other polarity continues after the first half cycle. a half-cycle time calculation means; a second integrated angle calculation means for calculating a second integrated angle within the second half cycle; and compares the first integrated angle and the second integrated angle and determines that the integrated angle is substantially the same, and , a determination means for determining dozing off when the ratio of the first half-cycle time and the second half-cycle time is equal to or greater than a predetermined value; Features of drowsy driving detection device.