JP4038571B2 - Dangerous state detection method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a dangerous state detection method and apparatus for detecting a dangerous state in a work environment .
[0002]
[Prior art]
In work environments that contain danger, such as construction sites, roads, and welfare facilities, not only cases where accidents actually occurred, but also cases where workers felt danger even if they did not result in an accident (so-called “ “Hidden / Hatto” cases) are accumulated and analyzed. Conventionally, such cases have been accumulated based on reports from workers who actually experienced dangerous situations.
[0003]
[Problems to be solved by the invention]
However, even though it is said that there are tens of times the actual number of accidents in the “Hidden / Hatto” scene, there are many cases where notification is not imposed on the workers simply because they feel a danger. For this reason, it is difficult to gather the number of reports necessary for analysis.
[0004]
In addition, since there is a tendency for notification from the worker to be delayed unless there is any actual harm, there is also a problem that it is difficult to reproduce an accurate situation because the worker's memory fades with time.
[0005]
The present invention has been made to solve such a problem, and a dangerous state detection method capable of improving the safety of a work environment by accurately detecting a state in which a worker feels danger, and The purpose is to provide a device .
[0006]
[Means for Solving the Problems]
The object of the present invention is a method for detecting a dangerous state in a work environment, the step of measuring the worker's GSR and the worker's heart rate or pulse rate as physiological information, and the physiological information A step of determining whether or not a dangerous state has occurred at a determination target time based on a series change, wherein the step of determining whether or not a dangerous state has occurred includes a GSR after a predetermined time t1 (sec) has elapsed from the determination target time; The predetermined electric conductance c1 (μ mho ) is greater than the GSR value at the determination target time. Or higher, and the instantaneous heart rate during the minimum value of the instantaneous heart rate or the instantaneous pulse rate, the determination target time predetermined time after t3 (sec) between the determination target time before the predetermined time t2 (sec) Alternatively, it is achieved by a dangerous state detection method comprising a step of determining that a dangerous state has occurred at the determination target time when the instantaneous pulse rate is greater than the minimum value .
[0008]
In this case, the predetermined time t1 is preferably 2.5 to 20 (sec), the predetermined electric conductance c1 is preferably 0.02 to 1.3 (μmho), and the predetermined time t2 Is preferably 3 to 66 (sec), and the predetermined time t3 is preferably 6 to 66 (sec).
[0009]
In particular, in order to obtain a detection sensitivity of 99% or more, the predetermined time t1 is more preferably 3.5 to 17.5 (sec), and the predetermined electric conductance c1 is 0.02 to 0.14. (μmho) is more preferable, the predetermined time t2 is more preferably 3 to 24 (sec), and the predetermined time t3 is more preferably 21 to 66 (sec). Here, the reason for considering the detection sensitivity as an evaluation index in determining the parameter value range is that this algorithm avoids the fact that a non-hazardous scene cannot be detected rather than a non-hazardous scene being falsely detected. This is because it is considered a priority.
[0010]
In the above-described dangerous state detection method, at least a part of the visual field of the worker is continuously captured to generate image information, and the determination time corresponding to the determination target time at which it is determined that a dangerous state has occurred. The method may further comprise specifying image information.
[0011]
Further, the object of the present invention is an apparatus for detecting a dangerous state in a work environment, the measuring means for measuring the worker's GSR and the worker's heart rate or pulse rate as physiological information, and the physiological information. Storage means for storing, and dangerous state determination means for determining whether or not a dangerous state has occurred at a determination target time based on a time series change of the physiological information, wherein the dangerous state determination means is predetermined from the determination target time. The GSR value after elapse of time t1 (sec) is greater than the GSR value at the determination target time by a predetermined electric conductance c1 (μ mho ). Or higher, and the instantaneous heart rate during the minimum value of the instantaneous heart rate or the instantaneous pulse rate, the determination target time predetermined time after t3 (sec) between the determination target time before the predetermined time t2 (sec) Alternatively, it is achieved by a dangerous state detection device that determines that a dangerous state has occurred at the determination target time when the instantaneous pulse rate is greater than the minimum value.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
First, a dangerous state detection method according to an embodiment of the present invention will be described. The present inventors have examined information effective for detecting a dangerous state among various physiological information obtained from the worker. As a result, the operator's GSR (galvanic skin response) and the operator It has been found through experiments that a dangerous state can be detected with high accuracy by focusing on the heart rate (or pulse rate).
[0013]
GSR is a change in skin electrical conductance associated with sweating when a worker receives a strong stimulus. When the GSR values before and after giving a stimulus such as a gunshot to the subject were measured, it was found that, for example, as shown in FIG. Here, the GSR value is skin electrical conductance and is expressed in units of μmho.
[0014]
As described above, since the GSR value is considered to be physiological information that is effective for detecting a dangerous state such as “Hidden / Hatto”, the present inventors have a total of four surfaces including the front surface, the left and right surfaces, and the floor surface. An event where a virtual construction work site is reproduced using a large VR (virtual reality) device capable of projecting a stereo image, and a subject such as a steel drop falls in this virtual work space. Was generated, and the subject's subjectivity was reported as “satisfactory / happy”, and the detection accuracy of the dangerous state based on the change in the GSR value of the subject was examined. The result is shown in FIG.
[0015]
FIG. 2 shows a case where it is determined that a dangerous state has occurred when an event occurs when the GSR value after t1 (sec) has elapsed from the occurrence of the event becomes higher than c1 (μmho) than the GSR value at the time of the event. (A) Sensitivity and (b) Specificity are shown in a bird's-eye view, with the horizontal axis being t1 (in increments of 3 seconds) and the vertical axis being c1 (in steps of 0.02 μmho). Here, the sensitivity is the ratio of the number of detections to the number of event occurrences, and the higher the sensitivity, the higher the detection rate at the time of event occurrence. The specificity is the ratio of the non-occurrence non-detection number to the number of tests (the number of detections that did not occur when no event occurred). The higher the specificity, the lower the false detection rate when no event occurs. ing.
[0016]
As shown in FIG. 2A, the sensitivity is almost 0 when t1 is 0, but increases as t1 increases, and there is a time lag from when the stimulus is actually received until the GSR value changes. You can see what is happening. On the other hand, as c1 increases, the sensitivity decreases even if t1 is the same. Thus, from the viewpoint of improving the sensitivity, it is preferable to increase t1 and decrease c1.
[0017]
On the other hand, as shown in FIG. 2B, it can be seen that as c1 approaches 0, the specificity rapidly decreases and the false detection rate increases. Also, when t1 is increased, the specificity tends to decrease. Thus, from the viewpoint of improving the specificity, it is preferable to decrease t1 and increase c1.
[0018]
From the above, it can be seen that in order to maintain both sensitivity and specificity satisfactorily, each of t1 and c1 should be set within a predetermined range. From the experimental results shown in FIG. 2, in order to achieve a detection sensitivity of 70% or more for t1, it is preferable to set it in the range of 2.5 to 20 seconds, and in order to achieve a detection sensitivity of 80% or more. , 3 to 20 seconds is more preferable, and in order to achieve a detection sensitivity of 99% or more, it is more preferable to set the range to 3.5 to 17.5 seconds. Further, c1 is preferably set in the range of 0.02 to 1.3 μmho in order to achieve a detection sensitivity of 70% or more, and 0. It is more preferable to set in the range of 02 to 0.7 μmho, and in order to achieve a detection sensitivity of 99% or more, it is more preferable to set in the range of 0.02 to 0.14 μmho. Here, the reason why the detection sensitivity is considered as an evaluation index in determining the parameter value range is that it is considered that the first priority is to avoid that this algorithm cannot detect a dangerous scene.
In addition, when the present inventors measured the instantaneous heart rate before and after applying a stimulus such as a gunshot to the subject, for example, as shown in FIG. It was found to be lower than before the outbreak. The instantaneous heart rate is a value obtained for each heartbeat of the subject.
[0019]
Thus, since the instantaneous heart rate is considered to be physiological information that is effective for detecting a dangerous state such as “Hidden / Hatto”, the change in the heart rate of the subject using the VR device in the same manner as described above. The accuracy of detection of dangerous conditions based on the above was investigated. The result is shown in FIG.
[0020]
FIG. 4 shows a dangerous state when an event occurs when the minimum value of the instantaneous heart rate during t2 (sec) before the event occurs is larger than the minimum value of the instantaneous heart rate during t3 (sec) after the event occurs. (A) Sensitivity and (b) Specificity are shown in a bird's eye view, where the horizontal axis is t2 (in 3 sec increments) and the vertical axis is t3 (in 3 sec increments) Yes. The range where the instantaneous heart rate is 30 or less is excluded as an artifact.
[0021]
As shown in FIG. 4A, from the viewpoint of improving sensitivity, it is preferable to decrease t2 and increase t3. On the other hand, as shown in FIG. 4B, from the viewpoint of improving the specificity, it is preferable to increase t2 and decrease t3. Therefore, in order to maintain both sensitivity and specificity satisfactorily, it can be seen that t2 and t3 may be set within predetermined ranges, respectively. From the experimental results shown in FIG. 4, t2 is preferably set in the range of 3 to 66 sec in order to achieve a detection sensitivity of 80% or more, and 3 in order to achieve a detection sensitivity of 90% or more. More preferably, it is set in the range of -57 sec, and more preferably in the range of 3-24 sec in order to achieve a detection sensitivity of 99% or more. Also, t3 is preferably set in the range of 6 to 66 seconds in order to achieve a detection sensitivity of 90% or more, and in the range of 21 to 66 seconds in order to achieve a detection sensitivity of 99% or more. More preferably, it is set. Here, the reason why the detection sensitivity is considered as an evaluation index in determining the parameter value range is that it is considered that the first priority is to avoid that this algorithm cannot detect a dangerous scene.
The present inventors have confirmed through experiments that the above-described examination of the instantaneous heart rate can provide the same result with respect to the instantaneous pulse rate. As for the instantaneous pulse rate, a value obtained for each beat can be used.
[0022]
Thus, the GSR value and the instantaneous heart rate (or instantaneous pulse rate) are both physiological information effective for detecting a dangerous state, but are qualitatively independent information between the two. Therefore, by determining that a dangerous state has occurred at the determination target time when both the GSR value and the instantaneous heart rate (or instantaneous pulse rate) satisfy the above-described conditions for an arbitrary determination target time, It is possible to improve the detection accuracy of occurrence of a dangerous state.
[0023]
Next, a dangerous state recording apparatus according to an embodiment of the present invention will be described. FIG. 5 is a block diagram showing a configuration of a dangerous state recording apparatus according to an embodiment of the present invention. As shown in FIG. 1, the dangerous state recording device 1 includes a physiological information measuring device 10, a determination processing device 20, an imaging device 30, and an image recording device 40.
[0024]
The physiological information measuring apparatus 10 is configured in a glove shape as shown in FIG. 6, and is provided with GSR deriving electrodes 12 and 14 at positions corresponding to the index and middle fingers when put on the left hand, and corresponds to a ring finger. A pulse derivation electrode 16 is provided at a position where the pulse is derived. The physiological information measuring device 10 is connected to the determination processing device 20.
[0025]
The determination processing device 20 includes a physiological information storage unit 22 and a dangerous state determination unit 24, and is configured to be attachable to the wrist of the left hand as shown in FIG. The physiological information storage unit 22 stores physiological information about the GSR and the pulse rate input from the physiological information measuring apparatus 10. The dangerous state determination unit 24 determines whether or not a dangerous state has occurred at a determination target time according to a predetermined determination condition based on a time-series change in physiological information stored in the physiological information storage unit 22, and generates a dangerous state. Is output to the imaging apparatus 30.
[0026]
The imaging device 30 includes an imaging unit 32, an image signal processing unit 34, an image information storage unit 36, and a dangerous state specifying unit 38, and is configured in a helmet shape as shown in FIG. ing. The imaging unit 32 is, for example, a CCD camera, and images at least a part of the worker's field of view. The image signal processing unit 34 converts the image signal input from the imaging unit 32 into digital information, and stores the generated image information in the image information storage unit 36. The dangerous state identification unit 38 extracts image information corresponding to the determination target time input from the dangerous state determination unit 24 from the image information storage unit 36 and outputs the image information to the image recording device 40.
[0027]
The image recording device 40 is configured to be worn on the right wrist and stores image information input from the imaging unit 30.
[0028]
In the dangerous state recording device 1 having the above configuration, the GSR value and the pulse rate of the worker are constantly measured by the physiological information measuring device 10 and stored as physiological information in the physiological information storage unit 22. In addition, the visual field of the worker is always captured by the imaging device 30 and stored in the image information storage unit 36 as image information.
[0029]
The dangerous state determination unit 24 starts the determination of the dangerous state based on the stored physiological information after a predetermined time has elapsed since the physiological information measuring device 10 started measurement. Specifically, first, a determination target time that is a determination target of whether or not a dangerous state has occurred is set, and the GSR value after elapse of t1 (sec) from the determination target time is the GSR value of the determination target time. More than c1 (μmho) and the minimum value of the instantaneous pulse rate during t2 (sec) before the determination target time is larger than the minimum value of the instantaneous pulse rate during t3 (sec) after the determination target time In this case, it is determined that a dangerous state has occurred at the determination target time. The values of t1, c1, t2, and t3 can be determined as appropriate within the above-described range, but in the present embodiment, t1 = 4.5, c1 = 0.12, t2 = 9, and t3 = 66. . The dangerous state determination unit 24 repeatedly performs the above-described determination of the dangerous state while changing the determination target time by a predetermined time (for example, 3 seconds), and captures each determination target time at which it is determined that the dangerous state has occurred. Output to.
[0030]
The imaging unit 30 extracts image information corresponding to each determination target time (for example, moving image information for about 3 seconds before and after the determination target time) from the image information storage unit 36 and outputs the image information to the image recording device 40. In this way, the image recording device 40 stores only image information corresponding to the state in which the worker felt dangerous. Therefore, by analyzing this image information, it is possible to easily identify a dangerous situation or place at the work site. Therefore, it is possible to improve the safety of the work environment.
[0031]
As mentioned above, although one Embodiment of this invention was explained in full detail, the specific aspect of this invention is not limited to the said embodiment. For example, the dangerous state recording apparatus of the present embodiment determines the dangerous state based on the worker's GSR value and the instantaneous pulse rate, but based on the worker's GSR value and the instantaneous heart rate, the present embodiment It is also possible to determine a dangerous state in the same manner as described above.
[0032]
Further, in the dangerous state recording apparatus of the present embodiment, the physiological information measuring apparatus 10 has a glove shape and the determination processing apparatus 20 can be attached to the wrist. However, the configuration is particularly limited as long as workability is not impaired. Instead, for example, the physiological information measuring device 10 may be configured as a sock, and the electrodes 12, 14, and 16 may be attached to the toes and the determination processing device 20 may be attached to the ankle.
[0033]
Further, as in the present embodiment, instead of extracting image information corresponding to the determination target time determined to be in a dangerous state, by providing an informing means that informs by voice immediately after determining the dangerous state, the operator can You may enable it to recognize a dangerous state on the spot.
[0034]
In the present embodiment, the dangerous state determination unit 24 determines the dangerous state in parallel with the measurement by the physiological information measuring device 10, but after the measurement by the physiological information measuring device 10 is completed, the dangerous state is determined. The state determination may be started.
[0035]
In the present embodiment, the dangerous state specifying unit 38 extracts image information corresponding to the determination target time when it is determined that a dangerous state has occurred, and stores the extracted image information in the image recording device 40. If it is possible to specify image information corresponding to the target time, another method may be used. For example, the dangerous state specifying unit 38 visually or visually analyzes image information corresponding to a predetermined time before and after the determination target time when it is determined that a dangerous state has occurred, among the image information stored in the image information storage unit 36. It is also possible to perform a process that can be identified by voice.
[0036]
In the present embodiment, the imaging device 30 stores all image information obtained by imaging in the image information storage unit 36. This image information storage unit 36 is old after a predetermined time has elapsed. It is also possible to use an image buffer or the like in which image information can be deleted and new image information can be written.
[0037]
【Example】
A dangerous scene was generated at the virtual construction work site using the VR device described above, and the detection accuracy of the “satisfying / rapid” state was confirmed for 21 subjects (18 to 29 years old). . As a detection algorithm, by measuring the subject's GSR and heart rate, the GSR value after 4.5 seconds from the determination target time is 0.12 μmho or more higher than the GSR value of the determination target time, and When the minimum value of the instantaneous heart rate for 9 seconds before the determination target time is larger than the minimum value of the instantaneous heart rate for 66 seconds after the determination target time, it was determined that a dangerous state occurred at the determination target time.
[0038]
As a result, sensitivity (sensitivity) = 1 (100%) and specificity = 0.90 (90%) in the resting (static) state, and sensitivity in the bicycle rowing (dynamic) state. = 1 (100%), specificity = 0.87 (87%), and good detection results were obtained for both sensitivity and specificity in both static and dynamic states.
[0039]
【The invention's effect】
As is apparent from the above description, according to the present invention, there is provided a dangerous state detection method and apparatus capable of improving the safety of the work environment by accurately detecting the state in which the worker feels dangerous. be able to.
[Brief description of the drawings]
FIG. 1 is a diagram illustrating an example of a time-series change of a GSR value.
FIG. 2 is a diagram showing (a) sensitivity and (b) specificity when determining the occurrence of a dangerous state based on a GSR value.
FIG. 3 is a diagram showing an example of a time-series change in instantaneous heart rate.
FIG. 4 is a diagram showing (a) sensitivity and (b) specificity when determining the occurrence of a dangerous state based on the instantaneous heart rate.
FIG. 5 is a block diagram showing a configuration of a dangerous state recording apparatus according to an embodiment of the present invention.
FIG. 6 is a schematic configuration diagram of a main part of the dangerous state recording apparatus.
FIG. 7 is a schematic configuration diagram of another main part of the dangerous state recording apparatus.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Danger state recording device 10 Physiological information measuring device 12, 14 GSR deriving electrode 16 Pulse deriving electrode 20 Judgment processing device 22 Physiological information storage unit 24 Danger state judging unit 30 Imaging device 32 Imaging unit 34 Image signal processing unit 36 Image information Storage unit 38 Hazardous state specifying unit 40 Image recording device

Claims (5)

A method for detecting a dangerous state in a work environment,
Measuring the worker's GSR and the worker's heart rate or pulse rate as physiological information; and
Determining whether or not a dangerous state has occurred at a determination target time based on a time-series change of the physiological information,
The step of determining whether or not the dangerous state has occurred includes
The GSR value after elapse of a predetermined time t1 (sec) from the determination target time is higher than the GSR value at the determination target time by a predetermined electric conductance c1 (μ mho ). Or higher, and the instantaneous heart rate during the minimum value of the instantaneous heart rate or the instantaneous pulse rate, the determination target time predetermined time after t3 (sec) between the determination target time before the predetermined time t2 (sec) Alternatively, a dangerous state detection method comprising a step of determining that a dangerous state has occurred at the determination target time when the instantaneous pulse rate is greater than the minimum value . The predetermined time t1 is 2.5 to 20 (sec),
The predetermined electrical conductance c1 is 0.02 to 1.3 (μmho), the predetermined time t2 is 3 to 66 (sec),
The dangerous state detection method according to claim 1 , wherein the predetermined time t3 is 6 to 66 (sec). Continuously imaging at least a portion of the worker's field of view to generate image information; and
The dangerous state detection method according to claim 1 , further comprising a step of specifying the image information corresponding to the determination target time when it is determined that a dangerous state has occurred. A device for detecting a dangerous state in a work environment,
Measuring means for measuring the worker's GSR and the worker's heart rate or pulse rate as physiological information,
Storage means for storing the physiological information; and
Based on the time series change of the physiological information, comprising a dangerous state determination means for determining the presence or absence of a dangerous state at the determination target time,
The dangerous state determination means is a predetermined time t1 from the determination target time. (sec) Is greater than the GSR value at the determination target time by a predetermined electric conductance c1 (μ mho ) The predetermined time t2 that is higher than the above and before the determination target time (sec) Between the instantaneous heart rate or the minimum value of the instantaneous pulse rate is a predetermined time t3 after the determination target time. (sec) A dangerous state detection device that determines that a dangerous state has occurred at the determination target time when the instantaneous heart rate is greater than the minimum value of the instantaneous heart rate or the instantaneous pulse rate. Imaging means for continuously imaging at least a part of the visual field of the worker and generating image information; and
The dangerous state detection device according to claim 4, further comprising a dangerous state identification unit that identifies the image information corresponding to the determination target time when it is determined that a dangerous state has occurred by the dangerous state determination unit.

Images