JP4663503B2 - Fall determination method and apparatus - Google Patents

Description

  The present invention relates to a fall judging method and apparatus for judging fall.

  In recent years, the number of elderly people living alone or living in welfare facilities for the elderly has been increasing year by year. The fall of the elderly is life-threatening if left unattended and requires early detection and emergency contact. For this reason, there is a need for a system that reliably detects and reports the fall of the elderly. Such a system includes a detection unit that detects at least one of tilt, vibration, and impact, and at least one of tilt, vibration, and impact that is equal to or higher than a predetermined level based on the detection output of the detection unit. A device that determines a fall when one of them is detected is known (Patent Document 1).

  When judging fall, when detecting only tilt, vibration, and shock as an index, for example, when a hand is touched or falls and the shock is alleviated, or an elderly person gets up In many cases, an erroneous determination is made when lying down. Further, if the threshold value used for determination is set high, a case where the fall cannot be determined occurs, and if the threshold value is set low, a case where the fall is not detected may occur.

  As an alternative to the above method, the present inventors have determined that the subject's trunk is based on sensors that detect accelerations in three axial directions and angular velocities around the three axes that are attached to the subject's trunk. A new method for calculating the vertical displacement and determining the subject's fall based on the calculated vertical displacement of the trunk of the subject is being studied (for example, JP-A-2005-237576). When calculating the vertical displacement of the trunk of the subject based on the sensor that detects the acceleration in the three-axis direction and the angular velocity around the three axes, for example, the vertical acceleration of the trunk is obtained, and this is integrated twice over time. By doing so, the vertical displacement of the trunk of the subject can be calculated.

According to this method, theoretically, the absolute displacement of the subject in the vertical direction can be calculated, and a fall is determined based on this. Therefore, it can be expected that erroneous determination can be reduced even when the impact is alleviated, or when an elderly person gets up or lies down.
JP-A-9-305875 JP 2005-237576 A

  However, as a result of the experiment, the result of calculating the vertical displacement of the subject's trunk based on the sensor that detects the acceleration in the three-axis direction and the angular velocity around the three axes is the result of the fall event that occurred in the subject. Depending on the way of falling, there may be results that deviate from the actual falling event. For example, when the subject falls and tumbles while tangling several steps, the calculation to calculate the vertical displacement by capturing the impact when first crawling or tangling several steps It may be activated in a superimposed manner. In such a case, the calculated vertical displacement may be calculated to be smaller than the actual value, and even if the subject actually falls, there may be cases where it is not determined that the fall has occurred. Therefore, in order to improve the accuracy of the fall determination, there is room for improvement in how the fall event is captured and how the vertical displacement is calculated.

  The fall determination method according to the present invention is based on triaxial acceleration data that are attached to a subject's trunk and that are obtained from sensors that detect angular velocity around the three axes and triaxial acceleration data. The step of calculating the suspicion of falling, the step of determining whether the event is suspected of falling based on the rate of jerk calculated in the step of calculating the degree of jerk, and the step of suspected to fall The vertical displacement calculation step of calculating the vertical displacement of the subject's trunk based on the acceleration and angular velocity data obtained from the sensor, and the vertical of the subject's trunk calculated in the vertical displacement calculation step And a fall determination step for determining whether or not the subject has fallen based on the displacement in the direction.

  According to the fall determination method according to the present invention, based on the jerk calculated from the acceleration data in the three-axis directions obtained from the sensor, it is determined whether or not there is a suspected fall event. The degree of jerk does not change significantly for minor impacts such as walking or whispering that does not cause a fall, but there is a significant change for the impact when a hip is attached or a hand is attached during a fall. Arise. For this reason, by using the jerk as a physical quantity for determining whether or not the event is suspected of falling, the falling event can be captured more appropriately, and the accuracy of calculating the vertical displacement can be improved.

  Hereinafter, an embodiment of a fall determination device embodying a fall determination method according to the present invention will be described with reference to the drawings.

  The fall determination device 10 is a device that determines a subject's fall. In this embodiment, as shown in FIG. 1, the fall determination device 10 includes a portable device 11 that is attached to the subject's body, and a fall is suspected. In addition, the information is communicated to a predetermined contact via a communication network 50 such as a cellular phone network or the Internet.

  As shown in FIG. 1, the portable device 11 includes an acceleration sensor 21, an angular velocity sensor 22, an A / D conversion unit 23 (amplifier), an internal clock 24, a CPU 25 (calculation unit), and a memory 26 (storage unit). ), A power source 27 (rechargeable battery), a jerk calculation unit 28, a fall suspect determination unit 29, a vertical displacement calculation unit 30, and a fall determination unit 31.

  The acceleration sensor 21 is configured by a sensor that detects acceleration in three axial directions orthogonal to each other. The angular velocity sensor 22 is configured by a sensor that detects an angular velocity around the axis with respect to the same three axes as the acceleration sensor 21. In this embodiment, the acceleration sensor 21 and the angular velocity sensor 22 detect acceleration and angular velocity with electric signals, respectively, and the detected electric signals are converted into signals indicating acceleration and angular velocity by the A / D conversion unit 23. It is stored in the memory 26.

  For example, H48A manufactured by Hitachi Metals, Ltd. may be used for the acceleration sensor 21, and ENC-03M manufactured by Murata Manufacturing Co., Ltd. may be used for the angular velocity sensor 22, for example. The acceleration sensor 21 and the angular velocity sensor 22 are not limited to this. Some acceleration sensors 21 detect gravitational acceleration. In this case, in order to obtain acceleration in each axial direction of the trunk, the gravitational acceleration is discounted from the detection amount obtained by the acceleration sensor 21. Find the acceleration.

  Moreover, in this embodiment, the sensor which detects the acceleration of the triaxial direction orthogonal to each other and the angular velocity around the three axes is composed of the acceleration sensor 21 and the angular velocity sensor 22 described above, and is attached to the trunk of the subject. Built in.

  The acceleration sensor 21 and the angular velocity sensor 22 may be configured by combining individual sensors, but a so-called orthogonal three-axis acceleration / angular velocity sensor in which the acceleration sensor 21 and the angular velocity sensor 22 are integrally configured may be used. good.

  For example, the portable device 11 is configured so that the three-axis sensors orthogonal to each other face the subject's front (z), the subject's body axis (y), and the subject's left and right (x), respectively. (Chest, abdomen, waist) Since the vertical displacement at the time of the fall is calculated, if it is assumed that the subject falls from a standing state, it can be expected that a larger displacement is obtained in the order of chest> abdomen> waist depending on the position where the sensor is attached.

  The internal clock 24 is a clock provided in the portable device 11. The memory 26 functions as a storage unit that stores various information of the mobile device 11. As a particularly important function, the memory 26 always stores the acceleration measured by the acceleration sensor 21, the angular velocity measured by the angular velocity sensor 22, and the time of the internal clock 24 in association with each other. In this embodiment, the memory 26 is used for recording these data while overwriting them for a predetermined time (for example, 1 minute). As a result, the required memory capacity can be reduced. The CPU 25 functions as a calculation means, and performs various calculation processes of the portable device 11 according to a predetermined program such as data writing data processing to the memory 26 and data transmission / reception. Further, the jerk calculation unit 28, the fall suspect determination unit 29, the vertical displacement calculation unit 30, and the fall determination unit 31 function the CPU 25 and the memory 26 according to predetermined programs, respectively, to realize their functions.

  The jerk calculation unit 28 executes the jerk calculation process, and may calculate the jerk from the acceleration data in the three-axis directions obtained from the acceleration sensor 21. The jerk is obtained by differentiating acceleration with respect to time. For example, the jerk in each axial direction is calculated by differentiating the acceleration with respect to time from the acceleration data in three axes obtained from the acceleration sensor 21. Good.

  The jerk is obtained by time-differentiating acceleration, and the degree of change in acceleration generated in the subject can be known. Since the jerk is the degree of change in acceleration, as shown in FIG. 2, the change is relatively small for slight whispering during walking and walking, and the subject fell compared to these events In some cases, significant changes occur. Therefore, by setting an appropriate threshold value for the jerk, an event that is suspected of falling can be captured more accurately. The jerk calculated by the jerk calculator 28 is stored in the memory 26.

  The fall suspect determination unit 29 executes a fall suspect determination step, and determines whether or not there is a suspected fall based on the jerk calculated by the jerk calculation unit 28. In this embodiment, a threshold setting unit 29a that sets a threshold for determining suspected fall is provided.

  As illustrated in FIG. 2, the threshold setting unit 29 a sets an appropriate threshold q for determining the suspected fall for the jerk calculated by the jerk calculating unit 28. Thereby, when the jerk calculated by the jerk calculating unit 28 exceeds the threshold q set by the threshold setting unit 29a, it can be determined that there is a possibility of falling.

  In this way, by determining whether or not there is a suspicion of falling based on the jerk, by calculating the vertical displacement of the subject's trunk for the event determined to be suspicious of falling Compared with the case where it is determined whether or not there is a suspicion of a fall based on the acceleration, the number of activations of the vertical displacement calculation unit can be reduced, and the calculation load can be reduced.

  If the threshold set in the threshold setting unit 29a is too high, the jerk may not exceed the threshold q in the event of a fall. Therefore, the threshold value q set in the threshold setting unit 29a is preferably set to an extent that is not too high so that an event in which a fall has occurred can be detected.

  Next, the vertical displacement calculation unit 30 executes a vertical displacement calculation step, and calculates the vertical displacement of the subject's trunk based on acceleration and angular velocity data obtained from the sensor. In this embodiment, the system is activated when the fall suspect determination unit 29 determines that a fall is suspected, and the vertical displacement of the subject's trunk is calculated for such an event. The vertical displacement may be calculated, for example, by calculating the acceleration in the vertical direction of the subject's trunk based on the acceleration and angular velocity data obtained from the acceleration sensor and the angular velocity sensor, and integrating twice in time.

  In this embodiment, the vertical displacement calculation unit 30 includes a maximum jerk search unit 30a, a maximum jerk search time setting unit 30b, a maximum jerk time storage unit 30c, a vertical acceleration calculation unit 30d, and a fall required time setting. A unit 30e and a multiple integral calculation unit 30f are provided.

  As shown in FIG. 2, the maximum jerk search unit 30 a executes a maximum jerk search process, and the jerk calculated by the jerk calculation unit 28 is stored in the threshold setting unit 29 a of the fall suspect determination unit 29. The maximum jerk time tb at which the maximum jerk is recorded is searched for within a predetermined max jerk search time ta from the time when the set threshold value q is exceeded. The maximum jerk search time ta for searching for the maximum jerk time tb is set by the maximum jerk search time setting unit 30b. The maximum jerk time tb is a time that is assumed to be the end of the overturn, and the maximum jerk time will be described below.

  In one suspicious fall event, a change in jerk that exceeds the threshold value q may be observed several times. In such a case, every time the jerk exceeds the threshold value q, it may be determined as a suspicion of falling, but in this case, the calculation may be complicated, and in each case, the displacement in the vertical direction is changed. The amount of calculation increases.

  According to the knowledge obtained from the experiments by the present inventors, the change in the jerk of various patterns of fall events was examined, and in one fall event, the subject put on a buttocks or a hand. , I found that when taking a passive, the greatest achievement. For example, if a subject has a whip that triggers a fall, and then falls a few steps while holding the foot, as shown in FIG. Although it changes markedly, the greatest change in jerk appears when the subject touches the hand or puts on the buttocks. In this way, even in other fall events, the greatest change in jerk appears when the subject touches his hand or puts his buttocks on. The point at which the greatest jerk is observed (maximum jerk) is the point when the subject falls and puts his hand or buttock on the basis of the vertical displacement of the subject's trunk According to the fall determination device 10 for determining a subject's fall, it can be assumed that the end point of the fall.

  The maximum jerk search time ta set in the maximum jerk search time setting unit 30b may be set assuming a time required for a fall. For example, in a normal fall, since it takes less than 1 second from the occurrence of a whip that triggers a fall, until it touches or puts on the buttocks, according to the knowledge of the present inventors, the maximum jerk The search time ta may be set to about 0.7 seconds. However, in the case where the subject generates a beating that triggers a fall, and then falls a few steps while holding his / her foot, the threshold is set by the threshold setting unit 29a. When the threshold value q is exceeded, it may be insufficient if the maximum jerk search time ta is set to about 0.7 seconds. Therefore, assuming such a case, the maximum jerk search time ta may be set slightly longer.

  The maximum jerk search unit 30a is activated when the jerk exceeds the threshold q set by the threshold setting unit 29a of the suspected fall determination unit 29, and is set in the maximum jerk search time setting unit 30b after being activated. A search is made for a time point tb at which the greatest jerk (maximum jerk) is observed among the jerk recorded within the maximum jerk search time ta. The maximum jerk time storage unit 30c records the maximum jerk time tb. In this embodiment, as described above, it is assumed that such maximum jerk time tb is the end time of the operation suspected of falling.

  Next, the vertical displacement calculation unit 30 calculates the vertical acceleration Ao of the subject's trunk based on the acceleration and the angular velocity detected by the sensor attached to the subject's trunk in the vertical acceleration calculation unit 30d.

  The vertical acceleration calculation unit 30d performs a vertical acceleration calculation step, and calculates the vertical acceleration Ao of the subject's trunk based on the acceleration and angular velocity detected by the sensor attached to the subject's trunk. In this embodiment, accelerations (Ax, Ay, Az) and angular velocities (dθ / dt, dψ / dt, dφ /) in three axial directions (x, y, z) received from the portable device 11 worn by the subject. Based on the data of dt), the vertical acceleration Ao of the trunk of the subject is calculated.

  Below, the example of a calculation which calculates | requires the vertical acceleration Ao is demonstrated.

  In FIG. 3, X, Y, and Z are absolute coordinate systems, z is the front of the subject, y is the body axis of the subject, and x is the left and right axial directions of the subject. As shown in FIG. 3, the absolute coordinate system O-XYZ is set so that the human traveling direction at t = 0 is the positive direction of the X axis. In FIG. 3, θ represents the displacement angle (pitch angle) around the x axis, ψ represents the displacement angle (yaw angle) around the y axis, and φ represents the displacement angle (roll angle) around the z axis. Yes. θ, ψ, and φ can be obtained by time-integrating angular velocities around the x-axis, y-axis, and z-axis, respectively.

  In the memory 26, sensor output history for a predetermined time may be sampled and recorded at a predetermined interval T. For example, when the sensor output history for 7 seconds is sampled at intervals of 10 ms, the memory 23 may be recorded with a data structure in which 7 seconds are divided into 700 for each fall and suspicious motion. The time can be defined by t = kT (k = 0, 1,..., N = 699). The first time extracted from the memory 23 is represented by t = 0 (k = 0). For example, a time of about 3 seconds before and about 1 second thereafter may be extracted from the memory 23 with the maximum jerk time point as the center.

  The accelerations observed at the time t = kT (k = 0, 1,..., N = 699) on the xyz coordinate axes are sequentially expressed as Ax [k], Ay [k], and Az [k]. Ax represents the acceleration in the x-axis direction, Ay represents the acceleration in the y-axis direction, and Az represents the acceleration in the z-axis direction. Similarly, the angular velocities around each axis are described by dθ / dt [k], dψ / dt [k], dφ / dt [k].

In addition, the rotation transformation matrix necessary for expressing the vector expressed on the new coordinate axis obtained by rotating in the positive direction about each coordinate axis of the absolute coordinate system as an absolute coordinate is given by Equation 1.

Further, a coordinate transformation matrix for expressing the sensor coordinate system vector at time kT in the absolute coordinate system is given by the recurrence formula of Formula 2.

An acceleration vector in sensor coordinates at time kT is represented by As [k] = [Ax [k], Ay [k], Az [k]] ^T , As [k] in absolute coordinates, and A [k] = If [Ax [k], Ay [k], Az [k]] ^T is described, Equation 3 is obtained. Since the vertical acceleration Az [k] can be calculated from Equation 3, the vertical displacement can be calculated by numerical integration.

If the sensor to be used is capable of detecting the DC component of acceleration, gravity acceleration that does not contribute to motion is superimposed on the vertical acceleration obtained by Equation 3, and therefore correction must be performed by Equation 4.

  In this embodiment, since the above-described maximum jerk time is assumed to be the end time tb of the motion suspected of falling, the vertical acceleration is calculated in several seconds before and after the maximum jerk time tb. The acceleration in the vertical direction of the subject may be calculated using the data on the angular velocity.

  Next, the vertical displacement calculation unit 30 calculates the vertical direction of the subject's trunk in relation to the time after the predetermined fall required time tc from the maximum jerk time tb searched by the maximum jerk search unit 30a. The vertical displacement of the subject's trunk is calculated by integrating the acceleration twice with the time required for the fall required time tc from the maximum jerk time tb.

  The fall required time setting unit 30e sets such a fall required time tc, and the multiple integral calculation unit 30f calculates the calculated vertical acceleration of the subject's trunk from the maximum jerk time tb. Integrate twice at a time that goes back tc.

  That is, the vertical displacement ΔZ of the subject's trunk can theoretically be obtained by time-integrating the vertical acceleration of the subject's trunk twice, and ΔZ = ∫∫ (Ao) dt is set to t = 0 to tb. Can be obtained by time integration twice. That is, the vertical displacement ΔZ can be obtained by, for example, an expression of vertical displacement ΔZ = ∫ (vertical velocity) dt = ∫ (∫ (vertical acceleration) dt) dt.

  In this embodiment, the maximum jerk time tb searched by the maximum jerk search unit 30a is assumed to be the fall end time tb, and the fall required time set in advance in the fall required time setting unit 30e from the fall end time tb. The vertical displacement of the subject's trunk is integrated by integrating the calculated vertical acceleration of the subject's trunk twice in the time required for the fall required time tc from the maximum jerk time tb. Is calculated.

  Note that the vertical displacement ΔZ (t) of the subject at the time of the fall is assumed to be a pattern like the graph shown in FIG. At this time, in order to facilitate the calculation of the vertical displacement ΔZ, as shown in FIG. 5, a time axis τ (τ = tc−t) obtained by reversing the time axis with respect to the vertical acceleration Ao (t) is taken, and Ao ( τ) may be calculated by time integration twice.

  The fall determination unit 31 performs a fall determination step, and determines the fall of the subject based on the vertical displacement of the subject's trunk calculated by the vertical displacement calculation unit 30. In this embodiment, as shown in FIG. 1, a threshold setting unit 31a is provided, and as shown in FIG. 6, a threshold r is set in advance for the vertical displacement ΔZ, and the vertical displacement ΔZ is larger than a predetermined threshold r. This is done by determining whether or not. The threshold value r may be set to an appropriate value in consideration of the height of the subject and the position of the trunk of the subject to which the sensor is attached (for example, the chest, abdomen, and waist).

  Since the fall determination device 10 determines the fall of the subject based on the vertical displacement ΔZ of the trunk of the subject, even if the impact is alleviated by touching or holding hands at the fall, When there is a suspicion, it can be detected reliably. Further, even when an elderly person gets up or lies down and the acceleration sensor reacts, the vertical displacement calculation unit 30 is activated based on the jerk, so that the jerk is not accompanied by a fall. It is possible to accurately determine whether or not it is caused by a fall without trajectory in a small motion. Thereby, erroneous determination is reduced, and more accurate determination is possible.

  The fall determination unit 31 not only determines the subject's fall based on the vertical displacement ΔZ of the trunk, but includes, for example, a change in the direction of gravity acceleration relative to the trunk before and after the fall, a displacement of the pitch angle and the yaw angle, The reliability of the fall determination can be further improved by determining the fall of the subject in consideration of other indicators such as the time (tb) from the start of the fall to the end of the fall.

  The change in the direction of the gravitational acceleration relative to the trunk before and after the fall may be detected by installing a sensor having a function capable of detecting the direction of the gravitational acceleration in the portable device 11. For example, since the acceleration sensor 21 can detect the gravitational acceleration, the gravitational acceleration can be calculated by comprehensively considering the triaxial acceleration sensor attached to the portable device 11. By determining whether or not there is a change in the direction of gravity acceleration before and after the fall, the certainty of the fall determination can be compensated. Specifically, it is preferable to determine whether or not there is a certain change in the axis for detecting gravitational acceleration when t = 0 and when t = tb. Thereby, for example, if the posture is lying on its back or lying on its back after falling, the direction of gravity acceleration with respect to the trunk changes, and this can be used to supplement the certainty of the fall determination.

  Alternatively, the angular velocity sensor may be integrated over time to observe the pitch angle or yaw angle displacement during the time t = 0 to tb, thereby supplementing the certainty of determination. Also, the smaller the time (tb) from the start of the fall to the end of the fall, the shorter the vertical displacement occurred, so whether the subject fell or the subject intentionally changed their posture. It becomes an index to judge.

  To determine the displacement of the pitch angle θ and the yaw angle φ and the time (tb) from the beginning of the fall to the end of the fall (tb), a threshold is set for these, and these indicators are used together to determine the fall of the subject. May be.

  In addition, taking into account the acceleration sensor data after it was determined to fall and the displacement of the pitch angle and yaw angle, the subject's injury status, such as the posture of the subject after the fall and whether the body is movable It is possible to obtain judgment materials for grasping the above.

  In the fall determination device 10, when it is determined that the subject has fallen by the fall determination process described above, the subject falls to a predetermined contact stored in a memory that is contacted when the subject determines to have fallen. It is informed that it has been determined that it has been done.

  In this embodiment, the GPS module 41, the position detection part 42, the modem circuit 43, the weak radio | wireless module 44, and the communication part 45 are provided as such a communication means. Also,

  The GPS module 41 is a satellite positioning device that measures the current position based on a signal received from a GPS satellite. The position detection unit 42 detects the position of the mobile device 11 based on the satellite positioning signal received by the GPS module 41. The modem circuit 43 amplifies the radio signal.

  The weak wireless module 44 receives a predetermined weak radio wave. In this embodiment, the weak radio module 44 prevents the GPS module 41 from being activated when a predetermined weak radio wave is received. That is, since the GPS module 41 consumes a large amount of power, if the GPS module 41 is always activated, the continuous use time of the mobile device 11 is shortened. Further, it is not necessary to activate the GPS module 41 while the subject is in a predetermined place such as home. Therefore, for example, a weak radio wave transmitter is installed in a predetermined place such as a home, and while the weak radio module 44 receives radio waves emitted by the weak radio wave transmitter, activation of the GPS module 41 may be limited. . Thereby, since the starting time of the GPS module 41 can be shortened only to required time, the continuous use time of the portable apparatus 11 can be improved. Further, as a weak radio wave transmitter installed at home or the like, a dedicated transmitter may be installed, or for example, radio waves of a wireless device attached to a home phone may be used.

  The communication unit 45 is a part that performs a function of performing communication. For example, a unit that performs communication using a mobile phone communication network may be used. By using such a communication device, necessary information is transmitted to the mobile phone 51 of a predetermined contact via the network 50 such as a mobile phone communication network or the Internet, or the required information is transmitted to the predetermined server computer 52. It is possible to make a transmission or to transmit a predetermined mail from the server computer 52 to a predetermined contact computer 53. It is preferable to add the result of the fall determination and the position information detected by the GPS module 41 and the position detection unit 42 to the transmitted information.

  According to the fall determination device 10, the fall suspect determination unit 29 determines whether or not there is a suspicion of the fall based on the jerk in each axis direction calculated from the triaxial acceleration data obtained from the sensor. Judgment was made. The degree of jerk does not change significantly for minor impacts such as walking or whispering that does not cause a fall, but it changes greatly for the impact when a hip is attached or a hand is attached during a fall. Arise. For this reason, the fall event can be captured more appropriately by using the jerk as a physical quantity for determining whether or not the event has a suspected fall. And compared with the case where the suspicion of a fall is determined using acceleration, the frequency | count of starting of the vertical displacement calculation part 30 decreases, and it can hold down calculation load small.

  Further, in this embodiment, the fall suspect determination unit 29 determines that the jerk calculated by the jerk calculation unit 28 exceeds the threshold q set by the threshold setting unit 29a within a predetermined maximum jerk search time ta. Thus, a maximum jerk search unit 30a for searching for the maximum jerk time tb at which the maximum jerk is recorded is provided. The vertical displacement calculation unit 30 calculates the vertical displacement ΔZ of the trunk of the subject based on the maximum jerk time tb searched by the maximum jerk search unit 30a. Furthermore, in this embodiment, the vertical displacement calculation unit 30 determines the acceleration and angular velocity obtained from the sensor with respect to a time that goes back a predetermined fall required time tc from the maximum jerk time tb searched by the maximum jerk search unit 30a. Based on the above data, the vertical acceleration of the subject's trunk is calculated, and the calculated vertical acceleration of the subject's trunk is 2 times the time required to fall from the maximum jerk time tb. Since the vertical displacement of the subject's trunk is calculated by integrating the number of times, the vertical displacement ΔZ of the subject's trunk can be calculated with higher accuracy, and determination with higher accuracy can be performed.

  As mentioned above, although the fall determination apparatus concerning one embodiment of the present invention was explained based on a drawing, the fall determination apparatus of the present invention is not limited to the above-mentioned embodiment.

  For example, the acceleration sensor and the angular velocity sensor of the fall determination device can be mounted on a belt buckle, for example.

  In addition, the fall determination device according to the present invention can be applied to fall detection of a humanoid robot or the like, and subjects include not only humans but also humanoid robots.

The figure which shows the structural example of the fall determination apparatus which concerns on one Embodiment of this invention. It is a figure which shows an example of the time change of a jerk. The figure which shows the axial direction which attaches an acceleration sensor. The figure which shows the relationship between vertical displacement (DELTA) Z and the time-axis t. The figure which shows the relationship between the vertical acceleration Ao of a test subject's trunk, and the time-axis (tau). The figure which shows the fall determination process based on vertical displacement (DELTA) Z.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Fall determination apparatus 11 Portable apparatus 21 Acceleration sensor 22 Angular velocity sensor 23 Memory 23 Conversion part 24 Internal clock 26 Memory 27 Power supply 28 Judgment calculation part 29 Judgment part 29a Threshold setting part 30a Maximum jerk search part 30c Maximum jerk time point memory part 30b Maximum jerk search time setting unit 30e Fall required time setting unit 30f Multiple integral calculation unit 30d Vertical acceleration calculation unit 30 Vertical displacement calculation unit 31 Fall determination unit 31a Threshold setting unit 41 Module 42 Position detection unit 43 Modulation / demodulation circuit 44 Weak wireless module 45 Communication unit 50 Network 51 Mobile phone 52 Server computer 53 Contact computer ta Maximum jerk search time tb Maximum jerk time (falling end time)
tc Fall time required

Claims (7)

A jerk calculation step of calculating jerk from data of accelerations in three axes orthogonal to each other attached to the subject's trunk, and acceleration in three axes obtained from a sensor that detects angular velocity around the three axes; ,
Based on the jerk calculated in the jerk calculation step, the suspected fall determination step that determines whether or not there is an event of suspected fall, and
A vertical displacement calculating step for calculating a vertical displacement of the trunk of the subject based on acceleration and angular velocity data obtained from the sensor for the event determined to be a suspicion of falling in the suspected falling determination step;
A fall determination method comprising: a fall determination step for determining whether or not a subject has fallen based on the vertical displacement of the trunk of the subject calculated in the vertical displacement calculation step. In the suspected fall determination step, the maximum jerk when the maximum jerk is recorded within the predetermined maximum jerk search time from when the jerk calculated in the jerk calculation step exceeds a preset threshold value It has a maximum jerk search process to search for
The fall determination according to claim 1, wherein the vertical displacement calculating step calculates a vertical displacement of the trunk of the subject based on the maximum jerk time point searched in the maximum jerk searching step. Method.   The vertical displacement calculating step is based on acceleration and angular velocity data obtained from the sensor for a time that is a predetermined time required to fall from the maximum jerk searched in the maximum jerk search step. The vertical acceleration of the subject's trunk is calculated, and the subject's body is integrated by integrating the calculated vertical acceleration of the subject's trunk twice in the time required to fall from the maximum jerk time point. The fall determination method according to claim 2, wherein the vertical displacement of the trunk is calculated. A sensor that is attached to the trunk of the subject and detects acceleration in three axial directions orthogonal to each other and angular velocity around the three axes;
A jerk calculator for calculating jerk from acceleration data in the three-axis directions obtained from the sensor;
Based on the jerk calculated by the jerk calculation unit, a fall suspect determination unit that determines whether there is an event of suspected fall, and
Based on acceleration and angular velocity data obtained from the sensor, a vertical displacement calculator that calculates the vertical displacement of the subject's trunk;
A fall determination unit that determines the fall of the subject based on the vertical displacement of the trunk of the subject calculated by the vertical displacement calculation unit;
For the event that the fall suspect determination unit determines that there is a suspected fall, the vertical displacement calculation unit calculates the vertical displacement of the subject's trunk, and the fall determination unit determines whether or not the subject has fallen A fall determination device characterized by:   The said fall suspected determination part is provided with the threshold value setting part which sets an appropriate threshold value for determining the suspected fall to the jerk calculated by the jerk calculation part. The fall judging device described. The fall suspect determination unit records the maximum jerk within a predetermined maximum jerk search time from when the jerk calculated by the jerk calculator exceeds the threshold set by the threshold setting unit. Equipped with a maximum jerk search unit that searches for the maximum jerk time point,
6. The fall determination device according to claim 5, wherein the vertical displacement calculation unit calculates the vertical displacement of the trunk of the subject based on the maximum jerk point searched by the maximum jerk search unit. .   The vertical displacement calculation unit is a test subject based on acceleration and angular velocity data obtained from the sensor with respect to a time retroactive to a predetermined fall required time from the maximum jerk point searched by the maximum jerk search unit. The vertical acceleration of the subject's trunk is calculated, and the subject's body is integrated by integrating the calculated vertical acceleration of the subject's trunk twice in the time required to fall from the maximum jerk time point. The fall determination device according to claim 6, wherein the vertical displacement of the trunk is calculated.

Images