JP6578874B2 - Walking cycle detection method and detection apparatus - Google Patents

Description

  The present invention relates to a walking cycle detection method and detection apparatus.

  In walking, from one foot landing to the next foot landing is said to be one walking cycle, which is widely used for detailed analysis of walking motion.

  As a method of measuring the walking cycle, a method is used in which images of the subject are taken from multiple directions, the distance between the ankles of both legs of the 3D human model is calculated, and the walking cycle is calculated from the minimum value of the distance between the ankles of both legs (patent) Reference 1), method of measuring walking cycle using a seat-type pressure sensor (Patent Document 2), measuring acceleration in the vertical direction or front-rear direction while walking with an accelerometer carried by the subject, and measuring the measured acceleration as a frequency A method is known in which a peak of a power spectrum is detected by analysis and a walking cycle is obtained based on the peak frequency (Patent Document 3).

JP 2009-189671 A JP 2014-94069 A JP 2005-342254 A

  However, as described in Patent Document 1, when the walking cycle is detected using the foot position information, the walking cycle cannot be accurately detected from the distance between the ankles of both legs in a subject having a left-right difference in the foot. . In addition, since it is necessary to shoot from multiple directions using a plurality of imaging devices, a dedicated large space where a large measuring device can be installed is required, and the detection system is complicated.

  When a seat type pressure sensor is used as described in Patent Document 2, the walking cycle when the subject walks on the seat type pressure sensor in a walking feature inspection hall or the like can be obtained accurately. The obtained walking cycle does not always coincide with the walking cycle in daily life.

  According to the method using an accelerometer carried by a subject as described in Patent Document 3, it is possible to detect a walking cycle in daily life. However, since it is necessary to obtain a power spectrum by performing frequency analysis on the detected acceleration, a large amount of calculation is required to obtain a walking cycle, and the walking cycle cannot be easily obtained in real time.

  On the other hand, the subject of this invention is related with making it possible to detect the walk cycle of everyday life correctly and in real time.

  The inventor paid attention to the fact that the hip rotates in response to whether the foot starting to move forward during walking is the right foot or the left foot, and from the acceleration detected by the accelerometer attached to the waist, We have devised an algorithm that can inspect in real time whether the moving foot is the right foot or the left foot, and further conceived the present invention to inspect the walking cycle from the inspection result.

That is, according to the present invention, the acceleration during walking of the subject is measured at a predetermined measurement cycle using an acceleration sensor mounted on the waist of the subject, and a time point that is a predetermined time t2 after a certain time point T0 is T2, and the time point T0 is higher than the time point T0. When the point in time in the predetermined time t3 is T3 (where t2 <t3),
Moving average Y _3-2 of acceleration in the left-right direction between time T3 and time T2,
Moving average Y _2-0 of acceleration in the left-right direction between time T2 and time T0,
Difference ΔY between moving average Y _2-0 and moving average Y _3-2 ,
To calculate
A threshold value y1 for determining that one of the left and right feet (hereinafter referred to as the first foot) has started moving forward, and that the other foot (hereinafter referred to as the second foot) has started moving forward. Set a threshold value y2 of ΔY for
Based on the magnitude relationship between ΔY at time point T0 and threshold y1, and the magnitude relationship between ΔY at time point T0 and threshold value y2, the time from time point T0 until the first foot starts moving forward until the second foot starts moving forward. It is determined whether it is in a state (hereinafter referred to as a first foot transition state) or in a state (hereinafter referred to as a second foot transition state) from when the second foot starts moving forward until the first foot starts moving forward. Periodically check the walking state,
Provided is a walking cycle detection method in which a repetition cycle of a first foot transition state or a second foot transition state is a walking cycle.

Further, the present invention is a walking cycle detection device that detects the walking cycle from the acceleration of the waist of the subject who is walking, as a walking cycle detection device that implements the above-described walking cycle detection method,
An acceleration sensor that is mounted on the waist and that can measure the lateral acceleration of the waist during walking, and an arithmetic device that detects the walking cycle from the acceleration measured by the acceleration sensor,
The arithmetic unit is
A function of extracting the acceleration in the left-right direction measured by the acceleration sensor at a predetermined measurement cycle;
When a certain time T0, a time point T2 past the predetermined time T0 is T2, and a time point T3 past the predetermined time T0 is T3 (where t2 <t3),
Moving average Y _3-2 of acceleration in the left-right direction between time T3 and time T2,
The moving average Y _2-0 of the lateral acceleration between the time point T2 and the time point T0, and the difference ΔY between the moving average Y _2-0 and the moving average Y _3-2 ,
The ability to calculate
When the threshold value of ΔY for determining that the first foot starts moving forward is y1, and when the threshold value of ΔY for determining that the second foot starts moving forward is y2, the ΔY and the threshold value y1 A walking state test is periodically performed to determine whether the time point T0 is in the first foot transition state or the second foot transition state based on the magnitude relationship between the current time T0 and the threshold value y2. A walking state inspection function and a function of detecting a repetition cycle of the first foot transition state or the second foot transition state and outputting as a walking cycle;
A walking cycle detection device is provided.

  ADVANTAGE OF THE INVENTION According to this invention, the walk cycle of a test subject's daily life can be detected correctly and in real time using the acceleration of a test subject's left-right direction measured with the acceleration sensor with which the test subject was mounted | worn.

  In particular, when the acceleration in the vertical direction of the subject is also used, it is possible to accurately and in real time detect from one foot landing to the next foot landing, which is useful for biophysical engineering. Walking data can be obtained.

FIG. 1 is a flowchart of a walking cycle detection method according to an embodiment. FIG. 2 is an explanatory diagram of the lateral acceleration and the vertical acceleration used for detecting the walking cycle when a triaxial acceleration sensor is used. FIG. 3 is an explanatory diagram of a method of calculating a moving average of accelerations in the left-right direction and a moving average of accelerations in the vertical direction. FIG. 4 is an explanatory diagram of the waveform of ΔY and the waveform of ΔX. FIG. 5 is a flowchart of the algorithm in the walking state inspection. FIG. 6 is obtained by superimposing the determination results of the walking state inspection and the landing inspection on the data of the triaxial acceleration in the triaxial acceleration sensor. FIG. 7 is a graph of determination values related to the walking cycle detected for the subject 1 by the apparatus of the example. FIG. 8 is a graph of determination values related to the walking cycle detected for the subject 2 by the apparatus of the example. FIG. 9 is a graph of determination values related to the walking cycle detected for the subject 3 by the apparatus of the example. FIG. 10 is a graph of determination values related to the walking cycle detected for the subject 4 by the apparatus of the example.

  Hereinafter, the present invention will be described in detail with reference to the drawings. In each figure, the same numerals indicate the same or equivalent components.

<Walking cycle detection device>
FIG. 1 is a flowchart of a walking cycle detection method performed by the walking cycle detection device 1 according to an embodiment of the present invention.

  This walking cycle detection device 1 is a portable device that can be worn on the waist, and includes an acceleration sensor 2 and an arithmetic device 3, and the acceleration sensor 2 measures the acceleration of the rotation of the waist accompanying walking, The arithmetic device 3 detects the walking cycle by processing the acceleration measured by the acceleration sensor 2 by a specific method.

<Acceleration sensor>
As the acceleration sensor 2, a sensor that can measure at least the lateral acceleration of the waist of the subject who is walking is used, and preferably a sensor that can also measure the vertical acceleration of the waist as described later. For example, as shown in FIG. 2, a triaxial acceleration sensor 2 is used, and the acceleration in the left-right direction (Y direction) measured by the acceleration sensor 2 is used. In addition to detecting the walking cycle, as will be described later, it is also detected when the foot is landing on the foot, and when detecting the walking cycle from the left and right foot landing to the next foot landing, In addition to the acceleration in the direction (Y direction), the acceleration in the vertical direction (that is, the vertical direction) (X direction) is used.

  The acceleration sensor 2 is mounted on the waist of the subject A, preferably on the left and right sides. As a result, the change in the acceleration of the hip rotation accompanying walking increases, so that the measurement becomes easy and the detection accuracy of the walking cycle is improved.

  Hereinafter, as an example, the present invention will be described in the case where an acceleration sensor is attached to the right waist, the right foot is the first foot, and the left foot is the second foot. However, the acceleration sensor is attached to the left waist and the left foot is the first foot. In the case where the right foot is the second foot, or when the acceleration sensor is attached to the center of the waist width, one of the left and right feet is the first foot, and the other foot is the second foot, the same can be detected. . In that case, the walking cycle can be detected more accurately by the present invention by appropriately setting thresholds y1 and y2 described later.

<Calculation device>
(Outline of walking cycle detection method)
In the arithmetic device 3, roughly, the right and left direction acceleration is extracted at a predetermined measurement cycle, and the right and left direction acceleration at each measurement time point is the right foot transition state from when the right foot starts moving forward until the left foot starts moving forward, A walking state test is performed to determine which of the left foot transition state from when the left foot starts to move forward until the right foot starts to move forward, and a periodic repetition of the determination result of the right foot transition state or the left foot transition state is detected. Thus, the walking cycle is detected.

(Measurement cycle)
The arithmetic device 3 has a function of extracting and storing a lateral acceleration signal output from the acceleration sensor 2 at a predetermined measurement cycle in order to perform the above-described walking state inspection. Considering that a person's walking cycle is generally 0.8 to 1.2 seconds, the acceleration measurement cycle is preferably 10 Hz to 250 Hz, and erroneous determination is performed when a walking state inspection is performed using a specific algorithm described later. From the point of prevention, it is more preferably 32 to 128 Hz.

(Moving average in the horizontal direction)
More specifically describing the walking state inspection, when the arithmetic unit inspects the walking state at an arbitrary time point T0, as shown in FIG. 3, the time point T2 past the predetermined time t2 from the time point T0 is the time point T2. When the time point T3 past the time T0 is T3 (where t2 <t3),
A moving average Y _3-2 of acceleration in the left-right direction between time T3 and time T2, and a moving average Y _2-0 of acceleration in the left-right direction between time T2 and time T0,
Difference calculated, further a moving average Y _2-0 and the moving average Y _3-2 and ΔY = Y _2-0 -Y _3-2
Is calculated.

As described above, ΔY is calculated by using the moving averages Y _2-0 and Y _3-2 in the left and right directions for a predetermined time before a certain time T 0, so that ΔY is the degree of the forward movement of the leg at the time T 0. Will be shown. Then, by sequentially calculating ΔY for the lateral acceleration measured at a predetermined cycle, ΔY draws a periodic waveform shown in FIG.

In calculating the moving average of the acceleration in the left-right direction, in order to detect the degree of acceleration in the left-right direction at the beginning of the transition, T2 is preferably set to a time point 64 ms to 1600 ms past a certain time point T0, and T3 is set to a certain time point T0. From 192 msec to 4800 msec.
In addition, the certain time T0 can be set as the current time, which makes it possible to detect the walking cycle in real time.

(ΔY thresholds y1, y2)
On the other hand, a threshold value y1 for determining that the right foot starts moving forward and a threshold value y2 for determining that the left foot starts moving forward are set in the arithmetic unit.
While ΔY reaches from the threshold value y1 to the threshold value y2, it is a walking state from when the right foot starts moving forward until the left foot starts moving forward, and in the present invention this is called a right foot transition state. Similarly, while ΔY reaches the threshold value y1 from the threshold value y2, it is a walking state from when the left foot starts moving forward until the right foot starts moving forward, and in the present invention this is referred to as a left foot transition state.

  In setting the threshold values y1 and y2, as shown in FIG. 4, the right leg transition state includes the right leg swing leg period, and the left foot transition state includes the left leg swing leg period. Further, in order to prevent erroneous determination in performing a walking state test for determining whether the time point is the right foot transition state or the left foot transition state from the value of ΔY, it is preferable that y1 ≠ y2. If the difference between the threshold values y1 and y2 is too small, an erroneous determination may occur when determining the landing time of the right footpad by combining the walking state inspection and the landing inspection as will be described later, and the difference between the threshold values y1 and y2 If it is too large, it will be difficult to detect the walking cycle for subjects with low hip rotation during walking. A preferable threshold is detected by a motion capture system (for example, a three-dimensional motion analysis system using a plurality of video cameras, a VICON MX system manufactured by VICON, or a VICON NEXUS) or a floor reaction force meter detected by the method of the present invention. For example, when the maximum value of the waveform of ΔY is +8 and the minimum value is −8, y1 is −6 to −1 and y2 is 1 to 6. It is preferable to do.

  The arithmetic device may be configured such that the set values of the threshold values y1 and y2 can be adjusted in accordance with the position of the waist where the acceleration sensor is mounted, the degree of hip rotation when the subject walks, and the like.

(Walking state inspection)
In the walking state test, it is determined whether the walking state at a certain measurement time T0 is the right foot transition state or the left foot transition state based on the magnitude relationship between the value of ΔY and the threshold values y1 and y2. This determination is performed periodically. Preferably, for each lateral acceleration extracted in a predetermined measurement cycle, the determination is sequentially performed in the same cycle as the acceleration measurement cycle. As a result, the walking cycle can be detected in real time at the same time as the acceleration during walking is measured.

(Algorithm for walking state inspection)
In order to reduce the amount of calculation required for the walking state inspection, the arithmetic device determines, for example, whether the walking state at a certain measurement time T0 is the right foot transition state or the left foot transition state by an algorithm shown in FIG. Hereinafter, the algorithm will be described assuming that the measurement time T0 is the current time.

  First, it is determined whether or not ΔY ≦ y1 and ΔY ≧ y2 are satisfied at the current time T0. When Y ≦ y1, the current time T0 is determined as the right foot transition state, and when Y ≧ y2, the current time T0 is determined as the left foot transition state. To do.

  On the other hand, when y1 <ΔY <y2 at the current time T0, as can be seen from FIG. 4, it cannot be determined whether ΔY is in the right foot transition state of (d) or the left foot transition state of (b). Therefore, whether or not ΔY ≦ y1 or ΔY ≧ y2 is satisfied at the next measurement time T0 ′ is determined. At this time T0 ′, when ΔY is ΔY ≦ y1 in (c), considering that the interval between the current time T0 and the next measurement time T0 ′ is the above-described measurement cycle, the current time T0 is shifted to the right foot of (d). The possibility of a state may be excluded. Accordingly, when ΔY ≦ y1 in (c) at the next measurement time T0 ′, the current time T0 is in the left foot transition state in (b). Similarly, when ΔY is ΔY ≧ y2 in (a) at the next measurement time T0 ′, the possibility that the current T0 is in the left foot transition state in (b) may be excluded. Therefore, when ΔY ≧ y2 in (a) at the next measurement time T0 ′, the current T0 is in the right foot transition state in (d).

  FIG. 5 shows a flow of determining whether or not ΔY ≦ y1 at the current time T0, and then determining whether or not ΔY ≧ y2, and whether ΔY ≦ y1 is true at the next measurement time T0 ′. The flow for judging false and then judging whether or not ΔY ≧ y2 is shown. In these judgments, first, whether or not ΔY ≧ y2 is judged, and then whether or not ΔY ≦ y1 is judged. Also good.

  If the walking state at the current time T0 cannot be determined because ΔY ≦ y1 or ΔY ≧ y2 at the next measurement time point T0 ′, ΔY ≦ at the next measurement time point after the next measurement time point T0 ′. Whether or not y1 or ΔY ≧ y2 is satisfied is determined, and the walking state at the current time T0 is determined in the same manner as the determination at the next measurement time point T0 ′.

  Thus, the walking state at a certain time T0 can be determined by a simple determination method without drawing a waveform of ΔY or calculating the slope of ΔY at a certain time T0. Therefore, ΔY can be calculated in real time with respect to the acceleration in the left-right direction calculated at a predetermined measurement cycle, and it can also be determined whether the right foot transition state or the left foot transition state in real time.

  The repetition cycle of the right foot transition state or the left foot transition state obtained by the walking state inspection corresponds to the walking cycle. Therefore, the walking cycle can be easily detected from the repetition cycle of the determination result.

  The determination value of the walking state test in FIG. 4 is obtained by giving a numerical value 50 at the time point when the right foot transition state is determined in the above-described walking state inspection, and by giving a numerical value 100 at a time point when the left foot and subsequent state is determined. Thus, by giving different numerical values to the determination values of the right foot transition state and the left foot transition state, the determination value becomes a rectangular pulse waveform, and one walking cycle can be extracted more easily.

  The walking cycle is generally the time from the landing of one foot to the next landing of that foot. However, the walking cycle is the same as the walking repetition cycle even when the landing is based on the landing or the arbitrary walking phase in one walking cycle (for example, ground contact with the foot, toe off). Therefore, the walking cycle can be obtained from the repetition cycle of the above determination result.

(Examination of walking state by other method)
In the present invention, the method for obtaining the walking cycle from the waveform of ΔY and the predetermined threshold value y1 or y2 is not limited to the method based on the algorithm described above. For example, as shown in FIG. 4, when the slope of ΔY is negative and ΔY falls below the threshold value y1, the right foot transition state starts, and the slope of ΔY at that time is positive and exceeds the threshold value y2. Since the time is the start time of the left foot transition state, the slope of ΔY and the time when ΔY takes thresholds y1 and y2 are detected from the start time of the right foot transition state or the start time of the left foot transition state, and the repetition cycle is detected. Thus, the walking cycle may be detected. From the viewpoint of simplifying the calculation, the above-described algorithm method is preferable.

(Detection of walking cycle based on the landing)
On the other hand, in the present invention, when the walking cycle from the right foot heel landing to the next right foot heel landing is obtained with reference to the foot landing, the acceleration in the vertical direction is measured together with the acceleration in the left-right direction. The acceleration in the vertical direction is preferably measured at the same measurement cycle in synchronization with the measurement cycle of the acceleration in the left-right direction.

When the right foot is landing on the heel, it is possible to detect the following using a combination of the landing test for determining the presence or absence of the foot landing at a certain time T0 using the vertical acceleration and the above-mentioned walking state test. it can. First, as shown in FIG. 3, a time T1 past time t1 from a certain time T0 and the above-mentioned times T2 and T3 are considered (where t1 <t2).
A moving average X _3-2 of vertical acceleration between time T3 and time T2, and a moving average X _2-1 of vertical acceleration between time T2 and time T1,
Further, the difference ΔX = X _2-1 −X _3-2 between the moving average X _2-1 and the moving average X _3-2 is calculated. In this way, in order to determine whether or not the foot has landed at a certain time T0, the difference ΔX between the past moving averages X _2-1 and X _3-2 not including the most recent acceleration at the certain time T0 is calculated. Is for detecting the degree of swinging up of the legs in the vertical direction. This ΔX has a periodic waveform as shown in FIG.

  In calculating the moving average of the acceleration in the vertical direction, in order to detect the swinging down of the foot before landing, T1 is a time point 32 to 800 milliseconds past a certain time point T0, and the time points T2 and T3 are as described above. The same is preferable. As in the case of obtaining the moving average of the acceleration in the left-right direction, when obtaining the moving average of the acceleration in the vertical direction, a certain time T0 can be set as the current time. It becomes possible to detect in real time.

(Implantation inspection)
On the other hand, a value x1 of ΔX for determining that the foot is landing is set in advance, and a landing test is performed to determine whether or not there is a landing based on the value of ΔX. More specifically, a threshold value x1 for determining that the right foot or the left foot has landed when ΔX is equal to or less than x1 is set in the arithmetic unit. In addition, the computing device is provided with a function of landing inspection that determines that the right foot or the left foot is landed when the slope of ΔX is negative and ΔX is equal to or less than the threshold value x1.

  As a more specific setting method of the threshold value x1, when the maximum value of the waveform of ΔX is +8 and the minimum value is −8, it is preferable to set x1 to −1 to −6. The numerical value of x1 is a motion capture system (for example, a three-dimensional motion analysis system using a plurality of video cameras, a VICON MX system or a VICON NEXUS) using a plurality of video cameras, and a floor reaction force meter. It is determined empirically so as to coincide with the time of landing of the saddle detected by the above.

  As can be seen from FIG. 4, when the presence of landing is determined in the landing test, that is, when the slope of ΔX is negative and ΔX is equal to or less than the threshold value x1, the time when the right foot landing and the left foot landing are reached. included. Therefore, in the arithmetic device, the first measurement time of the section determined to be landing in the landing inspection and determined to be the right foot transition state in the above-described walking state inspection is determined as the right footpad landing time. To do. Then, by detecting the repetition cycle at the time of landing on the right footpad, the walking cycle based on the landing on the right footpad is detected.

  The determination value of “landing inspection + walking state inspection” in FIG. 4 gives a determination value of 50 when it is determined that there is landing in the above-described landing inspection and in the walking state inspection that the right foot transition state is determined, In addition, a determination value 100 is given. In this rectangular pulse waveform, the falling portion to the determination value 50 is when the right footpad is landing, so that one walking cycle based on the right footpad landing can be easily extracted.

  FIG. 6 is a graph in which time is taken on the horizontal axis, and the acceleration in the horizontal direction, the acceleration in the vertical direction, and the acceleration in the front-rear direction measured by the triaxial acceleration sensor are taken on the vertical axis. The waveform of the judgment value of “+ walking state inspection” is entered. Thus, although it is extremely difficult to extract the walking cycle directly from the raw acceleration data, it is possible to easily extract one walking cycle based on the right footpad landing from the determination of the landing examination and the walking state examination. I understand.

  The ability to easily extract the walking cycle based on the right footpad landing is, for example, one stride from the right footpad landing to the next right footpad landing when analyzing acceleration data during walking of many subjects. Since the data can be extracted, it is very significant in terms of biomechanics, and is useful for analysis of walking in daily life.

The present invention can take various forms.
For example, the walking cycle detection device can include a display, a printer, or the like as a display device that displays the walking cycle output by the arithmetic device, the determination value related to the walking cycle, the time of landing on the right foot, and the like.

  The walking cycle detection device can be configured as a portable device in which an acceleration sensor and a calculation device are incorporated, preferably a display. On the other hand, when an acceleration sensor incorporated in a mobile phone or the like is used, it is preferable that the arithmetic device constituting the walking cycle detection device is provided separately from the acceleration sensor. In this case, the acceleration data measured by the acceleration sensor is transferred to the arithmetic device through the public telecommunication line, the walking cycle detected by the arithmetic device, or the determination value related thereto is transferred to the mobile phone etc. through the public telecommunication line, For example, it may be configured to be displayed on an operation screen of a mobile phone or the like.

Hereinafter, the present invention will be specifically described with reference to the drawings.
Example A data logger equipped with a three-axis acceleration sensor (manufactured by Accors Corporation) is used as an acceleration sensor, and acceleration data is transmitted to the PC local environment using Bluetooth, and the flow shown in FIG. A gait cycle detection device that detects the gait cycle based on the landing point was made. In this case, the lateral acceleration and the vertical acceleration are measured at 64 Hz, and the walking state inspection and the landing inspection can be performed in real time for each measured acceleration. In addition, when the landing inspection determines that there is a landing and the walking state inspection determines the right foot transition state, the determination value is 350, and the other determination values are 450.

  Four healthy men in their twenties were subjects (subjects 1 to 4). Each subject walks on the floor reaction force meter (BP400600-1000P, manufactured by Amti) with the acceleration sensor attached to the right waist (near the upper anterior iliac spine), and the floor reaction force meter is used during this walking. While measuring force, the walk period detection test which outputs the above-mentioned judgment value with the walk period detector of an example was done. This walking cycle detection test was performed three times for each subject. The results are shown in FIGS.

  In general, the walking cycle from starboard landing to the next starboard landing is measured using a floor reaction force meter as a standard measurement method. Two-mountain type floor reaction force is measured during the two-foot support period following the standing up, right foot landing, the right-legged stance phase, and the both-foot support period, and the floor reaction force becomes zero during the right leg swing phase period.

  7 to 9, the time point when the determination value by the walking cycle detection device of the embodiment decreases from 450 to 350 substantially coincides with the time point when the floor reaction force by the floor reaction force meter suddenly rises.

  Next, the difference between the time point when the determination value becomes 350 in the walking cycle detection test of the example and the start point of rising of the floor reaction force measured by the floor reaction force meter Calculated as detection error. Moreover, the average of the absolute value of each error was calculated as an average error. The results are shown in Table 1.

  In this walking cycle detection test, since the acceleration measurement cycle is 64 Hz (15.625 msec), the detection error at the time of landing is generally within the measurement interval of acceleration. According to this walking cycle detection device, the right footpad landing It can be seen that the time can be detected accurately.

DESCRIPTION OF SYMBOLS 1 Walking cycle detection apparatus 2 Acceleration sensor or 3-axis acceleration sensor 3 Arithmetic apparatus A Test subject

Claims (9)

The acceleration in the left-right direction during walking of the subject is measured at a predetermined measurement cycle using an acceleration sensor mounted on the subject's waist, the time point T2 is a predetermined time t2 from a certain time point T0, and the predetermined time t3 is a time point T3 from the time point T0. When the past time point is T3 (where t2 <t3),
Moving average Y _3-2 of acceleration in the left-right direction between time T3 and time T2,
Moving average Y _2-0 of acceleration in the left-right direction between time T2 and time T0,
Difference ΔY between moving average Y _2-0 and moving average Y _3-2 ,
To calculate
A threshold value y1 for determining that one of the left and right feet (hereinafter referred to as the first foot) has started moving forward, and that the other foot (hereinafter referred to as the second foot) has started moving forward. Set a threshold value y2 of ΔY for
Based on the magnitude relationship between ΔY at time point T0 and threshold y1, and the magnitude relationship between ΔY at time point T0 and threshold value y2, the time from time point T0 until the first foot starts moving forward until the second foot starts moving forward. It is determined whether it is in a state (hereinafter referred to as a first foot transition state) or in a state (hereinafter referred to as a second foot transition state) from when the second foot starts moving forward until the first foot starts moving forward. Periodically check the walking state,
A method for detecting a walking cycle in which a repetition cycle of a first foot transition state or a second foot transition state is a walking cycle. When the threshold values y1 and y2 are set in the walking state test so that the first foot transition state is determined when ΔY is equal to or less than the threshold value y1, and the second foot transition state is determined when ΔY is equal to or greater than the threshold value y2. ,
Determining whether or not ΔY ≦ y1 and ΔY ≧ y2 at the time T0,
When ΔY ≦ y1, the time point T0 is determined as the first foot transition state,
When ΔY ≧ y2, the time T0 is determined to be the second foot transition state,
y1 <When [Delta] Y <y2, 'to determine the presence or absence of satisfaction of delta Y ≦ y1 and [Delta] Y ≧ y2 in, the time T0' next measurement time T0 when the [Delta] Y ≦ y1 in, the time T0 transition 2nd foot It is determined that
The method of detecting a walking cycle according to claim 1, wherein when ΔY ≧ y2, the time point T0 is determined to be the first foot transition state. Measure the acceleration in the vertical direction during walking of the subject at a predetermined measurement cycle using the acceleration sensor,
When the point in time T1 past the point T0 is T1 (where t1 <t2),
Moving average X _3-2 of vertical acceleration between time T3 and time T2,
Moving average X _2-1 of vertical acceleration between time T2 and time T1,
Difference ΔX between moving average X _2-1 and moving average X _3-2
To calculate
1st foot or the bipedal sets a threshold value x1 of ΔX for determining that landed,
Perform an implantation test to determine the presence or absence of the first or second foot ,
Claim that the first measurement time of the section in which it is determined that the first foot or the second foot is landing in the landing inspection and the first foot transition state is determined in the walking state inspection is the time when the first foot is landing on the heel Item 3. A method for detecting a walking cycle according to item 1 or 2.   The method for detecting a walking cycle according to any one of claims 1 to 3, wherein the measurement cycle is 10 Hz to 250 Hz. The time point T2 is a time point 64 ms to 1600 ms past a certain time point T0,
The method of detecting a walking cycle according to any one of claims 1 to 4, wherein the time point T3 is set to a time point 192 msec to 4800 msec past a certain time point T0.   The method for detecting a walking cycle according to any one of claims 3 to 5, wherein the time point T1 is a time point 32 to 800 milliseconds past a certain time point T0. A walking cycle detection device that detects a walking cycle from the acceleration of the waist of a subject who is walking,
An acceleration sensor that is mounted on the waist and that can measure the lateral acceleration of the waist during walking, and an arithmetic device that detects the walking cycle from the acceleration measured by the acceleration sensor,
The arithmetic unit is
A function of extracting the acceleration in the left-right direction measured by the acceleration sensor at a predetermined measurement cycle;
When a certain time T0, a time point T2 past the predetermined time T0 is T2, and a time point T3 past the predetermined time T0 is T3 (where t2 <t3),
Moving average Y _3-2 of acceleration in the left-right direction between time T3 and time T2,
The moving average Y _2-0 of the lateral acceleration between the time point T2 and the time point T0, and the difference ΔY between the moving average Y _2-0 and the moving average Y _3-2 ,
The ability to calculate
When the threshold value of ΔY for determining that the first foot starts moving forward is y1, and when the threshold value of ΔY for determining that the second foot starts moving forward is y2, the ΔY and the threshold value y1 A walking state test is periodically performed to determine whether the time point T0 is in the first foot transition state or the second foot transition state based on the magnitude relationship between the current time T0 and the threshold value y2. A walking state inspection function and a function of detecting a repetition cycle of the first foot transition state or the second foot transition state and outputting as a walking cycle;
A walking cycle detection device. In the walking state examination, when ΔY is less than or equal to threshold y1, the first foot transition state is determined, and when ΔY is greater than or equal to threshold y2, thresholds y1 and y2 are set so that the second foot transition state is determined. In addition,
Determining whether or not ΔY ≦ y1 and ΔY ≧ y2 at the time T0,
When ΔY ≦ y1, the time point T0 is determined as the first foot transition state,
When ΔY ≧ y2, the time T0 is determined to be the second foot transition state,
y1 <When [Delta] Y <y2, 'to determine the presence or absence of satisfaction of delta Y ≦ y1 and [Delta] Y ≧ y2 in, the time T0' next measurement time T0 when the [Delta] Y ≦ y1 in, the time T0 transition 2nd foot It is determined that
The walking cycle detection device according to claim 7, wherein when ΔY ≧ y 2, the time point T 0 is determined as the first foot transition state. The acceleration sensor can measure the acceleration in the vertical direction along with the lateral acceleration of the waist during walking,
When the time point T1 past the time point T0 is T1 (where t1 <t2),
Arithmetic unit is
Moving average X _3-2 of vertical acceleration between time T3 and time T2,
Moving average X _2-1 of vertical acceleration between time T2 and time T1,
Difference ΔX between moving average X _2-1 and moving average X _3-2
The ability to calculate
And said [Delta] X, determines landing the first foot or absence of implantation of the second foot on the basis of the threshold value x1 set as [Delta] X when the first foot and second foot is determined to be landed The ability to inspect,
The first measurement time of the section determined to have the first foot or the second foot landing in the landing inspection and the first foot transition state in the walking state inspection is output as the time of landing on the first foot. function,
The walking cycle detection device according to claim 7 or 8, comprising: