JP2016137227A - Walking measurement system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a walking measurement system capable of accurately grasping walking property of a subject.SOLUTION: A walking measurement system 100 comprises a laser range sensor 10 and an electronic controller 20. The laser range sensor 10 emits laser beam so as to scan along a horizontal direction, and based on a reflection state of the laser beam, acquires two-dimensional plane distance information related to a distance with leg parts of a subject on which the laser beam is reflected, over time. A data analysis part 24 of the electronic controller 20 performs filtering processing which uses a walking model in which, plural walking phases are defined, the plural walking phases contain a first walking phase when both leg parts of the subject are support legs, based on at least the two-dimensional plane distance information, and associates a position and a speed of the leg parts of the subject with a time, then calculates the position and the speed.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a walking measurement system.

  2. Description of the Related Art Conventionally, a walking measurement system that measures walking characteristics (leg transport) during walking of a subject is known. For example, Patent Document 1 describes a footprint analyzer that records walking training data, quantitatively grasps the training situation, and applies the training data to a diagnostic device by comparison with past data. ing. The footprint analysis apparatus described in Patent Document 1 includes an image capturing device, an image calculation device, and a display device. From the image captured by the image capturing device, each of the floor surface contact position and the floor surface contact position of each foot. The time-dependent change for each foot is calculated by the image calculation device, and the time-dependent change for each foot in the floor contact position is displayed on the display device.

JP 2002-345785 A

  By the way, in the expanding aging society, the fall of the elderly is an important problem, and it has been reported that 30% of the elderly over 75 years old experience a fall at least once a year. Falling is considered to occur in various scenes of daily life, and it is generally reported that it occurs due to various risk factors. Actually, physical strength measurement and walking training for prevention of falls of elderly people are carried out in community health activities, and an effective and simple fall risk evaluation system is required.

  A Timed Up and Go (TUG) test has been proposed as a gait test that can evaluate a subject's walking ability, dynamic balance, agility, and other functional movement ability. In the TUG test, the subject stands up from a sitting position, walks toward a marker such as a cone (also called a road cone or pylon) ahead of a predetermined distance (for example, 3 m), turns the marker, and then turns the chair. Go back to sit down again.

  Here, in such a walking test, a leg is observed with a sensor, and filtering processing using a walking model is performed based on the observation result to calculate the position and speed of the leg. In general, the walking model assumes that a person is walking at a constant speed, and enough consideration is given to when a stationary person starts walking or when a walking person stops. It has not been. Therefore, for example, in the TUG test, when the subject stands up and sits down on the chair again, it may be difficult to accurately grasp the walking characteristics of the subject.

  This invention is made | formed in view of the said situation, and makes it a subject to provide the walk measurement system which can grasp | ascertain a walk characteristic of a test subject accurately.

  A gait measurement system according to the present invention is a gait measurement system used for a test of a subject's gait, which emits a detection wave so as to scan along a horizontal direction, and detects the detection wave based on a reflection state of the detection wave. A distance information acquisition unit that acquires distance information regarding the distance to the object that reflects the light over time, and a first case where both legs of the subject are support legs based at least on the distance information acquired by the distance information acquisition unit A data analysis unit that performs a filtering process using a walking model in which a plurality of walking phases including the walking phase is defined, and estimates the position and speed of the leg of the subject in association with time.

  In this gait measurement system, the plurality of gait phases defined in the gait model includes a first gait phase in which both legs of the subject are support legs, that is, a gait phase at rest. Therefore, it is possible to cope with a case where a stationary person starts walking and a person who is walking stationary, and it is possible to accurately grasp the walking characteristics of the subject.

  In the walking measurement system according to the present invention, the data analysis unit detects the position of the leg at the predetermined processing time as the observation position from the distance information, and the walking phase identified based on the observation position is earlier than the predetermined processing time. When a change deviating from the periodic change with respect to the walking phase at the time, the observation position may be set as a false detection value. Thereby, the erroneous detection of the observation position can be determined using the fact that the human walking phase changes periodically.

  In the gait measurement system according to the present invention, the data analysis unit performs a support leg swing leg determination to determine whether each of the leg parts is a free leg or a support leg, and walks from the determination result of the support leg swing leg determination. The walking phase of the model may be specified, and the filtering process may be performed based on the acceleration corresponding to the specified walking phase. A person advances walking with one leg as a supporting leg and the other leg as a free leg. Both legs move periodically while changing the free leg and the supporting leg by landing. In the walking measurement system according to the present invention, it is possible to accurately grasp the walking characteristics of the subject in consideration of the acceleration change in the behavior of the legs.

  In the gait measurement system according to the present invention, the data analysis unit specifies the gait phase as the first gait phase when both legs are supporting legs, the subject's left leg is a free leg, and the right leg Is a support leg, and when the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a positive value, the walking phase is identified as the second walking phase, When the subject's left leg is a free leg and the right leg is a support leg, and the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a negative value The walking phase is identified as the third walking phase, and the subject's left leg is a supporting leg and the right leg is a free leg, the position vector of the right leg relative to the left leg and the right leg If the inner product with the velocity vector is a positive value, the walking phase is identified as the fourth walking phase, and the subject's left leg is the supporting leg. If the right leg is a free leg and the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg is a negative value, the walking phase is the fifth walking phase. May be specified. In this case, it is possible to specify the walking phase in consideration of the relative positional relationship between the legs and the moving speed.

  In the walking measurement system according to the present invention, the data analysis unit may specify that the walking phase is the sixth walking phase and determine that the subject is running when both legs are swing legs. In this case, it can be determined whether or not the subject is traveling.

  In the walking measurement system according to the present invention, the data analysis unit may perform a filtering process using a Kalman filter. In this case, for example, filtering processing considering noise and the like included in the distance information is possible.

  ADVANTAGE OF THE INVENTION According to this invention, the walk measurement system which can grasp | ascertain a test subject's walk characteristic accurately can be provided.

It is a block diagram which shows the structure of the walking measurement system which concerns on one Embodiment. It is the schematic which shows the TUG test to which the walking measurement system of FIG. 1 was applied. It is a figure which shows an example of the analysis result by the conventional walk measurement system. It is a top view explaining the laser range sensor of the walking measurement system of FIG. It is a 1st figure explaining the example of the observation pattern used for the detection of a leg part observation position. It is a 2nd figure explaining the example of the observation pattern used for the detection of a leg part observation position. It is a 3rd figure explaining the example of the observation pattern used for detection of a leg part observation position. It is a 4th figure explaining the example of the observation pattern used for detection of a leg part observation position. It is a figure explaining a walking model. It is a figure explaining the detection of the leg predicted position using a gate. It is a figure explaining the interpolation of the leg observation position at the time of concealment. It is a figure explaining the measurement error of the leg observation position in each observation pattern. It is a figure explaining calculation of floor contact time. It is a figure explaining a stride, a step, and stride length. It is a flowchart which shows the process in the walking measurement system of FIG. It is a flowchart which shows the process of the interpolation of the leg observation position at the time of concealment. It is a figure explaining the effectiveness of interpolation of a leg observation position at the time of hiding. It is a figure which shows an example of the analysis result by the walking measurement system of FIG. It is a figure which shows the example of the tracking result of both legs when a left leg is temporarily hidden from the laser range sensor. It is a graph which shows the result of a tracking success rate. An example of the tracking result when the left leg is hidden from the laser range sensor in the turning phase is shown. It is a graph which shows RMSE regarding the movement locus | trajectory and landing position of both legs. It is a figure which shows an example of the analysis result of the movement locus | trajectory of both legs, and a landing position.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. In the following description, the same reference numerals are used for the same or corresponding elements, and duplicate descriptions are omitted.

  FIG. 1 is a block diagram illustrating a configuration of a walking measurement system according to an embodiment. FIG. 2 is a schematic diagram illustrating a TUG test to which the walking measurement system of FIG. 1 is applied. FIG. 3A and FIG. 3B are diagrams showing examples of analysis results obtained by a conventional walking measurement system. In FIG. 3, the line indicates the movement trajectory of the leg. On the line, a small mark indicates a free leg, and a large mark indicates a landing position.

  As illustrated in FIGS. 1 and 2, the walking measurement system 100 includes a chair 2, a marker 3, a laser range sensor (Laser Range Sensor, distance information acquisition unit) 10, a pressure sensor 15, and an electronic control device 20. The monitor 30 and the pole 40 are provided. The walking measurement system 100 can be suitably applied to, for example, a TUG test that is one of walking tests for evaluating functional movement ability that integrates walking ability, dynamic balance, agility, and the like. The walking measurement system 100 can perform quantitative walking measurement and walking ability evaluation.

  In the TUG test, the test subject 1 stands up from the state of sitting on the chair 2, walks toward the marker 3 such as a cone ahead of a predetermined distance (for example, 3 m), turns the marker 3, and then returns to the chair 2. Sit back on chair 2 again. The TUG test is one of the physical strength measurement manuals of the Ministry of Health, Labor and Welfare. In the TUG test, walking at the maximum walking speed that does not run is required. As the marker 3, various markers can be used as long as they serve as marks for the turn position. In the TUG test, one or a plurality of (here, two) poles 40 as reference members are used to align the marker 3 and the laser range sensor 10. The pole 40 has a cylindrical outer shape with a predetermined height. The predetermined height is a height at which the upper end of the pole 40 is located above the height position of the laser light emitted from the laser range sensor 10 (the height of the optical window portion of the laser range sensor 10). The shape of the pole 40 is not particularly limited, and various poles are used. The walking measurement system 100 measures and evaluates walking parameters in such a TUG test.

  The walking measurement system 100 can define a plane coordinate system having a straight line connecting the centers of the two poles 40 as one axis by detecting the two poles 40 with the laser range sensor 10. The floor surface to be installed can be this plane coordinate system. In this case, if the positional relationship between the two poles 40 and the marker 3 is strictly adjusted (specifically, from the midpoint of the line segment connecting the two poles 40, 3 m above the vertical line of this line segment) Even if the installation position and / or orientation of the laser range sensor 10 is roughly aligned, the relative positional relationship between the laser range sensor 10 and the marker 3 can be grasped strictly. It becomes possible.

  The walking measurement system 100 according to the present embodiment has a leg observation position (hereinafter simply referred to as “observation position”) that is an observation position of the leg by a characteristic pattern that is seen when both legs of the subject 1 are scanned by the laser range sensor 10. Are also tracked using a Kalman filter, and walking parameters are calculated based on the acquired movement trajectories of both legs.

  Specifically, in the gait measurement system 100, in order to reduce the loss or misidentification of the legs with respect to the situation where both legs of the subject 1 approach or the situation where the legs cannot be observed from the laser range sensor 10 temporarily. In addition, the leg observation position is acquired based on five types of observation patterns. In addition, by setting the effective area (gate) considering the leg state (free leg / support leg, speed, hiding), one-to-one correspondence between the leg observation position and both legs being tracked (correlation processing) )I do. In the following, “leg part” is also simply referred to as “leg”, “support leg part” is also simply referred to as “support leg”, and “free leg part” is also simply referred to as “free leg”. The “left leg” is also simply referred to as “left leg”, and the “right leg” is also simply referred to as “right leg”. “One leg” is also simply referred to as “one leg”, and “both legs” is also simply referred to as “both legs”.

  Actual human walking (the speed of both legs) changes periodically. In particular, in the TUG test, since the subject 1 is instructed to perform at the maximum walking speed that does not run, the speed change of both legs is large. Therefore, in the walking measurement system 100, an acceleration / deceleration model considering the walking phase is applied as a walking model. When a person walks, he / she performs a periodic motion such that one leg is a pivot leg (support leg) and the other leg (free leg) is swung. “Walking phase” is a classification of the periodic movement of both legs during walking of a person based on the swinging state of the support leg and the free leg (see, for example, FIG. 9).

  Further, since the TUG test includes a turn operation, a situation where one leg is always hidden from the laser range sensor 10 (hereinafter, also simply referred to as “hidden”) occurs. The distance data of the laser range sensor 10 is disturbed during the turn operation, and the leg observation position may not be correctly detected even if the observation pattern is used. In addition, both legs perform a turning motion in a turning phase during a turn motion from a linear motion in a forward phase prior to the turn motion. Therefore, the traveling direction changes rapidly. Therefore, as shown in FIG. 3 (a), leg portions are easily lost or misidentified due to changes in hiding or movement. Therefore, in the walking measurement system 100, correlation processing is performed in consideration of the periodicity of the human walking phase.

  When one leg cannot be observed from the laser range sensor 10, the predicted position of the Kalman filter is acquired as the observation position. Since the swing leg moves close to a circular motion during the turn motion, its moving direction changes from moment to moment. While the free leg is hidden by the supporting leg, as shown in FIG. 3B, the obtained leg position is used to predict the position of the free leg from the moving direction of the free leg observed immediately before the free leg is hidden. The measurement accuracy of the sensor deteriorates, and misidentification and loss of information easily occur. Therefore, a method for interpolating and correcting the leg position while the leg is hidden is necessary.

Therefore, in the walking measurement system 100, in the walking test in which there is a situation in which one leg is always hidden from the laser range sensor 10 as in the TUG test, both legs are tracked and there is no misidentification, and the measurement accuracy is improved. ,
・ Application of walking model considering walking phase,
・ Correlation processing considering periodicity of walking phase,
-Interpolation and / or correction of data while legs are hidden using spline curve-Relative distance of legs from laser range sensor 10 and / or observation noise based on observation state (observation pattern) Configuration,
To implement. Then, walking parameters for quantitatively evaluating the functional movement ability are calculated from the movement trajectories of both legs.

FIG. 4 is a plan view for explaining the laser range sensor 10 of the walking measurement system 100 of FIG. The laser range sensor 10 is a sensor for acquiring the movement locus of both legs. The laser range sensor 10 acquires two-dimensional plane distance information (hereinafter also simply referred to as “distance information”) that is a distance to an object around the sensor in a two-dimensional plane at a certain height. As shown in FIG. 1, FIG. 2 and FIG. 4, the laser range sensor 10 emits a laser beam (detection wave) L so as to scan along the horizontal direction, and based on the reflection state of the laser beam L. over time to obtain a two-dimensional plane distance information about the distance between the legs of the subject 1 that reflects the laser beam L F (right leg F _R and the left leg portion F _L).

  Specifically, the laser range sensor 10 emits the laser beam L and reflects the laser beam L with a rotating mirror, thereby scanning the laser beam L in a fan shape in the horizontal direction in the measurement region. More specifically, the laser range sensor 10 scans the laser beam L in a fan shape in the horizontal direction so that the subject 1 and the poles 40 installed on both sides of the subject 1 are included in the measurement region. Then, for example, the reflected light of the laser light L reflected by the leg F of the subject 1 is received, and the detection angle (scanning angle) of the reflected light and the time from the emission to reception of the laser light L (propagation time) are measured. Then, distance information including information related to the angle and distance from the leg F is detected.

  For example, the laser range sensor 10 scans the subject 1 with the laser light L before or after the test trial and detects leg information regarding the width of the leg F of the subject 1 (the length corresponding to the diameter of the shin part). . The laser range sensor 10 outputs the detected distance information and leg information to the electronic control unit 20.

  The laser range sensor 10 is configured such that the height of the light window that emits the laser light L can be adjusted. The laser range sensor 10 is installed so that the laser light L is emitted at a height position corresponding to the height of the shin part of the subject 1 (that is, the height from the ankle to the lower knee). The height of the optical window portion of the laser range sensor 10 is set to a position higher than the height of the marker 3 so that the laser beam L is not reflected by the marker 3. The height of the optical window part of the laser range sensor 10 is such that the leg part F can be observed even when the leg part F of the subject 1 leaves the floor during the swing period, and the width of the leg part F is maximum. In consideration of the average height, for example, it is 0.15 m to 0.27 m (here, 0.27 m) from the floor (ground).

  The laser range sensor 10 is disposed in a space formed below the seat portion of the chair 2 so that the emission direction of the laser light L faces the marker 3 side. The laser range sensor 10 emits laser light L to the subject 1 from the rear side of the subject 1 approaching the marker 3 (moving away from the chair 2), that is, from the front side of the subject 1 moving away from the marker 3 (approaching the chair 2). Irradiate. The laser range sensor 10 is installed so that the emission direction of the laser light L is horizontal with respect to the floor surface.

  As the laser range sensor 10, a sensor having a ranging range of 0.1 to 30 m, a ranging angle of 270 deg, a ranging accuracy of ± 30 mm, an angular resolution of 0.25 deg, and a sampling period of 25 ms / scan is used. Yes. Incidentally, the laser range sensor 10 is not limited in its type and specification (spec), and various types can be used according to the measurement environment, for example. Note that the laser range sensor 10 may be surrounded by a housing in order to suppress the adverse effects of EMI (electromagnetic noise), for example.

  The pressure sensor 15 is a sensor that detects a rising time when the subject 1 rises from the chair 2 and a sitting time when the subject 1 is seated on the chair 2. The pressure sensor 15 is provided on the seat surface of the chair 2. The pressure sensor 15 detects pressure information related to the applied pressure and outputs the pressure information to the electronic control unit 20. The pressure information of the pressure sensor 15 here is a voltage value, and the voltage value is A / D converted via the microcomputer board and input to the electronic control unit 20.

  The electronic control unit 20 performs a calculation based on the distance information acquired by the laser range sensor 10 to identify the leg position that is the position of the leg F of the subject 1 over time, and acquires the walking characteristics of the subject 1. The electronic control unit 20 outputs a start signal (described later) of at least one of sound (voice) and display to the subject 1. The start signal is output from the electronic control unit 20 according to the operation of the measurer. The electronic control unit 20 stores the time of the start signal. The start signal may be executed by a device different from the electronic control device 20 or a measurer. In this case, the time of the start signal is input to the electronic control device 20 by a device different from the electronic control device 20 or a measurer.

  For example, a personal computer or a dedicated control computer is used as the electronic control device 20 and includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like. The electronic control device 20 includes a sensor data acquisition unit 21, a storage unit 22, a seating determination unit 23, and a data analysis unit 24 as functional elements.

  The sensor data acquisition unit 21 acquires distance information detected by the laser range sensor 10 and pressure information detected by the pressure sensor 15 in synchronization. The sensor data acquisition unit 21 may acquire the distance information in synchronization with the time of the start signal stored in the electronic control device 20. The sensor data acquisition unit 21 acquires leg information detected by the laser range sensor 10. The sensor data acquisition unit 21 outputs the acquired distance information, leg information, and pressure information to the storage unit 22. The sensor data acquisition unit 21 outputs the acquired pressure information to the seating determination unit 23.

  The storage unit 22 stores the distance information and pressure information acquired by the sensor data acquisition unit 21 in association with the processing time (sampling time). The storage unit 22 stores the leg information of the subject 1 acquired by the sensor data acquisition unit 21. The seating determination unit 23 determines whether the subject 1 is seated based on the pressure information of the pressure sensor 15 acquired by the sensor data acquisition unit 21. For example, the seating determination unit 23 determines that the user is seated when the voltage value from the pressure sensor 15 rises larger than a threshold value. The seating determination unit 23 determines that the user is not seated when the voltage value from the pressure sensor 15 is less than or equal to the threshold value. The seating determination unit 23 outputs the determination result to the storage unit 22, and the storage unit 22 stores the determination result.

  The data analysis unit 24 calculates and acquires the position and speed of the leg F of the subject 1 in association with time based on at least the distance information acquired by the laser range sensor 10. The data analysis unit 24 obtains the movement trajectory and walking parameters of both legs by analyzing the stored data stored in the storage unit 22 after the test trial. The data analysis unit 24 includes a leg detection unit 26, a leg tracking unit 27, and a walking parameter calculation unit 28. The walking parameters include general walking parameters and walking parameters related to the TUG test.

  The general walking parameter is a walking parameter for evaluating a fall risk in a general walking test, and includes a landing position, a walking speed [m / second], a step count [step], a walking rate [step / second], Step length [m], stride length [m], step [m], and support leg time (one leg support leg time [s] and both leg support leg time [s]) are included.

  The walking parameters related to the TUG test are parameters for evaluating the risk of falls in the TUG test. The reaction time (rise time) [seconds], TUG performance time [seconds], turn direction, first step time (first step of getting out of bed) Time) [seconds] and the shortest distance [m] from the marker 3.

  The walking parameters are the forward phase after starting the movement and before turning around the marker 3, the turning phase while turning around the marker 3, and the time from turning around the marker 3 until sitting. It is calculated by being classified into a return phase. In other words, the walking parameters are calculated for each of these phases. At this time, for the stride length, stride length, step distance, single leg support leg time, and both leg support leg time, the average, standard deviation, and coefficient of variation are calculated for each phase.

  The landing position is a position where the leg F has landed on the floor surface. The walking rate is the number of steps per unit time. One leg support leg time is the time from landing (flooring) to leaving the same leg. Both legs supporting leg time is the time from the landing of one leg to the leaving of the other leg. The stride is the distance from the landing position (for example, the heel position) of one leg to the landing position of the other leg. The stride length is the distance between adjacent landing positions for the same leg. The step is the distance between the heels of both legs on the front face (when the subject 1 is viewed from the front).

  The reaction time is the time from the start time of the walking test (the time of the start signal) until the subject 1 stands up (the buttocks leave the seat of the chair 2). The TUG performance time is the performance time of the TUG test, and is the time from the start time to the end of the walking test (from the time when the subject 1 seated on the chair 2 starts moving to the time of sitting again). The first step time is the time from the start time of the walking test until the subject 1 raises the leg F of the first step (gets out of bed). The shortest distance from the marker 3 is the shortest distance between the center of the marker 3 and the movement locus of the leg F.

[Leg detection]
5 to 8 are diagrams for explaining examples of observation patterns used for detection of leg observation positions. FIG. 5A and FIG. 7A are diagrams for explaining an example of an SL (Single Leg) pattern. FIG. 5B and FIG. 7B are diagrams for explaining an example of an LT (two Legs Together) pattern. FIG. 5C and FIG. 8A are diagrams for explaining examples of FS_O (Forward Straddle Observable) patterns. FIGS. 6A and 8B are diagrams illustrating an example of an FS_U (Forward Straddle Unobservable) pattern. FIG. 6B is a diagram illustrating an example of a UO (Unobservable) pattern.

The leg detection unit 26 detects a leg observation position that is an observation position of the leg F of the subject 1. Specifically, considering that the two-dimensional plane distance information of the leg F has a characteristic shape, the edge detection process is performed on the two-dimensional plane distance information to detect the position of the edge E of the leg F. A leg observation position is detected as a candidate point of the leg F from the position of the edge E by pattern recognition using a plurality of observation patterns. At this time, using the legs information about the width w _l leg F of the subject 1, and the known width w _l leg F. The leg detection by the leg detection unit 26 will be specifically described below.

When the subject 1 exists in the measurement range of the laser range sensor 10 and there is no obstacle between the laser range sensor 10 and the subject 1, the two-dimensional plane distance information acquired by the laser range sensor 10 is as shown in FIG. It becomes a characteristic shape as shown in FIG. The leg detection unit 26 calculates a leg observation position (observed value) y ^f _k (corresponding to the vector on the left side of the following formula 8) by a leg detection method considering five types of observation patterns.

Leg detector 26, first, the i-th distance information from the right acquired by the laser range sensor 10 and l _i, the width w _l leg F of the two-dimensional plane distance information of the laser range sensor 10 adjacent the subject 1 On the other hand, when the following formula 1 is satisfied, an edge e ^h _m (h = B, F) is detected. At this time, if l _i > l _{i + 1} , the edge e ^B _m = i and the edge e ^F _{m + 1} = i + 1 are set. In the case of l _i <l _{i + 1} , the edge e ^F _m = i and the edge e ^B _{m + 1} = i + 1. The upper suffix B indicates the edge E on the back side with respect to the laser range sensor 10, and the upper suffix F indicates the edge E on the near side. The subscript m (m = 1,..., M _k ) indicates the mth detected edge E. M _k is the total number of edges E detected at time k.

Then, four edge ^e _h n ^consecutive, based on _{^{e h n + 1, e h}} n + 2, e h n + 3 positional relation and the edge of _{^{e h n + 1, e h}} n + 2 between the width _{w e} of, classified into five types of observed pattern Then, the leg observation position y _k that is an observation value is calculated. Here, n = 1,..., M _k −3, and the subscript k indicates the observation position at time k.

As shown in FIGS. 5A and 7A, the observation pattern O1 as the SL pattern is an observation pattern when the leg F can be completely observed from the laser range sensor 10 alone. In the observation pattern O1, the combination of the edges E satisfies (e ^B _n , e ^F _{n + 1} , e ^F _{n + 2} , e ^B _{n + 3} ), and the width between the edges E satisfies 0.5 w _l <w _e ≦ 1.5 w _l . Fulfill. At this time, the legs observation position y _k, in consideration of the width w _l leg F of the subject 1 is obtained as shown in Figure 5 (a). For example, the leg observation position y _k ^j is obtained from a line passing through (e ^F _{n + 1} + e ^F _{n + 2} ) / 2 and 1/2 of the width w _l of the leg F as shown in the figure.

As shown in FIG. 5B and FIG. 7B, the observation pattern O2 as the LT pattern is an observation pattern when the leg portions F are aligned and can be completely observed from the laser range sensor 10. Observed pattern O2, a combination of edge E is ^{_{(e B n, e F n}} + 1, e F n + 2, e B n + 3), (e F n, e B n + 1, e F n + 2, e B n + 3) or ^(e _{B n} , it fulfills ^{_{e F n + 1, e B}} n + 2, e F n + 3), the width between the edge E satisfies _{1.5w l <w e <3.0w l} . At this time, the leg observation position y _k is obtained as shown in FIG. 5B, considering that both legs of the subject 1 are aligned. For example, the leg observation position y _k ^{j + 1} is obtained from a line passing through (e ^F _{n + 1} + 3e ^F _{n + 2} ) / 4 and 1/2 of the width w _l of the leg F as shown in the figure. For example, the leg observation position y _k ^j is obtained from a line passing through (3e ^F _{n + 1} + e ^F _{n + 2} ) / 4 and 1/2 of the width w _l of the leg F as shown in the figure.

As shown in FIGS. 5C and 8A, the observation pattern O3 as the FS_O pattern has a stepped pattern due to the influence of one leg F or a cane, and the width between the positions of the edges E Is an observation pattern in the case where the width of the leg F of the subject 1 is ½ or more of the width F _l and the center position of the leg F can be observed from the laser range sensor 10. Observed pattern O3 are combinations ^{_{(e F n, e B n}} + 1, e F n + 2, e B n + 3) of the edge E or ^{_{(e B n, e F n}} + 1, e B n + 2, e F n + 3) fulfills edge width between E satisfies _{0.5w l ≦ w e <1.5w l} . At this time, the leg observation position y _k is obtained as shown in FIG. 5C in consideration that the center position of the leg F of the subject 1 can be observed from the laser range sensor 10. For example, the leg observation position y _k ^j is obtained from a line passing through (e ^B _{n + 1} + e ^F _{n + 2} ) / 2 and 1/2 of the width w _l of the leg F as shown in the figure.

As shown in FIGS. 6A and 8B, the observation pattern O4 as the FS_U pattern has a width between the edges E that is the width of the leg F of the subject 1 with respect to the observation pattern O3 that is the FS_O pattern. less than half the w _l, and, an observation pattern when the center position of the leg portion F can not be observed directly from the laser range sensor 10. Observed pattern O3 are combinations ^{_{(e F n, e B n}} + 1, e F n + 2, e B n + 3) of the edge E or ^{_{(e B n, e F n}} + 1, e B n + 2, e F n + 3) fulfills edge The width between E satisfies 0.3 w _l <w _e <0.5 w _l . At this time, since the center position of the observed pattern unlike O1~O3 leg F can not be directly observed by laser range sensor 10, the legs observation position y _k is virtually calculated as shown in FIG. 6 (a) The For example, the leg observation position y _k ^j is _{obtained from a} line passing through e ^B _{n + 1} and a line passing through e ^F _{n + 2} and 1/2 of the width w ₁ of the leg F as shown in the figure. A hollow square in FIG. 8B indicates a virtual edge.

  As shown in FIG. 6B, the observation pattern O5 as the UO pattern is obtained when one leg F is hidden by the other leg F or when the leg F is raised higher than the laser range sensor 10. This is an observation pattern when it cannot be observed.

Incidentally, even in a situation where the leg portion F can not be observed directly from the laser range sensor 10, by obtaining the virtually legs observation position y _k defines the observed pattern O4 is FS_U pattern, measurement of the position of the legs F Improvement in accuracy is expected. In the detection of the leg observation position y _k , it can be determined that the position of the edge E that does not satisfy the observation patterns O1 to O4 is not the leg F. Further, in order to avoid unnecessary correlation processing, an area that is a predetermined length wider than the area of the walking path is set as a measurement area, and the leg observation position y _k outside the measurement area and the position of the edge E corresponding thereto are excluded.

In the detection of the leg observation position y _k , not the center between the positions of the edge E but, for example, a position that is inside the position ½ of the width w ₁ of the leg F from the position of the edge E is the leg observation position y. _k is detected. As described above, by detecting the leg observation position y _k of the leg F according to the plurality of observation patterns O1 to O5, even when one leg F is overlapped and hidden when viewed from the laser range sensor 10, The leg F can be detected suitably.

[Leg Tracking]
The leg tracking unit 27 predicts the position and velocity of the leg F by a Kalman filter prediction process, performs a correlation process with the leg observation position y _k acquired by the leg detection unit 26, and tracks both legs. When there are a plurality of tracking targets and leg observation positions y _k , the leg tracking unit 27 has an effective region (gate) centered on the predicted position in order to limit the leg observation positions y _k that can be correlated with the tracking target. And is associated with the leg observation position y _k existing in the gate.

The leg tracking unit 27 applies the walking model (acceleration model) in consideration of the walking phase, the correlation processing in consideration of the periodic change of the human walking phase, and the legs are hidden in the turning phase or the like. Data interpolation and correction using the Catmull-Rom spline curve at the leg observation position y _k of the, and setting of observation noise based on the relative distance of the leg F and the observation state (observation patterns O1 to O4) . Thereby, in a walking test in which there is a situation where one leg is always hidden from the laser range sensor 10 at high speed, tracking of both legs without missing or misidentification is realized and measurement accuracy is improved. The leg tracking by the leg tracking unit 27 will be specifically described below.

  The leg tracking unit 27 includes a leg estimation unit 27a, a supporting leg swing leg determination unit 27b, a walking phase determination unit 27c, an acceleration input estimation unit 27d, a predicted position calculation unit 27e, a correlation processing unit 27f, a hidden data interpolation unit 27g, And an observation noise setting unit 27h.

  The leg estimation unit 27a performs a prediction process using a walking model in consideration of the walking phase, and estimates the position and speed of the leg F. In the prediction process, a Kalman filter is used.

  FIG. 9 is a diagram illustrating a walking model. When a person walks, he walks by swinging the other leg (free leg) while using one leg as a pivot leg (support leg). Both legs move periodically while changing the free leg and the supporting leg by landing. As shown in FIG. 9, it is possible to generate a walking model that simplifies human walking by paying attention to changes in the moving speed v and acceleration a of the leg F.

In the gait model, the following six gait phases are defined, including when stationary. Further, the walking phase changes periodically (for example, the phase always shifts to phase 2 after phase 1 and does not shift to phase 3 and phase 4). Therefore, as will be described later, erroneous identification in the correlation processing is reduced in consideration of periodic changes in the walking phase.
・ Phase 0: Both legs are support legs when stationary ・ Phase 1: Left leg is free leg (acceleration, swing-up movement), Right leg is support leg ・ Phase 2: Left leg is free leg (deceleration, facing landing) Motion), right leg is support leg ・ Phase 3: left leg is support leg, right leg is free leg (acceleration, swinging motion)
・ Phase 4: The left leg is the support leg, the right leg is the swing leg (motion toward deceleration, landing)
・ Phase 5: Both legs are swing legs (not seen during walking)

  In the TUG test, since subject 1 is instructed to perform at the maximum walking speed that does not run, the speed change of both legs is large. Therefore, a model that performs acceleration / deceleration is applied as a walking model.

  When tracking the center of gravity of a person, nonlinear models that include angle and angular velocity as state variables are often used. On the other hand, in the TUG test, it is desirable to apply a linear movement model having no directivity with respect to the moving direction because the moving direction of both legs changes rapidly during the turn operation. Therefore, a linear movement model having no directivity in the movement direction is applied as the walking model.

The discrete time state equation of each leg F when considering a walking model with acceleration / deceleration is given by the following Equation 2. In the following Equation 2, the superscript f indicates the left leg and the right leg, respectively, with f = L and R. The state variable has a value shown in Equation 3 below. Equation 4 below is the position and speed of each leg F of the subject 1 at time k.

u ^f _k is an acceleration input corresponding to the walking phase, and is represented by the following Equation 5. That is, the filtering process is based on the acceleration corresponding to the walking phase. It is assumed that the system noise Δx ^f _k of the following Equation 6 is given by normal white noise having an average of 0 and a covariance Q ^f _k . At this time, the matrices A, B _u , and B of the state equation are expressed by the following formula 7.

Further, if the leg observation position y ^f _k that is the observation position of each leg F acquired using the laser range sensor 10 is expressed by the following formula 8, the observation equation is given by the following formula 9. Here, it is assumed that w ^f _k is normal white noise having an average of 0 and a covariance R ^f _k . At this time, the matrix C is expressed by Equation 10 below.

Based on the speed of the leg F, the support leg free leg determination unit 27b performs support leg swing leg determination to determine whether the leg F is a support leg or a free leg. The support leg free leg determination unit 27b determines that the right leg is a support leg when the following Expression 11 is satisfied, and determines that it is a free leg when the following Expression 12 is satisfied. v _{st_th} indicates the speed threshold of the supporting leg, and v _{sw_th} indicates the speed threshold of the free leg. The support leg swing leg determination unit 27b similarly performs the support leg swing leg determination by the following formula 11 and the following formula 12 in which the right leg and the left leg are interchanged.

The walking phase determination unit 27c determines the walking phase based on the relative positional relationship and the speed of both legs, and specifies which of the phases 0 to 5 is the walking phase. Specifically, first, when both legs are supporting legs, the phase is specified as zero. Next, when the left leg is a free leg and the right leg is a support leg, the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg (see Equation 13 below) takes a positive value. Sometimes it is specified as phase 1, and when it takes a negative value, it is specified as phase 2. Similarly, when the left leg is a supporting leg and the right leg is a free leg, when the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg takes a positive value, Specify phase 4 when taking a negative value. Finally, when both legs are swing legs, the phase 5 is specified. If the walking phase is determined to be phase 5 by the walking phase determination unit 27c, the data analysis unit 24 determines that the subject 1 is traveling, for example, a notice that the subject 1 is traveling via the monitor 30. Output a call.

The acceleration input estimation unit 27d estimates the acceleration input in the human walking model based on the walking phase and past data. The acceleration input estimation unit 27d sets u ^L _k = 0 and u ^R _k = 0 when the walking phase is phase 0. When the walking phase is phase 1, the acceleration input estimation unit 27d estimates the acceleration input u ^L _k as shown in the following formula 14. When the walking phase is phase 2, the acceleration input estimation unit 27d estimates the acceleration input u ^L _k as shown in the following formula 15. g ^f _k is an acceleration function, and is calculated as an average value of norms of acceleration vectors in the free leg for the past N _ac steps. The same processing is performed in the phases 3 and 4 where the right leg is the free leg.

The predicted position calculation unit 27e is a predicted leg position (hereinafter, also simply referred to as “predicted position”) y ^ ^f _{k / k−} which is the predicted position of the leg F at time k according to the state equation shown in Equation 2 above. ₁ is obtained from Equation 16 below. _{^{Here, x ^ f k / k-}} 1 is pre-state estimated value at time _{^{k, x ^ f k-1}} / k-1 is a posteriori state estimate at time k-1.

FIG. 10 is a diagram for explaining the detection of the predicted leg position y ^ ^f _{k / k−1} using the gate G. The circle in the figure indicates the leg position of the free leg, and the double circle in the figure indicates the leg position of the support leg. In the drawing, the predicted leg position y ^ ^f _{k / k−1} of the leg F at time k is indicated by “x”.

The correlation processing unit 27f performs a correlation process between the leg observation position y ^j _k and the leg predicted position y ^ ^f _{k / k−1} in consideration of the periodicity of the walking phase, and the leg portion is determined based on the relative positional relationship. Identify the location. Correlation processing section 27f, as shown in FIG. 10, in order to limit the leg observation position y ^j _k, which may correspond to the tracked, effective around the leg predicted position y ^ f ^{_k / _k-1} region The gate G which is and is associated with the leg observation position y ^j _k in the gate G. When the leg observation position y ^j _k exists in the gate G, the correlation processing unit 27f associates only the leg observation position y ^j _k in the gate G, for example, the leg observation position y ^j _k . It is specified as the leg position. Further, when the leg observation position ^y _{j k} is not in the gate G, and associates the leg predicted position ^y _{^ f k / k-1,} for example, the leg predicted position ^y _{^ f k / k-1} foot Specify as part position.

In the correlation processing unit 27f, a cost function related to the association between the observation position for performing the correlation processing using the gate G for a plurality of tracking targets and the both legs to be tracked is shown in the following Expression 17. d _{f, j} is the correspondence between the j-th observation position y ^j _k and the predicted leg position y ^ ^f _{k / k−1} of each leg (f = L, R indicates the left leg and the right leg, respectively). The cost is given by Equation 18 below.

j = 0 indicates that the observation position included in the gate G is a false detection. λ _{f, j} is the Mahalanobis distance, and is given by the following Equation 19. S ^f _k is a covariance matrix of observation prediction errors (y ^j _k −y ^ ^f _{k / k−1} ). d _max is set to a value larger than that of the gate G shown in Equation 20 below. When the following Expression 20 is satisfied, it is determined that the leg observation position y ^j _k is included in the gate of the leg prediction position y ^ ^f _{k / k−1} . G is a gate, and is determined based on a chi-square (χ ² ) distribution with ^two degrees of freedom in the case of this embodiment in which the leg observation position y ^j _k is a two-dimensional vector and uses a Kalman filter. For example, when the content probability P _G = 0.95, G = 5.99.

Here, in the TUG test, it is assumed that one leg is hidden from the laser range sensor 10 in the turning phase, and the traveling direction changes rapidly at that time. Further, in the turning phase, it is assumed that the distance information of the laser range sensor 10 is disturbed and the observation position (leg observation position y ^j _k ) is erroneously detected when the leg pattern is used. For this reason, as shown in FIG. 3A, erroneous identification of both legs is likely to occur due to hiding or a change in the moving direction. Therefore, the correlation processing unit 27f executes correlation processing in consideration of the periodicity of the person's walking phase as described below, determines the probability of the observation position, and deviates from the periodicity of the person's walking phase. The observed position is regarded as a false detection.

That is, in order to prevent erroneous association with respect to erroneous detection of the observation position, the cost df _{, j} is designed in consideration of the periodic change of the human walking phase. First, a Kalman filter filtering process is performed once on all observation positions associated with the predicted positions predicted by the predicted position calculation unit 27e. Thereby, the position and speed of the leg F relating to the observation position are estimated. Based on the estimated position and speed of the leg F, the walking phase (phases 0 to 5) is determined and specified as in the determination in the walking phase determination unit 27c. If the identified walking phase deviates from a periodic change in the person's walking phase, the observation position is determined to be erroneous detection, and the cost is d _{f, j} = d _max . Then, the association that minimizes c _{a, b} while satisfying c _{a, b} <2d _max is selected.

The phase change deviating from the periodicity of the human walking phase is defined as follows.
・ Phase 0⇒Phase 5
・ Phase 1⇒Phase 0, Phase 3, Phase 4, Phase 5
・ Phase 2⇒Phase 1, Phase 4, Phase 5
・ Phase 3⇒Phase 0, Phase 1, Phase 2, Phase 5
・ Phase 4⇒Phase 2, Phase 3, Phase 5

  Fig.11 (a) is a figure which shows the movement locus | trajectory of both the leg parts acquired by the conventional method. FIG. 11B and FIG. 11C are diagrams illustrating virtual calculation of the observation position using the Catmull-Rom spline curve. In FIG. 11A, the movement locus 60R indicates the movement locus of the right leg, and the movement locus 60L indicates the movement locus of the left leg. On each movement locus 60R, 60L, a large point indicates a support leg, and a small point indicates a free leg.

  Generally, when the leg F cannot be observed from the laser range sensor 10, the predicted position by the Kalman filter is specified as the leg position. However, during the turn operation of the TUG test, the direction of the swinging leg changes from moment to moment because the free leg moves close to a circular motion. Therefore, if the prediction is made based on the direction immediately before the free leg is hidden by the support leg, the prediction may be greatly different from the actual movement (inside the circle in FIG. 11A). reference). As a result, misidentification and loss of information are likely to occur. Therefore, a method for interpolating and correcting the position at the time of hiding (that is, while one leg cannot be observed from the laser range sensor 10) is necessary.

  Therefore, the hidden-time data interpolation unit 27g virtually calculates an observation position from when it cannot be observed once until it is observed again (temporarily undetectable) using a curve. The curve is a curve obtained based on the observation position before observation and the observation position after observation again. Although it does not specifically limit as a curve, For example, a spline curve and a Catmull-Rom spline curve are mentioned. Then, the hidden data interpolation unit 27g considers the walking model and updates the state quantity such as the covariance matrix of the Kalman filter, so that the prediction process using the Kalman filter is performed again using the virtually calculated observation position. Perform filtering processing. This interpolates and corrects the observation position while hidden.

As shown in FIG. 11B and FIG. 11C, the hidden data interpolation unit 27g uses the Catmull-Rom spline curve that can define a smooth curve from the four points that have passed, Is virtually calculated. If N _uo observation positions from time k to time (k + N _uo +1) are not obtained, two observation positions y _k−Nbo, y _k before hiding and two observation positions after hiding Using y _{k + Nu + 1 and} y _{k + Nuo + Nao + 1} , a virtual observation position y ′ _{k + i} during which the observation could not be performed is calculated as shown in the following Equation 21. The dash symbol indicates the observation position calculated virtually. The example shown in FIG. 11B shows the case where N _bo = 1 and N _ao = 1, and the example shown in FIG. 11C shows the case where N _bo > 1, N _ao > 1 (here, N _Bo = 2 and N _ao = 2). In the example shown in FIG. 11B, the observation positions y ′ _{k + 1} and y ′ _{k + 2} are virtually calculated using the observation positions y _k−1 , y _k , y _{k + 3} and y _{k + 4.} To do. In the example shown in FIG. 11C, the observation positions y ′ _{k + 1} and y ′ _{k + 2} are virtually calculated using the observation positions y _k−2 , y _k , y _{k + 3} and y _{k + 5.} To do.

The hidden data interpolation unit 27g does not use the virtually calculated observation position as the measurement value, but considers the walking model and updates the state quantity such as the covariance matrix of the Kalman filter, so that the Kalman filter is re-started from time k. The prediction process and the filtering process are performed. Specifically, the hidden-time data interpolation unit 27g tentatively determines the predicted position until the leg F can be observed from the laser range sensor 10 (detected from the distance information) until it can be observed. And Then, after the observation position becomes observable, the process returns to the point where observation becomes impossible, and the prediction process and the filtering process are performed again based on the calculated virtual observation position y ′ _{k + i} . The hidden data interpolation unit 27g is expected to not only improve the accuracy of the measurement value but also reduce misidentification by correcting the observation position while hidden.

Each graph shown in FIG. 12 includes RMS (Root Mean Square) measurement errors of the leg observation position y _k in each of the x direction and the y direction, an average value thereof, and an approximate straight line of the average value. Are shown in each observation pattern. It is conceivable that the accuracy of the observation position changes according to the distance from the laser range sensor 10 and the observation state (observation pattern).

Therefore, the observation noise setting unit 27h changes the value (standard deviation) of the observation noise R ^f _k based on the relative distance l from the laser range sensor 10 and the observation state (observation pattern). At this time, the standard deviation σ ^p _{x, y} of the observation position in the x direction and the y direction is approximated linearly as shown in Equation 22 below. Superscripts p = SL, LT, FS_O, and FS_U indicate each observation pattern (see FIGS. 5 and 6). The measurement accuracy at rest is obtained by comparison with a three-dimensional motion analysis device, and the coefficient ^ap is set.

As shown in FIG. 12, in each observation pattern, the relative distance from the laser range sensor 10 is changed to be stationary for about 5 seconds, and the RMS of the measurement error of the leg observation position y _k in each of the x direction and the y direction is obtained. An approximate straight line of the average value is obtained. The obtained linear approximation coefficient is set as the coefficient ^ap . By changing the observation noise according to the distance from the laser range sensor 10 and the observation state (observation pattern), it is expected that the estimation accuracy of the filtering process is improved.

[Calculation of walking parameters]
The walking parameter calculation unit 28 calculates a walking parameter from the acquired movement trajectory of both legs. The walking parameter calculation unit 28 includes a support leg time calculation unit 28a, a landing position calculation unit 28b, a stride, step and stride length calculation unit 28c, a first step time calculation unit 28d, a reaction time calculation unit 28e, A TUG performance time calculation unit 28f.

The support leg time calculation unit 28a obtains the one leg support leg time and the both leg support leg time. In the present embodiment, in order to obtain the one leg support leg time and the both leg support leg time, based on the analysis results of the subject experiment conducted by attaching pressure sensors to the heel and metatarsal part of both legs, The bed leaving time is strictly calculated after the movement trajectory is acquired. First, the time when the state of the leg F changes from the supporting leg to the free leg is calculated as the bed leaving time t ^step _FO . Next, as shown in FIG. 13, the time when the speed in the swing leg after leaving the floor takes the maximum value v ^sw _max and the speed becomes equal to or less than a ^v _IC · v ^sw _max Calculated as time t ^step _IC . From the measurement experiment using the pressure sensor 15, in the present embodiment, a ^v _IC = 0.5 is set. Then, the difference between the bed leaving time and the floor contacting time is obtained as the one-leg supporting leg time. Further, the difference between the bed leaving time and the landing time one processing time before the leg F on the opposite side is obtained as the both leg supporting leg time.

  The landing position calculation unit 28b first determines the landing position in order to calculate the stride length, stride length, and step length. It is difficult for the landing position calculation unit 28b to directly measure the position of the leg F because the laser range sensor 10 acquires the movement trajectory of the leg F at the height of the shin. Therefore, the landing position calculation unit 28b takes into account that the leg F is perpendicular to the floor surface at the time when the speed of the leg F becomes the minimum in the support leg, and the leg F at that time. Is calculated as the landing position.

FIG. 14 is a diagram for explaining the stride length, step length, and stride length. As shown in FIG. 14, the stride, stride, and stride length calculation unit 28 c calculates a stride 71, a stride w _3, and a stride length 72 based on the calculated landing position 70. The stride 71 and Fu隔w ₃ and stride length 72, for example, in the example with reference to the landing position 70a, can be calculated as follows. The stride 71 is calculated as the distance from the landing position 70b of the leg F on the opposite side of the landing position 70a one processing time before the landing position 70a. Fu隔w ₃ shows one processing unit time before, and single treatment time after the opposite leg F of the landing position 70b, relative to a straight line connecting the 70c, is calculated as the distance of a perpendicular line from the landing zone 70a. The stride length 72 is calculated as the distance from the landing position 70d of the same leg F one processing time before the landing position 70a.

  The first step time calculation unit 28d calculates a time difference from the time of the start signal to the time when the leg F of the first step leaves the floor as the first step time. The reaction time calculation unit 28e calculates, as the reaction time, the time at which the voltage value from the pressure sensor 15 installed on the chair 2 fluctuates from the sitting state (the time when the voltage value falls below the threshold). That is, the rising of the subject 1 from the chair 2 is detected based on the pressure information detected by the pressure sensor 15. Then, the reaction time calculation unit 28e calculates the time difference from the time of the start signal to the reaction time as the reaction time.

  The TUG performance time calculation unit 28f uses the change in the voltage value of the pressure sensor 15 when the subject 1 sits on the chair 2 again, and sets the rise time of the voltage value as the end time. That is, the seating of the subject 1 on the chair 2 is detected based on the pressure information detected by the pressure sensor 15. Then, the TUG performance time calculation unit 28f calculates the time difference from the start signal to the end time as the test performance time of the TUG test.

  The walking parameter calculation unit 28 calculates the shortest distance from the marker 3 based on the movement trajectory of both legs and the center position of the marker 3. The walking parameter calculation unit 28 calculates other general walking parameters from the acquired movement trajectory of both legs.

  Returning to FIG. 1, the monitor 30 is an output unit that outputs an analysis result analyzed by the data analysis unit 24. The monitor 30 displays the analysis result analyzed by the data analysis unit 24 and feeds back to the subject 1. The monitor 30 can also output distance information, leg information, and pressure information acquired by the sensor data acquisition unit 21 or stored in the storage unit 22. The monitor 30 may be configured to output sound or the like instead of or in addition to display. Further, instead of or in addition to the monitor 30, a printer that prints out the calculation result or the like on a paper medium may be provided. The monitor 30 may be provided integrally with the electronic control unit 20, and may be a display unit such as a personal computer or a dedicated control computer.

  Next, the case where the walking characteristic in the TUG test is measured using the walking measurement system 100 of the present embodiment will be described with reference to the flowcharts shown in FIGS. 15 and 16.

As shown in FIG. 15, the measurement process in the TUG test is divided into three phases: calibration before trial, measurement during trial of the TUG test, and data analysis after trial. In the calibration phase, alignment of the laser range sensor 10 with two poles 40, and, the measurement of the width w _l leg F of the subject 1 it is performed. After the calibration is completed, for example, the electronic controller 20 notifies the subject 1 with a sound of a start signal (instruction to start the TUG test), and proceeds to the measurement phase. In the measurement phase, the distance information of the laser range sensor 10 and the pressure information of the pressure sensor 15 are acquired and stored in synchronization. Then, after the seating of the subject 1 is detected from the pressure information, the process proceeds to the analysis phase. In the analysis phase, the data analysis unit 24 tracks both legs based on the stored data to acquire a movement trajectory, and calculates a walking parameter based on the acquired movement trajectory. This will be specifically described below.

  First, calibration is executed before the subject 1 starts the operation. In calibration, scanning by the laser range sensor 10 is executed. That is, the laser beam L is emitted from the laser range sensor 10 so as to scan along the horizontal direction, and distance information is acquired based on the reflection state of the laser beam L (S1). Thereby, the positions of the leg F and the two poles 40 of the subject 1 as shown in FIG. 4 are observed.

From the observed width between the edge position of the leg portion F, the width of the leg portion F of the subject 1 (leg information) w _l is detected (S2). The detected leg information is stored in the storage unit 22. By determining the positional relationship between the two poles 40 with the laser range sensor 10, whether or not the laser range sensor 10 is installed at an appropriate position and direction with respect to the marker 3 is confirmed by a measurer or the like. Then, alignment of the laser range sensor 10 is performed (S3). As described above, the installation accuracy of the laser range sensor 10 may be rough as long as the relative positional relationship between the two poles 40 and the marker 3 is strictly determined.

  Until the countdown at the start of the TUG test is completed, a series of processes in S1 to S3 is repeatedly performed (S4). When the countdown of the start of the TUG test is completed, a start signal is notified to the subject 1 and the TUG test is started (S5).

  During the trial of the TUG test, scanning by the laser range sensor 10 is executed, and distance information regarding the distance of the leg F of the subject 1 is acquired over time by the sensor data acquisition unit 21 (S6). The sensor data acquisition unit 21 acquires pressure information from the pressure sensor 15 (S7). Then, the seat determination unit 23 determines whether or not the subject 1 is seated based on the pressure information (S8). When it is determined that the subject 1 is not seated, it is still in the middle of the TUG test and the process proceeds to S6 again. When it is determined that the subject 1 is seated, the following data analysis is performed on the assumption that the TUG test has ended.

  In the data analysis unit 24, leg detection by the leg detection unit 26 and leg tracking by the leg tracking unit 27 are executed for the stored distance information (S9 and S10). That is, as described in detail in the above description of each component, the leg detection unit 26 detects the observation position of the leg F. The leg estimation unit 27a performs a filtering process using a walking model (see FIG. 9) in consideration of the walking phase, and the position and speed of the leg are estimated and acquired. At this time, the support leg swing leg determination unit 27b determines the support leg swing leg determination, the walk phase determination unit 27c determines the walking phase, and the acceleration input estimation unit 27d estimates the acceleration input.

  The correlation processing unit 27f associates the observation position in the gate G with the predicted position as the center, and acquires the position of the leg F. At this time, the filtering process is performed once for all the associated observation positions, the position and speed of the leg F are estimated, the walking phase is specified, and the change in the walking phase is abnormal (previous processing time) In this case, the observation position is erroneously detected. When the walking phase is specified as phase 5 during the leg tracking, it is determined that the subject 1 is running. When it is determined that the subject 1 is traveling, for example, the monitor 30 may call attention and prompt the user to perform measurement again.

  Subsequently, for all the stored data (distance information stored in association with the time) stored in the storage unit 22, a data end determination is performed to determine whether each process related to data analysis has ended (S11). . If No in S11, the process returns to S9. On the other hand, in the case of Yes in S11, it is determined that the data analysis has been completed. As described above, the movement trajectory of both legs in the TUG test is acquired. Then, the walking parameter calculation unit 28 calculates a walking parameter from the acquired movement trajectory of both legs (S12). Thereafter, the analysis result of the stored data analysis is output to the monitor 30 (S13).

  In the data analysis unit 24, the following processing is performed by the hidden data interpolation unit 27g, and when one leg is temporarily hidden from the laser range sensor 10, the observation position therebetween is interpolated. That is, first, for example, based on the observation state from the laser range sensor 10 (result of the above-mentioned observation pattern, etc.), it is determined whether or not one leg F is hidden when viewed from the emission direction of the laser light L. (S21). In the case of No in S21, that is, when the leg F is hidden at the current processing time (for example, in the case of processing times k + 1 and k + 2 in FIG. 11B), the predicted predicted position is the observation position. Obtained (S22). In the case of Yes in S21, it is determined whether or not there is a cover before a certain processing time (S23).

In the case of Yes in S23 (for example, in the case of processing times k + 3 and k + 4 in FIG. 11B), the leg position while being hidden is interpolated (S24). That is, the observation position y ^′ _k when hidden is virtually calculated using the Catmull-Rom spline curve. Then, in order to update the state quantity such as the covariance matrix of the Kalman filter, the prediction process and the filtering process of the Kalman filter are performed again using the virtually calculated observation position y ′ _{k + i} (S25). After S21, if the answer is No in S23, and after S25, the process proceeds to S21 again until the processing is completed for all the stored data.

  As described above, in the walking measurement system 100, the filtering process using the walking model is performed, and the position and speed of the leg F of the subject 1 are calculated. The plurality of walking phases defined in the walking model includes a phase 0 in which both legs of the subject 1 are support legs, that is, a walking phase at rest. Therefore, it is possible to cope with a case where a stationary person starts walking and a person who is walking stationary, and the position and speed of the leg F can be calculated with high accuracy. It is possible to accurately grasp the walking characteristics of the subject 1.

  In the walking measurement system 100, when the observation position is detected from the distance information and the walking phase specified based on the observation position has changed from the periodic change with respect to the walking phase at the previous processing time, The observation position is set as a false detection value. Thereby, the erroneous detection of the observation position can be determined using the fact that the human walking phase changes periodically.

  In the walking measurement system 100, the walking phase is specified from the determination result of the support leg swing leg determination. Filtering processing is performed based on the acceleration corresponding to the identified walking phase. A person advances walking with one leg F as a supporting leg and the other leg F as a free leg. Both legs move periodically while changing the free leg and the supporting leg by landing. The walking measurement system 100 can accurately grasp the walking characteristics of the subject 1 in consideration of the acceleration change in the behavior of the leg F.

  In the walking measurement system 100, when both legs are supporting legs, the walking phase is specified as the first walking phase. When the left leg is a free leg and the right leg is a support leg, and the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a positive value, walking The phase is identified as the second walking phase. If the left leg is a free leg and the right leg is a support leg, and the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a negative value, walking The phase is identified as the third walking phase. If the left leg is a supporting leg and the right leg is a free leg, and the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg is a positive value, walking The phase is identified as the fourth walking phase. When the left leg is a supporting leg and the right leg is a free leg, and the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg is a negative value, walking The phase is identified as the fifth walking phase. In this case, it is possible to specify the walking phase in consideration of the relative positional relationship between the legs and the moving speed.

  In the walking measurement system 100, the plurality of walking phases include phase 5 when both legs are free legs. If the identified walking phase is phase 5, the data analysis unit 24 determines that the subject 1 is running (running). Thereby, the presence or absence of the test subject 1 can be determined. Moreover, based on the determination result determined that the subject 1 is traveling, a warning can be output via the monitor 30 and the TUG test can be interrupted.

  In the walking measurement system 100, the data analysis unit 24 performs a filtering process using a Kalman filter and estimates the position and speed of the leg F. As a result, for example, filtering processing considering noise and the like included in the distance information can be performed.

  The gait measurement system 100 uses the laser range sensor 10 and the pressure sensor 15 that are small and capable of acquiring distance information on a two-dimensional plane to track both legs and improve measurement accuracy with little loss or misidentification in the measurement of the TUG test. Can be realized.

  The gait measurement system 100 does not analyze in real time during the trial of the gait test, but analyzes (post-analysis) by the data analysis unit 24 after the gait test is completed. As a result, it is possible to calculate a precise movement trajectory and walking parameters. In the walking measurement system 100, it is of course possible to analyze the walking characteristics by the data analysis unit 24 while trying the walking test.

  Here, in the TUG test for a plurality of healthy persons (seven persons), measurement was performed by the walking measurement system 100, and at the same time, measurement was performed by a three-dimensional motion analysis apparatus (motion capture system). Then, the analysis results of the walking measurement system 100 and the three-dimensional motion analysis device were compared. As a result, it was confirmed that the walking measurement system 100 improved the tracking performance of both legs and the movement trajectory accuracy. In addition, the validity of the walking parameters calculated by the walking measurement system 100 could be confirmed.

  FIG. 17 is a diagram for explaining the effectiveness of the interpolation of the leg position when hidden. FIG. 17A is a diagram illustrating the movement trajectory of both legs acquired by the conventional method, and FIG. 17B is a diagram illustrating the movement trajectory of both legs acquired from the walking measurement system 100. In FIG. 17A, the movement locus 60R indicates the movement locus of the right leg, and the movement locus 60L indicates the movement locus of the left leg. In FIG. 17B, the movement locus 61R indicates the movement locus of the right leg, and the movement locus 61L indicates the movement locus of the left leg. On each movement locus 60R, 60L, 61R, 61L, a large point indicates a supporting leg, and a small point indicates a free leg.

  As shown in FIG. 17, in the walking measurement system 100 that interpolates the leg position while one leg is hidden using a Catmull-Rom spline curve, it is hidden in the state of a free leg whose movement is likely to change, such as during a turn operation. In this case, it can be confirmed that the leg position and the movement locus can be corrected. In the walking measurement system 100, it is possible to suppress the movement trajectory (refer to the inside of the round frame in FIG. 17A) greatly different from the actual motion.

  FIG. 18 is a diagram illustrating an example of an analysis result (movement trajectory and landing position of both legs) by the walking measurement system 100. In FIG. 18, the movement locus 61 </ b> R indicates the movement locus of the right leg, and the movement locus 61 </ b> L indicates the movement locus of the left leg. On each of the movement loci 61R and 61L, a large point indicates a support leg, a small point indicates a free leg, a large circle indicates a landing position, and a number near the landing position indicates the number of steps.

  In the walking measurement system 100, not only the walking parameters but also the movement trajectories 61R and 61L of both legs as shown in FIG. 18 can be output immediately after the test measurement. The walking measurement system 100 can present the result to the subject 1 not only numerically but also visually. Therefore, it is particularly useful in TUG test measurement for elderly person fall risk assessment in community health activities.

  The walking measurement system 100 has the following effects for solving the problems.

  In a technique related to conventional walking analysis, a floor reaction force meter and a three-dimensional motion measuring device are generally used as a walking measurement system often used when measuring walking parameters. However, when measuring walking of several meters, the system scale is large and the introduction cost is high. For this reason, many qualitative evaluations of walking training that are actually carried out in nursing homes for elderly people are mainly qualitative, mainly by observation by third parties, making quantitative evaluation difficult. In this respect, according to the walking measurement system 100, quantitative evaluation of walking training can be realized at low cost. That is, the problem of providing a walking measurement system capable of realizing quantitative evaluation of walking training at a low cost is solved.

  Usually, in the walking test, the test execution time from the start to the end of the test (for example, in the TUG test, the time from when the subject 1 seated in the chair 2 starts moving to the seat again) is measured using a stopwatch or other instrument. Measured by a test measurer. Therefore, an error may occur in the test execution time depending on the operation of the test measurer. In this respect, according to the walking measurement system 100, it is possible to accurately measure the test execution time from the start to the end of the test. That is, the problem of providing a walking measurement system that can accurately measure the test execution time from the start to the end of the test is solved.

Usually, in the walking test, the test measurer counts the number of steps of the subject 1 with assistance. In this case, step length 71, stride length 72, it is difficult to measure the detailed gait parameters such Fu隔w _3. According to the walking measurement system 100, detailed walking parameters can be measured. That is, the problem of providing a walking measurement system capable of measuring detailed walking parameters is solved.

  For example, in the TUG test, the walking speed is faster than in an MTST (Multi-Target Stepping Task) test and the like, and the walking has a rhythmic movement close to that of an actual walking. In a conventional walking measurement system, it may be difficult to cope with such a movement. According to the walking measurement system 100, walking with a fast walking speed and rhythmic movement can be handled. That is, it solves the problem of providing a walking measurement system that can cope with walking with high walking speed and rhythmic movement.

  For example, in the TUG test, when the subject 1 turns the marker 3, a situation where one leg is hidden with respect to the laser range sensor 10 is likely to occur. Since the change in the movement of the leg F becomes noticeable during the turn, it may be difficult to predict with the conventional technique. According to the walking measurement system 100, it is possible to accurately predict the leg F that is hidden when the subject 1 turns. That is, it solves the problem of providing a walking measurement system capable of accurately predicting the leg F hidden when the subject 1 turns the marker 3.

  For example, the TUG test includes not only a phase in which the subject 1 goes straight, but also a phase in which the subject 1 turns the marker 3 (turns around the marker 3 around the marker 3 so that the traveling direction changes by 180 °). In the conventional walking measurement system, the subject 1 is not sufficiently considered to turn, and the measurement accuracy may be reduced. According to the walking measurement system 100, it is possible to increase measurement accuracy in a walking test including a phase in which the subject 1 turns. That is, the problem of providing a walking measurement system capable of improving measurement accuracy in a walking test including a phase in which the subject 1 turns is solved.

  The preferred embodiments of the present invention have been described above. However, the present invention is not limited to the above-described embodiments. The present invention can be modified without departing from the scope described in the claims or applied to other embodiments. May be.

  In the above embodiment, the walking measurement system 100 is adapted to measure walking characteristics in the TUG test, but is not limited to this, for example, walking characteristics in MTST, Rhythmic Stepping Exercise (RSE), and various other walking trainings. The gait measurement system 100 can be applied to the measurement.

  In the above embodiment, the laser range sensor 10 is used as the distance information acquisition unit, but various sensors and devices can be used as long as the distance information at a position at a predetermined height from the floor surface can be acquired. For example, an infrared sensor or the like that can acquire the above-described two-dimensional plane distance information may be used as the distance information acquisition unit. Although the said embodiment is provided with one laser range sensor 10, you may provide multiple. In the said embodiment, although the pole 40 was used as a reference member, it is not limited to this. As the reference member, various members may be used as long as the laser range sensor 10 can be aligned with the marker 3.

  In the above embodiment, the laser range sensor 10 is arranged so that the laser beam L is irradiated from the rear side of the subject 1 approaching the marker 3 (moving away from the chair 2), but the laser is emitted from the front side of the subject 1 approaching the marker 3. You may arrange | position the laser range sensor 10 so that the light L may be irradiated. In some cases, the laser range sensor 10 may be arranged so that the laser beam L is irradiated from the side of the subject 1 approaching the marker 3. In the above embodiment, the laser range sensor 10 is arranged in the lower part of the chair 2, but the arrangement position of the laser range sensor 10 is not limited, and the laser range sensor 10 is arranged according to the irradiation mode of the laser light L to the subject 1. That's fine.

  In the above embodiment, the subject 1 has risen (moved out) from the chair 2 and seated in the chair 2 based on the pressure information of the pressure sensor 15, but in addition to the determination by the pressure sensor 15 or the pressure sensor 15 is omitted. The rising and the seating may be determined based on the detection result of the laser range sensor 10. In the above embodiment, the rising and sitting of the subject 1 are detected based on the pressure information, but only the rising may be detected based on the pressure information, or only the sitting may be detected based on the pressure information.

  The present invention can be understood as a walking measurement method implemented by the walking measurement system 100. The present invention can be understood as a walking measurement program that causes a computer to execute the walking measurement method (each process of the walking measurement system 100).

  Hereinafter, based on an Example, this invention is demonstrated more concretely. In addition, this invention is not limited to a following example.

(Subjects and methods)
In the Example, the measurement result by the gait measurement system 100 was verified in the TUG test with seven healthy subjects as subjects. Specifically, the TUG test when the subject turns the marker 3 counterclockwise and the TUG test when the subject turns clockwise are each performed twice (that is, a total of 28 TUG tests). . Simultaneously with the implementation of the TUG test, measurement was performed with a three-dimensional motion analyzer. By comparing the analysis result of the walking measurement system 100 and the analysis result of the three-dimensional motion analysis device, the tracking performance by the walking measurement system 100 and the measurement accuracy of the walking parameters acquired by the walking measurement system 100 were verified.

The specifications of the laser range sensor 10 to be used are as follows.
[Model name] UTM-30LX (Hokuyo Automatic Co, Ltd., 2015)
[Ranging range] 0.1-30m, maximum detection distance 60m, angle 270 °
[Ranging accuracy] 0.1-10m: ± 30mm, 10-30m: ± 50mm
[Angular resolution] 0.25 ° (360 ° / 1440 divisions)

The system parameters are as follows.
Sun Printer Im Δt (ms): 25 (40 Hz)
Covariance Q: diag (15.0 ² , 15.0 ² )
Covariance R: diag (0.04 ² , 0.04 ² )
Support leg speed threshold (m / s): 0.47
Swing leg speed threshold (m / s): 0.93
Gate G (content probability P _G = 0.999): 13.82
Number of steps of acceleration function N _ac : 40

  In the example, the walking measurement system 100 and the three-dimensional motion analysis apparatus were aligned using two poles 40. In the three-dimensional motion analysis apparatus, seven cameras were used, and the sampling period was set to 5.0 ms (200 Hz). In the analysis of the three-dimensional motion analyzer, the joint position of the lower limb of the subject is calculated using the Plug-In Gait model, and the x direction at the measurement height of the laser range sensor 10 is calculated from the coordinates of the knee and the center position of the ankle joint. And the position of the y direction was calculated. The measurement results at the calculated measurement height were compared, and the position measurement accuracy of the leg F of the walking measurement system 100 was verified. The landing position measured by the walking measurement system 100 was compared with the center position of the ankle joint during landing, and the accuracy was verified. The determination of landing (bed contact and getting out of bed) was performed using pressure sensors attached to the heels of both legs of the subject 1 and the second metatarsal bone.

(Validity verification for correlation processing considering periodicity of walking phase)
In order to verify the effectiveness of correlation processing in consideration of the periodicity of the human walking phase, a method using correlation processing by the GNN method (hereinafter referred to as “Method 1”) and a method in consideration of the periodicity of the walking phase ( In the following, “Method 2”) and the tracking performance of both legs were compared. The Global Nearest Neighbor (GNN) method is a method of selecting a combination that minimizes the cost function for associating a plurality of observation positions with tracking targets. In the GNN method, even when there are observation positions where the gates G to be tracked overlap, it is possible to perform a reasonable association.

  FIG. 19 is a diagram illustrating a result of tracking both legs when the left leg is temporarily hidden from the laser range sensor 10 and the distance information of the laser range sensor 10 is disturbed. In FIG. 19, the line indicates the movement trajectory of the leg F, the small mark on the line indicates the free leg, and the large mark indicates the landing position. 19 (a) to 19 (c) show the results of tracking both legs by Method1, and FIGS. 19 (d) to 19 (f) show the results of tracking both legs by Method2. 19 (a) and 19 (d) show the results at the same time, FIGS. 19 (b) and 19 (e) show the results at the same time, and FIGS. 19 (c) and 19 (f) The result at the same time. The order of FIGS. 19A to 19C and the order of FIGS. 19D to 19F correspond to the order of time.

In the example shown in FIG. 19A, the left leg is hidden by the right leg, and since there is no corresponding left leg observation position, the filtering process is not performed and the left leg gate G is enlarged. After that, as shown in FIG. 19B, two observation positions y ¹ _k and y ² _k are detected, but the observation position y ¹ _k is erroneously detected due to the disturbance of the distance information of the laser range sensor 10. It is. At this time, in the Method 1 GNN method, since the association is always performed when the observation position is included in the gate G, the association is also performed for the erroneously detected observation position. In this case, as shown in FIG. 19C, it can be confirmed that erroneous identification (replacement) of both leg portions occurs.

On the other hand, in the case of Method 2 in which the periodicity of the walking phase is taken into account, as shown in FIG. 19 (e), the observation position y ¹ _k is erroneously detected and not associated with the left leg, and the left leg is observed. Assume that the position does not exist. This is because the walking phase changes from phase 3 to phase 2 with respect to the association between the observation position y ¹ _k and the left leg and the association between the observation position y ² _k and the right leg (that is, human This is because the association is performed to perform a change deviating from the periodicity of the walking phase. Then, as shown in FIG. 19 (f), an appropriate association is performed with an observation position that satisfies the change in the walking phase (that is, an observation position that has changed according to the periodicity of the person's walking phase). Thereby, it can be confirmed that both legs can be tracked without erroneous identification.

  FIG. 20 is a graph showing the results of the tracking success rate. The success of the tracking here is defined as a case where there is no misidentification and all the landings can be detected without erroneous detection. As shown in FIG. 20, in Method 1 according to the comparative example, the tracking success rate was 50% (14 times / 28 times). On the other hand, the tracking success rate of Method 2 considering the periodicity of the gait phase was 89.3% (25 times / 28 times). Thereby, it can confirm that misidentification can be reduced by considering the periodicity of a person's walk phase.

(Effectiveness verification for interpolation of observation position while hidden)
In order to verify the effectiveness of the observation position interpolation method while the leg F is hidden from the laser range sensor 10, Method 2 that does not perform observation position interpolation and Method that performs observation position interpolation (hereinafter “Method 3”). ") And the tracking performance of both legs.

  FIG. 21 is a diagram illustrating an example of a tracking result when the left leg is hidden from the laser range sensor 10 in the turning phase. FIG. 21A shows the tracking result according to Method 2, and FIG. 21B shows the tracking result according to Method 3. In FIG. 21, the line indicates the movement trajectory of the leg F, the small mark on the line indicates the free leg, and the large mark indicates the landing position. A large circle indicates the landing position, and the number near the landing position indicates the number of steps.

  In Method 2, when the leg F cannot be observed, the predicted position based on the movement model of the leg F is acquired as the observation position. That is, in Method 2, during the turning phase, while the free leg is hidden by the support leg, the predicted position is predicted based on the movement direction observed immediately before the free leg is hidden, and this predicted position is set as the observed position of the free leg. . In the turning phase, the moving direction of the leg F changes rapidly. Therefore, in Method 2 as described above, as shown in the round frame of FIG. 21A, the measurement accuracy of the position of the leg portion F is degraded, and there is a possibility that erroneous identification is likely to occur. On the other hand, in Method 3, while the leg F is hidden, the observation position of the hidden leg F is interpolated using a spline curve. Thereby, as shown in FIG.21 (b), it can confirm that both leg tracking performance and measurement accuracy improve further. As shown in FIG. 20, the tracking success rate of Method 3 was 96.4% (27 times / 28 times).

  FIG. 22 shows RMSE (root) regarding the movement trajectory and landing position of both legs at the height (installation height) of the optical window of the laser range sensor 10 when the analysis result of the three-dimensional motion analysis apparatus is a true value. It is a graph showing mean squared error). 21 (a) indicates a false detection of the landing position. As shown in FIGS. 20-22, according to the present Example, it can confirm that the movement locus | trajectory and landing position of both legs can be measured with high precision.

  FIG. 23 shows an example of the analysis result of the movement trajectory and landing position of both legs. The walking measurement system 100 can output the walking parameters and the movement trajectory illustrated in FIG. 23 immediately after the TUG measurement. The walking measurement system 100 can present the result to the subject 1 not only numerically but also visually. Therefore, it turns out that it is especially useful in TUG measurement for elderly person fall risk evaluation in community health activities.

  DESCRIPTION OF SYMBOLS 1 ... Test subject, 10 ... Laser range sensor (distance information acquisition part), 24 ... Data analysis part, 100 ... Walking measurement system, F ... Leg part, L ... Laser beam (detection wave).

Claims (6)

A gait measurement system used for a subject's gait test,
A distance information acquisition unit that emits a detection wave so as to scan along the horizontal direction and acquires distance information regarding a distance from an object that reflects the detection wave based on a reflection state of the detection wave over time;
Based on at least the distance information acquired by the distance information acquisition unit, a walking model in which a plurality of walking phases including a first walking phase when both legs of the subject are support legs is defined is used. A gait measurement system comprising: a data analysis unit that performs filtering processing and calculates the position and speed of the leg of the subject in association with time. The data analysis unit
The position of the leg at a predetermined processing time is detected as an observation position from the distance information,
When the walking phase identified based on the observation position changes deviating from a periodic change with respect to the walking phase at a time prior to the predetermined processing time, the observation position is set as a false detection value. The walking measurement system according to claim 1. The data analysis unit
Performing support leg swing leg determination to determine whether each of the both leg parts is a free leg or a support leg,
Identify the walking phase of the walking model from the determination result of the support leg swing leg determination,
The walking measurement system according to claim 1, wherein the filtering process is performed based on an acceleration corresponding to the identified walking phase. The data analysis unit
When both the leg portions are support legs, the walking phase is identified as the first walking phase,
The subject's left leg is a free leg and the right leg is a support leg, and the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a positive value In this case, the walking phase is identified as the second walking phase,
The subject's left leg is a free leg and the right leg is a support leg, and the inner product of the position vector of the left leg with respect to the right leg and the velocity vector of the left leg is a negative value In this case, the walking phase is identified as a third walking phase,
The subject's left leg is a supporting leg and the right leg is a free leg, and the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg is a positive value In this case, the walking phase is identified as the fourth walking phase,
When the subject's left leg is a supporting leg and the right leg is a free leg, the inner product of the position vector of the right leg with respect to the left leg and the velocity vector of the right leg is a negative value In this case, the walking measurement system according to claim 3, wherein the walking phase is identified as a fifth walking phase.   5. The data analysis unit according to claim 3, wherein when the both legs are swing legs, the data analysis unit identifies the walking phase as a sixth walking phase and determines that the subject is running. 6. Walking measurement system.   The gait measurement system according to claim 1, wherein the data analysis unit performs the filtering process using a Kalman filter.