Walking training device

Abstract

To estimate a foot joint kick-out timing from indirect detection by an angle sensor.SOLUTION: A walking training device 1 includes a saddle support type load-relieving device 21 having a saddle 210, and a treadmill 22 for relatively moving the saddle 210 in a front direction with respect to a floor surface. The walking training device 1 includes the inertial sensor 121 that is attached to the waist of a person sitting on the saddle 210 and detects a tilting motion on the frontal plane of the pelvis during the walking training, and a timing estimation unit 112 that estimates a second extreme value point of the same polarity (e.g., a second negative peak) detected on the other side as a foot joint kick-out timing after the inertial sensor 121 detects that the direction of the pelvis crosses a vertical axis on the frontal plane from one of the right and left sides to the other.SELECTED DRAWING: Figure 6

Description

[Technical Field]

[0001] The present invention relates to a saddle-supported, weight-relieving walking training device. [Background technology]

[0002] In stroke rehabilitation, weight-supported treadmill training is one promising approach to restoring gait in patients. The latest Japan Stroke Society Stroke Treatment Guidelines 2021 (revised 2023) state that "weight-supported treadmill training (BWSTT) is appropriate in the subacute phase" and that "it is appropriate to add both BWSTT and robot-assisted gait training to standard gait training." The recommendation was classified as Level B (moderate recommendation) with a high level of evidence (consistent evidence from multiple high-quality RCTs) (Non-Patent Document 1). While lower limb robotic therapy is still in its infancy, scientific evidence regarding the usefulness of weight-supported treadmill training is steadily accumulating.

[0003] During healthy walking, the hip joint and ankle joint account for more than 80% of the positive power of the hip, knee, and ankle joints, and it is known that energy is consumed mainly during the phases of "hip extension" and "ankle push-off" (Non-Patent Documents 2 and 3). For example, as shown in Patent Document 1, in a configuration that combines a saddle-supported weight-relief device with a treadmill, the treadmill belt carries the leg backward, supporting hip extension. Meanwhile, ankle push-off has been addressed using exoskeleton-type ankle robots (Non-Patent Document 4) and functional electrical stimulation (Non-Patent Document 5). Furthermore, it is known that the timing of ankle push-off during treadmill walking coincides with the zero crossing point of the anterior-posterior component (AP-GRF) when a ground reaction force sensor that directly measures stepping force is used (Non-Patent Documents 5 and 6). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2018-139975

Non-licensed literature

[0005]

Non-licensed literature 1

Non-licensed Document 2

Non-licensed Document 4

Non-licensed Document 5

[0006] On the other hand, in situations where a floor reaction force sensor is not used or is difficult to use, there is a demand for the development of technology to predict the timing of ankle kick-off, and many methods have been proposed using devices and equipment such as image sensors, motion capture, force sensors, goniometers, gyro sensors, acceleration sensors, electromyography sensors, switches, etc. However, these methods have several issues, such as the need for dedicated equipment, the time required for preparation, and the need to attach devices and equipment to the subject that restrict movement (Non-Patent Document 7).

[0007] The present invention has been made in view of the above, and provides a saddle-supported, weight-relieving walking training device that can estimate the timing of ankle push-off using a simple detection means. [Means for solving the problem]

[0008] The walking training device of the present invention includes a saddle-supported unloading device with a saddle, and a walking mechanism that moves the saddle relatively in a forward direction with respect to the floor surface. The walking training device is equipped with an angle sensor that is attached to the center of the waist of a person sitting on the saddle and detects the lateral tilt of the pelvis in the frontal plane during walking training, and an estimation means that, after the angle sensor detects that the orientation of the pelvis has crossed the vertical axis in the frontal plane from one side to the other, estimates the second extreme point of the same polarity detected on the other side as the timing of ankle push-off.

[0009] In addition, the walking training device of the present invention includes a saddle-supported unloading device with a saddle and a walking mechanism that moves the saddle relatively forward relative to the floor surface, and is equipped with an angle sensor attached to the saddle that detects the rotational movement of the saddle on a horizontal plane during walking training, and an estimation means that, after the angle sensor detects that the orientation of the saddle has crossed the sagittal horizontal axis on the horizontal plane from one side to the other, estimates the first inflection point detected on the other side as the timing of ankle push-off.

[0010] This invention focuses on the finding that the timing of ankle push-off during walking coincides with the zero crossing point of the anterior-posterior component (AP-GRF) of a ground reaction force sensor, which directly measures tread force. Based on this finding, we investigated various movements that may be correlated with this timing, assuming the application of saddle-supported weight-relief. As a result, we experimentally verified that ankle push-off timing can be estimated from information related to pelvic movement during gait training, which can be indirectly obtained using a simple angle sensor. According to this invention, we found that movements highly correlated with ankle push-off timing during saddle rotation on the horizontal plane or pelvic lateral tilt in the frontal plane were detected using data from an angle sensor attached to a saddle that can rotate on the horizontal plane or to the center of the waist of a person straddling the saddle. We verified that ankle push-off timing can be indirectly estimated by detecting horizontal saddle rotation or pelvic lateral tilt using such an angle sensor. The estimated timing signal (estimated signal) can be preferably used to determine the timing of ankle push-off support for walking assistance. For example, this includes functional electrical stimulation, driving the actuator part of a support robot, and notifying the timing of support.

[0011] The present invention also includes an assist drive unit that provides walking assistance and an assist signal generation unit that supports ankle joint push-off, wherein the assist signal generation unit generates an assist signal to be output to the assist drive unit from a signal estimated as the ankle joint push-off timing. With this configuration, the assist signal is supplied to the assist drive unit in correspondence with the ankle joint push-off timing.

[0012] The assist drive unit includes positive and negative electrodes to which functional electrical stimulation is applied, and the assist signal generator generates a stimulation signal to be applied to the positive and negative electrodes as the assist signal. With this configuration, the ankle push-off is supported by the stimulation signal in accordance with the timing of the ankle push-off.

[0013] The assist drive unit is an exoskeleton-type ankle robot having an actuator that assists in pushing off from the ankle joint, and the assist signal generator generates a signal to drive the actuator of the exoskeleton-type ankle robot as the assist signal. With this configuration, the actuator of the exoskeleton-type ankle robot is driven by the drive signal in accordance with the timing of the ankle joint push-off, thereby assisting the push-off from the ankle joint. [Effects of the Invention]

[0014] According to the present invention, it is possible to estimate the timing of kicking off at the ankle joint from indirect detection using an angle sensor attached to the saddle or the waist. [Brief explanation of the drawings]

[0015] [Figure 1] 1A and 1B are diagrams showing the configuration of the walking training device, in which (A) is a schematic diagram of the entire device as seen from the side, and (B) is a perspective view of the device's exterior showing the detailed structure of the saddle. [Figure 2] (A) is a horizontal view illustrating the rotation of the saddle, and (B) is a frontal view illustrating the lateral tilt of the human pelvis. [Figure 3] These are diagrams of experimental results for the right leg to examine the relationship between ground reaction force and ankle push-off timing. (A) is a time chart showing the relationship with pelvic lateral tilt data during unweighted walking, and (B) is a time chart showing the relationship between pelvic lateral tilt data and saddle horizontal rotation data during unweighted walking on a saddle. [Figure 4] This is a diagram of the experimental results for the left leg to examine the relationship between floor reaction force and ankle push-off timing, and is a time chart showing the relationship between pelvic lateral tilt data and saddle horizontal rotation data during saddle-unloaded walking. [Figure 5] (A) is a diagram showing the degree of agreement between walking phase R in floor reaction force and walking phase P at the characteristic point of lateral pelvic tilt, and (B) is a diagram showing the degree of agreement between walking phase R in floor reaction force and walking phase Q at the characteristic point of horizontal saddle rotation. [Figure 6] 1 is a configuration diagram showing an embodiment of a walking training device according to the present invention. [Figure 7] 10 is a flowchart of a process I for estimating the timing of ankle joint kick-off due to pelvic lateral tilt. [Figure 8] 10 is a flowchart of a process II for estimating the timing of ankle kick-off due to horizontal rotation of the saddle. DETAILED DESCRIPTION OF THE INVENTION

[0016] This invention uses a gait training device equipped with a saddle-supported weight-relief device to estimate the timing of a person undergoing gait training at the ankle joint push-off using an angle sensor attached to the saddle or waist. Ankle push-off is a movement with a high positive power rate, and by estimating its timing, an exoskeleton-type ankle robot or functional electrical stimulation is performed in accordance with the movement, effectively supporting gait training. While the timing of ankle push-off during treadmill walking can be directly measured using a floor reaction force sensor, as described in Non-Patent Documents 5 and 6, measuring push-off timing is not easy in situations where a floor reaction force sensor is not used or is difficult to use.

[0017] Therefore, we investigated methods for measuring ankle push-off timing using other sensors, and as one aspect of this, we investigated and verified the possibility of estimating ankle push-off timing. More specifically, we conducted experiments to investigate the possibility of estimating ankle push-off timing from measurements of either the subject's pelvic posture information or the posture information of the saddle on which the subject sits (pelvis / saddle) during training. Here, we hypothesized that (1) measurements of pelvic lateral tilt / saddle horizontal rotation are useful for estimating ankle push-off timing, and (2) because the equipment required to measure pelvic lateral tilt / saddle horizontal rotation is an angle sensor including an inertial sensor (an inertial measurement unit: a device that detects angular velocity and acceleration), this method minimizes cost, time, and labor in clinical settings. The verification experiments are described below.

[0018] <Verification experiment> 1. Subjects The subjects were five young, healthy adults (males: 25.8±5.6 years old, 1.76±0.07 m, 69.6±4.5 kg). The content and purpose of the experiment were explained to the subjects in advance, and their consent to participate was obtained. The experiment was conducted with the approval of the Osaka University Research Ethics Committee.

[0019] 2.Equipment 2.1 Gait training device Figure 1 shows the configuration of the walking training device used in this experiment. Figure 1(A) is a schematic diagram of the entire device as seen from the side, and Figure 1(B) is a perspective view of the device showing the detailed structure of the saddle. In Figure 2, (A) is a horizontal view illustrating the rotation of the saddle, and (B) is a frontal view illustrating the lateral tilt of the human pelvis.

[0020] The walking training device 20 is a combination of a saddle-supported unloading device 21 and a treadmill 22 as an example of a walking mechanism. The saddle-supported unloading device 21 includes an aluminum base 211 and an aluminum arm 212 that is pivotally supported on a horizontal axis 211a in the sagittal plane of the base 211 and pitches around the axis. The arm 212 has a seesaw structure, with a saddle 210, for example, for a unicycle, attached to one end via a mounting fixture 213 so that the subject can straddle it, and a weight 216 hanging from the other end. The weight 216 has, for example, a ring shape and is loosely fitted into a support 215, and the weight-relief rate for the subject can be adjusted by changing the number of weights hung from a chain 217.

[0021] Saddle 210 includes saddle body 210a having a seat shape and, if necessary, cover 210b made of an elastic material that covers saddle body 210a. Saddle 210 includes horizontal shaft 2131 that is parallel to the front-to-rear direction and is attached to mounting fixture 213 at the tip of arm 212, and vertical shaft 2132 that is attached via horizontal shaft 2131, and saddle body 210a is attached via vertical shaft 2132. With this configuration, saddle 210 rolls around horizontal axis 2131, which serves as a roll axis, and yawing around vertical axis 2132, which serves as a yaw axis, both within a predetermined angular range (see FIGS. 1(B) and 2(A)).

[0022] The saddle 210 can freely rotate around a fixed axis parallel to the axis of the arm 212 (roll) and around a fixed axis perpendicular to the axis of the arm 212 in the sagittal plane (yaw), and these degrees of freedom can be individually restricted depending on the application. Note that the degree of freedom around the fixed axis perpendicular to the sagittal plane (pitch) is always open, as this is handled by the seesaw mechanism of the gait training device 20. This experiment was performed using a two-degree-of-freedom mechanism in which the roll rotation of the saddle 210 was fixed and the pitch and yaw rotations were open. Hereinafter, the yaw rotation of the saddle 210 will be referred to as horizontal saddle rotation (Figure 2(A)), and this movement will be the focus of attention.

[0023] The treadmill 22 includes a pair of rollers 221 having parallel horizontal rotation axes, a belt 222 stretched between the pair of rollers 221, and a drive unit (not shown) that drives the belt 222 in a circular motion. The belt 222 drives in a circular motion to assist the subject sitting on the saddle 210 in walking.

[0024] The floor reaction force sensor 31 is known and comprises a sensor section and a floor reaction force calculation section. As shown in Fig. 1(A), the sensor section comprises a rectangular plate of a predetermined size laid on the back surface of the belt 222, and load sensors (mechanical-electrical conversion elements) at each corner of the plate. More specifically, each load sensor detects changes in the resistance value of a strain gauge corresponding to the load applied to the plate as changes in current value, thereby directly measuring the floor reaction force as three directional components (vertical direction: V-GRF, front-back direction: AP-GRF, left-right direction: ML-GRF).

[0025] 2.2 Inertial Sensors Figure 2(A) is a horizontal plane view illustrating the rotation of the saddle 210, and Figure 2(B) is a frontal plane view illustrating the lateral tilt of the human pelvis. Inertial sensors (Xsens Mtw Awinda, Movella Inc., USA) 121 and 122 were used to measure the rotation of the saddle 210 and the direction of the lateral tilt of the subject's pelvis during the walking experiment. As shown in Figure 1(A), the inertial sensor 121 was attached to the subject's lower back, on the spinal side midway between the two anterior superior iliac spines on the left and right. The inertial sensor 122 was attached to the underside of the saddle 210. The inertial sensors 121 and 122 each output a counterclockwise (CCW) rotation as a positive value (deg) corresponding to the displacement (direction) of the vertical axis (Up) and the sagittal horizontal axis (Front), respectively, with the vertical axis (Up) and the sagittal horizontal axis (Front) as reference 0°. The sensors output a clockwise (CW) rotation as a negative value (deg) corresponding to the displacement of the vertical axis (Up) and the sagittal horizontal axis (Front), respectively.

[0026] 2.3 Experimental Procedure The subjects were instructed to walk at a constant speed on the treadmill 22 in both unloaded and saddle-supported unloaded states. In both cases, the walking speed was 1.0 m / s and the walking time was 90 seconds. During unloaded walking, the saddle 210 was adjusted so that 20-30% of the body weight was unloaded. During unloaded walking, the saddle 210 could be freely rotated in the pitch and yaw directions.

[0027] In the unloaded state, floor reaction force data for 20 gait cycles and lateral tilt displacement data of the inertial sensor 121 attached to the pelvis were obtained. In addition, in the saddle-supported unloaded state, floor reaction force data for 20 gait cycles and lateral tilt displacement data of the pelvis from the inertial sensors 121 and 122 and rotational displacement data of the saddle 210 were obtained. The floor reaction force data was measured to obtain the true value of the timing of the ankle joint push-off movement.

[0028] 3. Experimental Results 3.1 Pelvic lateral tilt and saddle horizontal rotation during unweighted / saddle-supported weight-bearing walking Figure 3 shows data on the lateral tilt displacement of the pelvis, horizontal rotation displacement of the saddle 210, and floor reaction force (vertical direction: V-GRF, anterior-posterior direction: AP-GRF, lateral direction: ML-GRF) for the right leg of a representative subject during unweighted and saddle-supported unweighted walking, obtained during the experimental procedure. More specifically, Figure 3(A) shows data on the lateral tilt displacement of the pelvis and floor reaction force (vertical direction: V-GRF, anterior-posterior direction: AP-GRF, lateral direction: ML-GRF) for the right leg during unweighted walking, and Figure 3(B) shows data on the lateral tilt displacement of the pelvis, horizontal rotation displacement of the saddle 210, and floor reaction force (vertical direction: V-GRF, anterior-posterior direction: AP-GRF, lateral direction: ML-GRF) for the right leg during saddle-supported unweighted walking. Note that data on the left leg during saddle-supported unweighted walking will be discussed later in Figure 4.

[0029] Each data point is the average of 20 gait cycles. Pelvis indicates pelvic lateral tilt, Saddle indicates saddle horizontal rotation, and the others indicate ground reaction forces. The standard deviation of each data point is indicated by the width of the light-colored band on the measurement result lines for pelvic lateral tilt and saddle horizontal rotation. The horizontal axis represents the gait phase, with the initial contact of one leg (here, the right leg) being set to 0%. The Push-off Initiation line shown in Figures 3(A) and 3(B) for the gait phase indicates the timing of the zero crossing point of the AP-GRF of the ground reaction force. Figure 4 also shows the results of an experiment to examine the relationship between the ground reaction force of the left leg and the timing of ankle push-off. It is a time chart showing the relationship between the pelvic lateral tilt displacement and the saddle horizontal rotation displacement during saddle-supported unweighted walking. As with the right leg, the Push-off Initiation line indicates the timing of the zero crossing point of the AP-GRF of the ground reaction force. Furthermore, from Figures 3(B) and 4, it was found that the graphs of pelvic lateral tilt and saddle horizontal rotation for the right and left legs are inverted at the reference [0°], that is, they are in a roughly bilaterally symmetrical relationship.

[0030] 3.2 Relationship between the characteristic points of pelvic lateral tilt / saddle horizontal rotation and the timing of ankle push-off Figure 5(A) shows the relationship between the second negative peak (walking phase P), which is the second minimum value of pelvic lateral tilt from the ground contact of the right leg, and the timing of ankle push-off obtained from floor reaction force (walking phase R) during saddle-supported unweighted walking. Figure 5(B) shows the relationship between the first inflection point (walking phase Q), which is the first convex to concave point of saddle horizontal rotation from the ground contact of the right leg, and the timing of push-off estimated from floor reaction force (walking phase R) during saddle-supported unweighted walking.

[0031] The graphs in Figures 5(A) and (B) show data for 100 walking cycles (20 walking cycles per person) obtained from five subjects, and are displayed so that each subject can be identified. More specifically, the data for each subject is displayed in five different ways: a small circle mark indicating each data point surrounded by a dashed line, a small circle mark with a bar mark added above it, a small circle mark with a bar mark added below it, a small circle mark with a dot mark added inside it, or a small circle mark only.

[0032] The diagonal solid line in the graph indicates a perfect match between the values ​​on the horizontal and vertical axes, and it was confirmed that the data points are distributed near the solid line in Figure 5(A), and above the solid line in Figure 5(B). Expressed as Spearman's rank correlation coefficient, ρ = 0.59 (p < 0.01) for Figure 5(A) and ρ = 0.67 (p < 0.01) for Figure 5(B).

[0033] 4. Discussion 4.1 Comparison of pelvic lateral tilt during unweighted and saddle-supported weight-bearing walking The Pelvis graphs shown in Figures 3(A) and (B) and Figure 4 indicate that the pelvic lateral tilt movement differs between unweighted walking and saddle-supported unweighted walking. These results suggest that saddle-supported unweighted walking results in unique pelvic movement not seen in unweighted walking (or natural walking). One factor is presumed to be the effect of the saddle 210 of the saddle-supported unweighted device on pelvic movement during walking. However, no significant difference was observed in the amplitude of pelvic lateral tilt between unweighted walking and saddle-supported unweighted walking, and it is believed that the effect of the saddle-supported unweighted device on the amplitude of pelvic lateral tilt was minimal under the conditions of this experiment (roll rotation fixed, pitch and yaw rotations open).

[0034] 4.2 Characteristics of pelvic lateral tilt / saddle horizontal rotation during ankle push-off in saddle-supported unweighted walking Figure 3 shows that pelvic lateral tilt and saddle horizontal rotation each have their own characteristics during ankle push-off during saddle-supported, unweighted walking. Pelvic lateral tilt exhibits a second negative peak after the tilt displacement crosses the vertical axis (changes in angle) from counterclockwise (CCW) to clockwise (CW), while saddle horizontal rotation exhibits a first inflection point after the rotation displacement crosses the sagittal horizontal axis (changes in angle) from clockwise (CW) to counterclockwise (CCW). Figure 4 also shows that pelvic lateral tilt exhibits a second positive peak after the tilt displacement crosses the vertical axis (changes in angle) from clockwise (CW) to counterclockwise (CCW), while saddle horizontal rotation exhibits a first inflection point after the rotation displacement crosses the sagittal horizontal axis (changes in angle) from counterclockwise (CW).

[0035] This suggests that the timing of ankle push-off can be estimated using either lateral pelvic tilt or horizontal saddle rotation. This characteristic was not observed in lateral pelvic tilt during unweight-bearing walking, but was unique to saddle-supported unweight-bearing walking.

[0036] 4.3 Estimation of ankle push-off timing using feature points of pelvic lateral tilt / saddle horizontal rotation during saddle-supported unweighted walking Figure 5(A) shows that the second negative peak of pelvic lateral tilt (gait phase P) and the timing of ankle push-off based on ground reaction force (gait phase R) generally coincide. The data points are distributed almost evenly above and below the solid line representing perfect agreement, which means that both earlier and later times than the actual timing of ankle push-off are estimated equally well. Spearman's rank correlation coefficient (ρ = 0.59, p < 0.01) for this data showed a moderate positive correlation.

[0037] Figure 5(B) shows that the first inflection point of saddle horizontal rotation (gait phase Q) and the timing of ankle push-off based on ground reaction force (gait phase R) generally coincide. Data points tend to be distributed above the solid line representing perfect agreement, which means that the estimated timing is earlier than the actual timing of ankle push-off. Spearman's rank correlation coefficient (ρ = 0.67, p < 0.01) for this data showed a high positive correlation.

[0038] When estimating the timing of ankle push-off, it is desirable to use either the observation results of pelvic lateral tilt or saddle horizontal rotation depending on the purpose. The method of estimation from pelvic lateral tilt requires the trainee to be fitted with an inertial sensor 121, but has the advantage of being able to estimate a time closer to the actual timing of ankle push-off. On the other hand, the method of estimation from saddle horizontal rotation estimates a time slightly earlier than the actual timing of ankle push-off, but has the advantage that because the inertial sensor 122 is attached to the saddle 210, there is no need to prepare a sensor for the trainee. Furthermore, if the property of estimating a time earlier than the actual timing is interpreted as providing a time leeway until the actual movement, this can be advantageous depending on the purpose of exercise support for the trainee.

[0039] In this experiment, we investigated whether it is possible to estimate the timing of ankle push-off by analyzing the lateral tilt displacement of the pelvis and the horizontal rotation displacement of the saddle 210 during saddle-supported unweighted treadmill walking. The findings obtained from this study are as follows (a) to (c).

[0040] (a) The lateral tilt of the pelvis differs between unweighted walking and saddle-supported weight-bearing walking.

[0041] (b) During saddle-supported unweighted walking, the second negative peak (second positive peak) of the pelvic lateral tilt, i.e., the second extreme point of the same polarity detected on the other side, is observed approximately in agreement with the timing of ankle push-off.

[0042] (c) In saddle-supported unweighted walking, the first inflection point of the saddle horizontal rotation was observed slightly before the timing of ankle push-off.

[0043] Based on the above, these findings (a)-(c) have the potential to be applied in clinical settings, and by doing so, it is possible to replace the functions of conventional equipment with a single inertial sensor, leading to significant savings in cost, time, and labor. They are also expected to contribute to the further development of lower limb robotic therapy. Findings (b) and (c) in particular show a useful correlation with ankle push-off timing, and it is thought that the means for detecting this timing could be replaced with angle sensors for pelvic lateral tilt / saddle horizontal rotation.

[0044] Next, a walking training device according to the present invention, which applies the above findings, will be described with reference to FIG. 6. FIG. 6 is a configuration diagram showing one embodiment of a walking training device 1 according to the present invention. The walking training device 1 includes a processing unit 10 for walking assistance, a saddle-supported unloading device 21 shown in FIG. 1, and a treadmill 22 that is used as needed. The treadmill 22 rotates a belt 222 stretched between a pair of rollers 221 at a required speed using a driving source (not shown) such as a motor, as shown in FIG. 1, and the walking trainee, supported by the crotch area of ​​a saddle 210, walks by alternately placing his or her left and right feet on the belt 222. Note that this embodiment does not include a floor reaction force sensor, and instead employs a sensor unit 12, which will be described later.

[0045] Furthermore, in this embodiment, a treadmill 22 is used as the walking mechanism, but the treadmill 22 is not essential, and for example, a form in which the belt is rotated manually may be used, or the walking mechanism may be one in which the saddle-supported unloading device 21 side moves relative to the floor surface at a predetermined speed.

[0046] The processing unit 10 includes an estimation processing unit 11 configured from a processor, and a sensor unit 12, an assistance signal generating unit 13, and an assistance driving unit 14 connected to the estimation processing unit 11. In addition, the estimation processing unit 11 is connected to a storage unit 110 that stores a processing program for executing walking assistance processing and data required for the processing.

[0047] When implemented, the sensor unit 12 is equipped with either an inertial sensor 121 or an inertial sensor 122 depending on the application. The inertial sensors 121 and 122 can have the same configuration, incorporate a gyroscope and an accelerometer, and perform repeated measurement operations to sequentially measure the orientation of the subject. The inertial sensor 121 is attached to the subject's waist as shown in FIG. 1A and measures the orientation of the pelvis in the frontal plane. As shown in FIG. 2B, the inertial sensor 121 uses the vertical axis (Up) as the reference (±0°), outputs a counterclockwise (CCW) orientation as a positive value [deg], and outputs a clockwise (CW) orientation as a negative value [deg]. This allows the clockwise and counterclockwise orientations relative to the front to be measured.

[0048] The inertial sensor 122 is attached to the back side of the saddle 210 as shown in Fig. 1(A) and measures the orientation of the saddle 210 in a horizontal plane. As shown in Fig. 2(A), the inertial sensor 122 uses the sagittal horizontal axis (Front) as a reference (±0°), outputs a counterclockwise (CCW) orientation as a positive value [deg], and outputs a clockwise (CW) orientation as a negative value [deg]. This allows the counterclockwise and clockwise orientations relative to the horizontal plane to be measured.

[0049] The assistance signal generator 13 generates an ankle joint push-off assistance signal corresponding to the ankle joint push-off timing estimated by the estimation processor 11 (described later), and outputs the signal to the assistance driver 14. The assistance driver 14 is applied with a functional electrical stimulation or exoskeleton-type ankle joint robot. In the case of functional electrical stimulation, the push-off force is supported by applying electrical stimulation via positive and negative electrodes 141 attached opposite the ankle joint plantar flexor muscles, such as the soleus muscle. The electrical stimulation can be set to, for example, an AC signal of 30 to 100 Hz at a current of a few tens of milliamperes for a duration of a few tenths of a second. In the case of an exoskeleton-type ankle joint robot, it is attached to the ankle joint and mechanically supports the push-off motion of the ankle joint via an actuator 142.

[0050] The estimation processing unit 11 functions as a sensor signal receiving unit 111 and a timing estimation unit 112 by executing a processing program in the storage unit 110 on the main memory. Note that it is desirable that the walking training device 1 be operated by a medical professional or the like.

[0051] The sensor signal receiving unit 111 repeatedly receives sensor data from one of the inertial sensors 121 and 122 that is to be used. The sensor signal receiving unit 111 receives data sampled at a predetermined period. The inertial sensors 121 and 122 and the sensor signal receiving unit 111 may be connected wirelessly or by wire.

[0052] The timing estimation unit 112 performs processing to estimate the time point corresponding to the ankle joint push-off timing from the repeatedly received sampling data, and outputs an estimated signal to the assistance signal generation unit 13. The assistance instruction signal is a signal that instructs the generation of an electrical stimulation signal if the assistance drive unit 14 is a functional electrical stimulator, and is a signal that generates a drive signal for an actuator 142 that mechanically performs an operation equivalent to ankle joint push-off if the assistance drive unit 14 is an exoskeleton-type ankle robot.

[0053] Next, we will explain ankle joint push-off timing estimation processes I and II. Figure 7 is a flowchart of ankle joint push-off timing estimation process I using pelvic lateral tilt, and Figure 8 is a flowchart of ankle joint push-off timing estimation process II using saddle horizontal rotation. Note that in Figures 7 and 8, inertial sensors 121, 122 are both set so that the CCW direction is represented by a positive value and the CW direction is represented by a negative value.

[0054] In FIG. 7, first, it is determined whether the orientation of the inertial sensor 121 has crossed the vertical axis, i.e., whether the detected data has changed from negative (CW orientation) to positive (CCW orientation) or from positive (CCW orientation) to negative (CW orientation) (step S11). If there is no change, the process returns and the operation is repeated periodically. On the other hand, if there is a change, it is determined whether the change is from CCW orientation to CW orientation with respect to the vertical axis (step S13). If the result is affirmative, it is determined whether a second negative peak (see FIG. 3(B)), which is the second minimum value in the CW orientation, has been detected (step S15). If it is determined that a second negative peak has been detected, an estimated signal is output, and based on that, a support signal is output to the right leg (step S17).

[0055] On the other hand, if in step S13 the change is not from a CCW to a CW direction with respect to the vertical axis (No in step S13), it is determined whether a second positive peak (see FIG. 4) which is a second maximum value in the CCW direction has been detected (step S19). If it is determined that a second positive peak has been detected, an estimated signal is output, and based on this, a support signal is output to the left leg (step S21). Note that whether the time point is a minimum value or a maximum value can be determined, for example, by sequentially calculating the difference between consecutively received detection data before and after, and determining whether the sign of the difference is reversed.

[0056] In FIG. 8, first, it is determined whether the orientation of the inertial sensor 122 has crossed the sagittal horizontal axis, i.e., whether the detected data has changed from negative (CW orientation) to positive (CCW orientation) or from positive (CCW orientation) to negative (CW orientation) (step S31). If there is no change, the process returns and periodically repeats the operation. On the other hand, if there is a change, it is determined whether the change is from CW orientation to CCW orientation with respect to the sagittal horizontal axis (step S33). If the determination is affirmative, it is determined whether a first inflection point (from convex to concave, see FIG. 3(B)) that is the first inflection point in the CCW orientation has been detected (step S35). If it is determined that the first inflection point has been detected, an estimated signal is output, and based on that, a support signal is output to the right leg (step S37).

[0057] On the other hand, if the change in step S33 is not a change from a CW direction to a CCW direction with respect to the sagittal horizontal axis (No in step S33), it is determined whether a first inflection point (from concave to convex, see FIG. 4) that is the first inflection point in the CW direction has been detected (step S39). If it is determined that the first inflection point has been detected, an estimated signal is output, and based on that, a support signal is output to the left leg side (step S41). Note that whether or not an inflection point has occurred can be determined, for example, by sequentially calculating the difference between consecutively received detection data before and after, and the difference (difference of differences), and determining whether or not the sign of the difference of differences is reversed.

[0058] In the ankle joint kick-off timing estimation processes I and II described in Figures 7 and 8, the measurement values ​​of the inertial sensors 121, 122 are set to a reference 0° on the vertical axis or sagittal horizontal axis, with positive and negative values ​​on the left and right sides of that, but instead, one revolution may be measured, for example, with the reference axis set to 180° and from 0° to 360° in one direction.

[0059] Furthermore, in the above embodiment, the inertial sensors 121 and 122 are used, but this is not limiting and an angle sensor capable of detecting a rotation angle may also be used. Furthermore, in this embodiment, the detected second positive peak, second negative peak, and first inflection point (convex to concave, concave to convex) are used to support the foot joint push-off motion, but they may also be used as support signals for guiding the push-off timing, for example. Furthermore, the support of the saddle 210 is not limited to a seesaw type, and may be a type that hangs from above. [Explanation of symbols]

[0060] 1. Walking training device 10 Processing section 11 Estimation processing unit 111 Sensor signal receiving unit 112 Timing estimation unit (estimation means) 12 Sensor section 121,122 Inertial sensor (angle sensor) 13 Support signal generation section 14 Support Drive Unit 141 Electrode 142 Actuator 20. Walking training device 21 Saddle-supported unloading device 22 Treadmill (walking mechanism)

Claims

1. A walking training device including a saddle-supported unloading device having a saddle and a walking mechanism that moves the saddle in a front direction relative to a floor surface, an angle sensor attached to the center of the waist of a person sitting on the saddle and detecting tilting movement of the pelvis in the frontal plane during walking training; and an estimation means for estimating, after the angle sensor detects that the orientation of the pelvis has crossed the vertical axis on the frontal plane from one side to the other, a second extreme point of the same polarity detected on the other side as the timing of kick-off of the ankle joint.

2. A walking training device including a saddle-supported unloading device having a saddle and a walking mechanism that moves the saddle in a front direction relative to a floor surface, an angle sensor attached to the saddle to detect a rotational movement of the saddle on a horizontal plane during walking training; and an estimation means for estimating, after the angle sensor detects that the orientation of the saddle has crossed the sagittal horizontal axis on the horizontal plane from one side to the other, the first inflection point detected on the other side as the timing of kicking off from the ankle joint.

3. an assist drive unit that performs walking assistance; a support signal generation unit that supports kicking off from the ankle joint; 3. The walking training device according to claim 1, wherein the support signal generator generates the support signal to be output to the support driver from the signal estimated as the ankle joint push-off timing.

4. The assistive driving unit includes positive and negative electrodes to which functional electrical stimulation is applied, The walking training device according to claim 3 , wherein the support signal generating unit generates, as the support signal, a stimulation signal to be applied to the positive and negative electrodes.

5. the assist drive unit is an exoskeleton-type ankle robot having an actuator that assists kicking off at the ankle joint, The walking training device according to claim 3 , wherein the support signal generating unit generates, as the support signal, a signal for driving an actuator of the exoskeleton-type ankle joint robot.

Images