JP2013048702A - Walking assistance device, and walking assistance program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To perform walking assistance with a constant load sense even when a walking manner varies.SOLUTION: This walking assistance device performs an assist of each joint such that an angular impulse (the time integration value of a joint moment) in one walking period of a wearer becomes constant regardless of a walking condition by paying attention to the angular impulse as a load amount of the wearer during the one walking period. Specifically, the walking assistance device sets an angular impulse obtained by subtracting an angular impulse assisted by the walking assistance device from the angular impulse in the one walking period in the walking condition that is a reference as a reference angular impulse z, and sets an assist rate β such that the angular impulse to be imposed on the wearer becomes equal to the reference angular impulse z even if the angular impulse increases by the change of the value of a walking parameter (the change of the walking manner). Thus, when the value of the walking parameter changes, walking with the constant load sense of the reference angular impulse z becomes possible.

Description

  The present invention relates to a walking support device and a walking support program, for example, a device that assists walking.

In recent years, wearable robots (power assist suits) that assist muscle strength used for human movement (walking, lifting, etc.) have been developed mainly in the nursing care business.
There are various types of wearable robots such as those that assist the muscle strength of the upper body, those that assist the muscle strength of the lower body, and those that assist the muscle strength of the whole body.
In addition, the use of the wearable robot ranges from a healthy person to an elderly person / person with a disability.

  The wearable robot, for example, analyzes the movement of the muscle from the electromyogram of the wearer or analyzes the movement of the wearer detected by the posture sensor arranged in each part of the joint, thereby obtaining the joint moment necessary for the movement. The necessary assist force corresponding to this is generated. Thus, the wearer can easily lift and walk heavy objects.

As such a technique, there is “a wearable movement assist device, a control method of a wearable movement assist device, and a control program” in Patent Document 1.
In this technique, a biological signal is detected by a sensor installed on the wearer, and the actuator is used to generate power according to the wearer's intention.

However, conventional wearable robots generate assist force at a fixed rate with respect to the joint moment generated at the joint of the wearer, and each walking scene (flat ground, stairs, slope, etc.) and each walking parameter value Assistance is not performed according to the difference (step length, pace, etc.).
For this reason, even if the assist is performed at the same rate, the wearer may feel that the assist power is insufficient depending on the walking scene and the walking method (difference in walking parameter value).

JP 2005-95561 A

  An object of the present invention is to provide walking support with a constant sense of burden even when walking is different.

(1) In the invention described in claim 1, reference angular impulse product storage means for storing a reference angular impulse product z that is a share of the walking support subject among angular impulse products necessary for walking in one walking cycle; A walking parameter value acquisition means for acquiring a value of a walking parameter for one walking cycle that defines a walking state in the walking support target person who is walking, and walking with the value of the walking parameter from the acquired walking parameter value An angular impulse product calculating means for calculating an angular impulse product M1 of one walking cycle necessary for the case, and an angular impulse product of one walking cycle borne by the walking support target with respect to the calculated angular impulse product M1 is the reference angular impulse product z The assisting force determining means for determining the assisting force and the determined assisting force are applied to the leg of the walking support target so that the walking support target can be applied to the joint of the leg. Providing a walking assistance device, characterized by comprising: a walking support means joint moment to reduce the assist walking of the walking support subject to produce a.
(2) In the invention described in claim 2, the overlapped walking distance and overlapping pace for one walking cycle are used as the walking parameter, and the walking parameter value acquisition means walks the overlapping walking distance and overlapping pace value for one walking cycle. The walking assist device according to claim 1, wherein the walking assist device is obtained as a parameter value.
(3) In the invention according to claim 3, there is provided regression equation storage means for storing a regression equation having a walking parameter as a variable and an angular impulse of one walking cycle as a solution, and the value of the walking parameter for deriving an optimal solution in the regression equation As a reference walk, the angular impulse product of one walking cycle that should be borne by the walking support target in the reference walk is defined as a reference angular impulse product, and the angular impulse product M1 uses the obtained walking parameter value as the regression equation. The walking support device according to claim 1, wherein the walking support device is calculated by substitution.
(4) In the invention according to claim 4, the regression equation storage means uses the value of each walking parameter and the angular impulse for one walking cycle measured by walking in various walking states as walking data, The walking support device according to claim 3, wherein a regression equation derived from a response surface method or multiple regression analysis is stored from walking data.
(5) In the invention according to claim 5, the walking parameter value acquisition means acquires a walking parameter for each walking cycle of the target leg when any one of the left and right legs is set as the target leg. The angular force product calculating means calculates the angular force product M1 of the target leg for each of the acquired walking parameters of each walking cycle, and the assist force determining means is calculated for the previous walking cycle. The walking support according to any one of claims 1 to 4, wherein the assist force of the next one walking cycle with respect to the target leg is determined from the angular force product M1 of the target leg. Providing equipment.
(6) The invention according to claim 6 further includes a walking surface determination unit that determines the type of walking surface on which the walking support target person walks, and the reference angular impulse storage unit stores the reference angular impulse product for each type of walking surface. z is stored, and the angular impulse product calculating means calculates an angular impulse product M1 of one walking cycle necessary when walking with the value of the walking parameter from the value of the acquired walking parameter and the determined type of walking surface. Then, the assist force determining means stores, for the calculated angular impulse product M1, the angular impulse product of one walking cycle borne by the walking support target person in correspondence with the determined type of walking surface. The walking assist device according to any one of claims 1 to 5, wherein the assist force is determined so as to be an angular force product z.
(7) In the invention according to claim 7, a walking parameter value acquisition function for acquiring a walking parameter value for one walking cycle that defines a walking state in a walking support target person during walking, and the acquired walking parameter value From the above, the angular impulse product calculation function for calculating the angular impulse product M1 of one walking cycle required when walking with the value of the walking parameter, and one walking that the walking support target person bears for the calculated angular impulse product M1 An assist force determining function for determining the assist force so that the angular impulse of the cycle becomes a reference angular impulse z that is a share of the walking support target among the angular impulses necessary for walking in one walking cycle; By applying the determined assist force to the legs of the walking support target person, the walking moment is reduced by reducing the joint moment generated by the walking support target person at the joint of the leg part. Providing walking support program for realizing a walking support function of supporting walking, to the computer.

  According to the present invention, even if the walking is different, the walking support target person can walk with a constant angular impulse load feeling by performing the walking support with the necessary assist force.

It is explanatory drawing showing the relationship between the time in 1 walk cycle, and a joint moment at the time of walking on a flat ground in two types of walking states. It is the figure which showed the mounting state of the mounting type robot. It is the figure which showed the system configuration | structure of the mounting | wearing type robot. It is a flowchart showing the operation | movement of the pre-processing in a walk assistance apparatus. It is explanatory drawing of the walk data used in a prior process. It is the graph which represented the regression type calculated | required from the walking data in three dimensions. It is a flowchart showing a walking movement determination process. It is a flowchart for demonstrating the procedure in which a control part performs walking assistance. It is a flowchart for demonstrating a back movement process. It is the flowchart which showed the operation | movement of the walk assistance process.

(1) Outline of Embodiment The wearable robot 1 (FIG. 2) is a wearer that analyzes the movement of muscles from the electromyogram of the wearer (walking support target) or is detected by posture sensors arranged in each part of the joint. The movement of each joint is calculated, the joint moment required for the movement is calculated, and the necessary assist force corresponding to this is generated to support the movement of each joint. For example, the hip joint assist actuator 17 supports the hip joint operation, the knee joint assist actuator 18 supports the knee joint operation, and the ankle joint assist actuator 19 supports the ankle joint operation.
That is, when the wearer generates a joint moment for walking, the wearable robot 1 drives each assist actuator to reduce the joint moment generated by the wearer.

In general, when a person walks, the joint moments required for each joint differ depending on the walking state such as the stride and pace. For example, each joint moment when the stride is widened or when walking with a fast foot is larger than each joint moment when the easiest way of walking is used as a reference.
Therefore, even when 50% of the joint moments are uniformly assisted, the joint moment and force that the wearer should bear will change.

Therefore, in the present embodiment, attention is paid to the angular impulse product (time integrated value of joint moment) as a burden amount of the wearer during one walking cycle, and the angular impulse product in the one walking cycle of the wearer is constant regardless of the walking state. Assist each joint so that
That is, the angular impulse product obtained by removing the angular impulse product assisted by the walking support device from the angular impulse product in one walking cycle of the reference walking state (= an angular impulse product borne by the wearer in the reference walking) is set as the reference angular impulse product z, and walking Even if the angular impulse is increased by changing the parameter value (the way of walking is changed), the assist rate β is set so that the angular impulse (bearing angular impulse) to be borne by the wearer becomes equal to the reference angular impulse z. Set.
As a result, even if the value of the walking parameter changes, it is possible to walk with a sense of burden due to a constant reference angular impulse z without being conscious of the force output by the person (burden angular impulse).
The assist rate refers to the ratio (unit:%) of the assist force to the force required for walking.

FIG. 1 shows the relationship between time and joint moment in one walking cycle when walking on a flat ground in two types of walking states.
In FIG. 1, the solid line represents the joint moment of a specific joint in one walking cycle. On the other hand, the dotted line is the joint moment borne by the wearer by assist, and the angular value product is obtained by time-integrating the absolute value over one walking cycle, and the total area of the hatched portion in the figure is Equivalent to.

  Note that one walking cycle is the time required for the movement from the left or right foot to ground until the next foot on the next landing (this is called “overlap walking”). The stance phase is about 60% in the first half and the swing phase is about 40% in the second half, and the direction of the moment is reversed according to both phases.

And the way of walking (walking state) in which the angular impulse product in one walking cycle becomes the minimum value is defined as the reference walking. The change of the joint moment in the standard walking is shown by the solid line in FIG. 1A, and the area of the solid line part is the minimum value of the angular impulse product without assistance, and this value is 20 (Nms). To do.
When assisting with the assist rate α = 50% with respect to the reference walking, the wearer only has to bear an angular impulse product (= reference angular impulse product z) of a total of 10 Nms shown by the oblique lines in FIG. Become.

On the other hand, the solid line in FIG. 1B represents a change in joint moment with respect to walking in one walking cycle actually performed in a walking method (walking state) different from the reference walking.
In the present embodiment, the angular impulse (bearing angular impulse) borne by the wearer is the same as the reference angular impulse z = 10 Nms, as indicated by the oblique lines in FIG. Thus, the assist rate β is determined, and the corresponding joint is assisted at the assist rate β.

In the present embodiment, a parameter for specifying how to walk (walking state) is set as a walking parameter, and an overlapping pace and an overlapping distance are used. Then, the angular impulse product (calculated from the joint moment) with respect to the walking parameter value when walking in various walking states is measured in advance as walking data.
A response surface method, multiple regression analysis, or the like is used for the walking data to derive a regression equation using the walking parameters as variables and the angular impulse of one walking cycle as a solution. This regression formula is for the right leg for each walking scene (walking surface types (flat, uphill, downhill, upstairs, downstairs) 5 x traveling direction (front, back) 2 types = 10 scenes). , For each left leg joint (hip joint, knee joint, ankle joint), that is, 10 × 2 × 3 = 60.
In each derived regression equation, the walking state having the value of the walking parameter that leads to the optimum solution (minimum angular impulse) is set as the reference walking.

And when walking is actually started, the angular impulse product M1 for the walking parameter value before one walking cycle is obtained from the regression equation, and it is assumed that the next one walking cycle is the same walking parameter value as the previous walking cycle, The assist rate β is determined so that the burden angular impulse product for the next overlapping step becomes equal to the reference angular impulse product z, and the assist rate β is assisted for the next overlapping step.
In the example shown in FIGS. 1A and 1B, the angular impulse product M1 obtained from the walking parameter value and the regression equation is 30 Nms, and the reference angular impulse product z is 10 Nms, so 20 Nms (= 30-10) is supported for walking. By assisting with the apparatus, burden angular impulse product = reference angular impulse product z.
Therefore, since the assist rate β = {(M1-z) / M1} × 100 = 66% (= (20/30) × 100) is obtained, 66% of the joint moment is assisted with respect to the next overlapping walking. Thus, the wearer can always walk with a constant angular impulse feeling.

(2) Details of Embodiment FIG. 2 is a diagram showing a wearing state of the wearing robot 1.
The wearable robot 1 is worn on the waist and lower limbs of the wearer to assist (assist) the wearer's walking.
The wearable robot 1 includes a waist attachment part 21, an upper leg attachment part 22, a lower leg attachment part 23, a foot attachment part 24, an upper leg connection member 26, a lower leg connection member 27, a control device 2, a toe reaction force sensor 10, Force sensor 11, toe posture sensor 12, heel posture sensor 13, waist posture sensor 14, upper leg posture sensor 15, lower leg posture sensor 16, hip joint assist actuator 17, knee joint assist actuator 18, ankle joint assist actuator 19 and the like are provided. Yes. In addition, except for the waist mounting part 21, the control device 2, and the waist posture sensor 14, they are provided on both the left and right feet, and the respective detected values are output.
However, the toe reaction force sensor 10 and the heel reaction force sensor 11 may be provided with a toe grounding sensor and a heel grounding sensor instead of both sensors in the embodiment in which the detection of the reaction force is unnecessary. .

The waist mounting portion 21 is attached around the waist of the wearer and fixes the wearable robot 1.
The waist posture sensor 14 is attached to the waist attachment portion 21 and detects the posture of the waist (roll angle, yaw angle, pitch angle) with a gyroscope or the like. Also, by differentiating these angles, the angular velocity and angular acceleration of the waist can be obtained.

The control device 2 is attached to the waist mounting portion 21 and controls the operation of the wearable robot 1.
The hip joint assist actuator 17 is provided at the same height as the hip joint of the wearer, and drives the upper thigh coupling member 26 in the front-rear direction with respect to the waist mounting portion 21. Note that the hip joint assist actuator 17 may be configured to be driven in the lateral direction as a three-axis actuator.

The upper thigh connecting member 26 is a rigid columnar member provided outside the upper thigh of the wearer, and connects the hip joint assist actuator 17 and the knee joint assist actuator 18.
The outer side of the upper thigh mounting part 22 is fixed to the inner side of the upper thigh coupling member 26, and the inner side is fixed to the upper leg of the wearer.
The upper leg posture sensor 15 detects the upper leg posture (roll angle, yaw angle, pitch angle). In addition, by differentiating these angles, the angular velocity and acceleration of the upper thigh can be obtained.

The knee joint assist actuator 18 is provided at the same height as the knee joint of the wearer, and drives the lower leg connecting member 27 in the front-rear direction with respect to the upper leg connecting member 26.
The lower leg connecting member 27 is a columnar member having rigidity provided outside the lower leg part of the wearer, and connects the knee joint assist actuator 18 and the ankle joint assist actuator 19.

The outer side of the lower leg mounting portion 23 is fixed to the inner side of the lower leg connecting member 27, and the inner side is fixed to the lower leg of the wearer.
The lower leg posture sensor 16 detects the lower leg posture (roll angle, yaw angle, pitch angle). Also, by differentiating these angles, the angular velocity and angular acceleration of the lower leg can be obtained.

The ankle joint assist actuator 19 is provided at the same height as the wearer's ankle joint, and drives the toe of the foot mounting portion 24 in the direction of moving up and down with respect to the crus coupling member 27.
The foot mounting portion 24 is fixed to the foot portion (instep and sole) of the wearer. Generally, the joint at the base of the toe is bent during walking, but the foot mounting portion 24 is also configured so that the base of the toe is bent according to the toes.

  The toe posture sensor 12 and the heel posture sensor 13 are respectively installed at the front end and the rear end of the foot mounting portion 24 and detect the toe and heel postures (roll angle, yaw angle, pitch angle), respectively. Further, by differentiating these angles, the angular velocity and angular acceleration of the toes and the heel can be obtained.

The toe reaction force sensor 10 is installed in front of the sole portion of the foot mounting portion 24 and detects the ground contact of the toe and the reaction force from the walking surface.
The heel reaction force sensor 11 is installed on the rear side of the sole of the foot mounting portion 24 and detects the ground contact of the heel and also detects the reaction force from the walking surface.
The wearable robot 1 configured as described above supports the walking of the wearer by driving the hip joint assist actuator 17, the knee joint assist actuator 18, and the ankle joint assist actuator 19.

FIG. 3 is a diagram showing a system configuration of the wearable robot 1.
The control device 2 includes an electronic control unit including a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a clock as means for measuring time, a storage unit 7, various interfaces, and the like. And each part of the wearable robot 1 is electronically controlled.

The control device 2 is also configured by executing various programs such as a walking support program stored in the storage unit 7 by the CPU, and includes a sensor information acquisition unit 3, various parameter calculation units 4, a walking motion determination unit 5, A walking scene determination unit 6 and a walking assist force determination unit 8 are provided.
The sensor information acquisition unit 3 acquires a detection value from each of the toe reaction force sensor 10 to the crus posture sensor 16. The detection values of each sensor acquired by the sensor information acquisition unit 3 are used for determination of walking motion (FIG. 7), determination of walking scene (FIGS. 8 and 9), calculation of walking parameters (FIG. 10), and the like. The

The various parameter calculation unit 4 calculates walking parameter values (overlapping pace and overlapping walking distance) by obtaining the angles and positions of the joints from the detection values acquired by the sensor information acquisition unit 3.
Here, an operation from the contact of one side of the heel to the contact of the next side of the heel is referred to as an overlapping step, and a series of operations in the overlapping step is referred to as a walking cycle. The distance between the two ground contact points in the overlapping step is referred to as the overlapping step distance, and the number of overlapping steps per minute (the number of overlapping steps / minute) is referred to as the overlapping step.

The walking motion determination unit 5 determines whether the wearer's motion is a motion other than walking such as a bending / stretching motion or a stepping motion, or an actual walking motion.
The walking scene determination unit 6 determines a walking scene in which the wearer is walking from the detection value acquired by the sensor information acquisition unit 3. There are two types of walking scenes that can be judged: walking direction (flat, upstairs, downstairs, uphill, downhill) for each of the five walking directions of forward walking and backward progress. There are 10 walking scenes.

  The walking assist force determining unit 8 determines the assist force to be output to the hip joint assist actuator 17, the knee joint assist actuator 18, and the ankle joint assist actuator 19 disposed on the left and right feet, and drives these assist actuators accordingly. . The assist force is a moment (torque) that the wearable robot 1 drives the assist actuator to act on the legs.

Next, various processes and operations in the walking support device configured as described above will be described.
FIG. 4 shows a pre-processing operation in the walking support device.
This process is a process for obtaining a reference angular impulse product z and a regression equation for each joint of the left and right feet in each of the 10 types of walking scenes described above. Since this pre-processing is not performed for each walking support device but is common to each wearer, it is processing that is appropriately performed along with walking data. In this state, the description will be made on the assumption that the control device 2 of the walking assistance device performs, but the reference angular impulse product z and the regression equation obtained by performing the pre-processing with another control device are stored and used in each walking assistance device. You may make it do.

First, in the pre-processing, the control device 2 acquires walking data for each walking parameter value measured in advance (step 301).
FIG. 5 shows the walking data to be acquired.
As shown in FIG. 5, the walking data is a value of an angular impulse with respect to the knee joint when walking with various walking parameter values (overlapping pace, overlapping walking distance). In the example of FIG. 5, the angular impulse of one walking cycle is measured as 18.3275 (Nms) when walking at an overlapping step distance of 1.0 m at an overlapping pace (walking tempo) with 30 overlapping steps per minute. .

The angular impulse value for each walking parameter value is measured and calculated as follows.
That is, to a subject with markers at predetermined positions such as joints, toes, and heels, walking corresponding to each walking parameter value, for example, walking with overlapping steps 30 (number of overlapping steps / min) and overlapping walking distance 1.0 m. And take pictures with multiple cameras.
From this image and by analyzing the movement of the marker, as shown in FIG. 1, the change of the joint moment for each joint is obtained, and the joint moment of one walk cycle is time-integrated to obtain one walk cycle. The angular impulse is obtained.
Instead of having the subject walk according to each walking parameter value, ask the subject to freely walk in various patterns, and find the walking parameter value and the change in joint moment from the movement of the marker at that time Thus, the angular impulse product (walking data) of one walking cycle corresponding to each walking parameter value may be obtained.

Returning to FIG. 4, the control device 2 uses the acquired walking data and the response surface method to derive a regression equation Z having the walking parameter value as a variable and the angular impulse of one walking cycle as a solution (step 302).
By obtaining a regression equation from the walking data, it is possible to obtain a reference walking and a reference angular impulse z (to be described later), which will be described later, or to be detected in walking support without measuring in detail the angular impulse product for all walking parameter values. An angular impulse product with respect to the walking parameter value can be obtained.
As an example, the regression equation Z (knee) obtained from the walking data for the knee joint illustrated in FIG.
In the expression, the parameter X represents the overlapping pace, and Y represents the overlapping distance. In the formula, @ 2 represents the square of the previous formula / numerical value.

Z (knee) = − 0.149 × X
+ 7.771 × Y
+ 0.825 × (X−42.7594) × (Y−1.1174)
+0.015×(X−42.7594)@2
+20.218×(Y−1.1174)@2
+11.070

FIG. 6 visualizes the regression equation Z (knee) for the knee joint obtained from the walking data of FIG. 5 with a three-dimensional graph.
As shown in FIG. 6, the regression equation Z is used to apply the overlapping gait value of the walking parameter value detected in the walking support to the X axis and the overlapping walking distance value to the Y axis. The angular impulse (without assist) required when walking with the direction parameter value is obtained.

Next, the control device 2 calculates the reference walking and the angular impulse of one walking cycle from the calculated regression equation Z (step 303).
That is, in the calculated regression equation Z, a walk having a walking parameter value for deriving an optimal solution (minimum angular impulse product) is set as a reference walking, and an angular impulse product of one walking cycle in the reference walking is calculated.
In the case of the regression equation Z (knee) derived from the above-mentioned walking data of FIG. 5, the reference walking is performed when the walking parameter value is the overlapping pace 51 (overlapping steps / minute) and the overlapping pace is 1.0 (m). The angular impulse (minimum value) is 11.7411 (Nms).

Then, the control device 2 determines the assist rate α in the reference walking (step 304).
Note that this assist rate is a default value, and the wearer of the walking assistance device may be able to adjust (increase / decrease) each one.

Next, the control device 2 stores the reference angular impulse product z and the regression equation for the reference walk in the storage unit 7 (step 305), and ends the process.
Here, the reference angular impulse product z is a value obtained by multiplying the angular impulse product M2 of one walking cycle in the reference walking by (1−assist rate α × 0.01), and the walking from the angular impulse product M2 of one walking cycle in the reference walking. It is an angular impulse product (= an angular impulse product which a wearer bears in a standard walk) excluding the angular impulse product (M2 × α × 0.01) assisted by the support device. That is, z = M2 × (1−α × 0.01).
The reference angular impulse product z and the regression equation stored in step 305 are calculated and stored for each of the ten walking scenes, the left and right legs, and each joint (hip joint, knee joint, ankle joint). . That is, the reference angular impulse product z and the regression equation of 10 × 2 × 3 = 60 in total are stored in the storage unit 7.

Next, walking support processing after mounting an actual walking support device will be described.
This walking support process is separately processed and controlled for each of the right leg and the left leg. In the following description, the leg on the side that is the target of processing and control will be described as the target leg.
In this walking support process, a walking motion determination process and a walking scene determination process are performed as subroutines, so both processes will be described first.
This walking motion determination process is executed by the control device 2 as the processing of the walking motion determination unit 5, and whether the wearer's motion is a motion other than walking such as a bending motion or a stepping motion, or actually walking This is a process for determining whether the operation is in progress.

FIG. 7 is a flowchart illustrating the walking motion determination process.
The control device 2 acquires a parameter value for determining walking motion from the sensor information acquisition unit 3 (step 311). Note that the walking motion determination process is continuously performed from the time when the walking support device is mounted and the power is turned on, and the parameter value to be acquired is continuously acquired.

And the control apparatus 2 performs walking movement determination from the parameter value to acquire (step 312).
In this walking movement determination, from the output value (parameter value) of each sensor, if the wearer is not moving, or if it is moving, for example, if it just bends and stretches, if only one leg is raised or lowered, The determination is made by distinguishing movements other than walking, such as when stepping on the field, and actual walking movements.
Specifically, the detected values of the toe posture sensor 12, the heel posture sensor 13, the waist posture sensor 14, the upper leg posture sensor 15 and the lower leg posture sensor 16 (hereinafter, abbreviated as posture sensors 12 to 16). From the output values of the toe reaction force sensor 10 and the heel reaction force sensor 11 (hereinafter abbreviated as reaction force sensors 10, 11), the foot on the same side is calculated by calculating the angle of each joint from Is determined to be a walking motion when it is determined that has touched the front or back.

  In addition, about determination of whether it is a walking motion, it is a walking motion when a standing phase and a swing phase are determined from the motion of the foot calculated from the detection values of the reaction force sensors 10 and 11 and the posture sensors 12 to 16. May be determined.

As a result of the walking motion determination (step 312), when it is determined that it is not a walking motion (step 313; N), the control device 2 determines whether or not the currently set assist rate is the default assist rate (step 312). 314).
If it is the default assist rate (step 314; Y), the control device 2 returns to step 311 without changing the assist rate, and continues monitoring and determination of walking motion.
On the other hand, when it is not the default assist rate (step 314; N), the control device 2 changes the currently set assist rate to the default assist rate (step 315), and then returns to step 311 to perform the walking motion. Continue monitoring and judgment.
When it is determined that the walking motion is determined in the walking motion determination performed in step 312 (step 313; Y), the control device 2 returns to the main routine (FIG. 10).

Next, the walking scene determination process will be described with reference to FIGS.
In this walking scene determination processing, any one of a total of 10 types of walking scenes, that is, two types of traveling directions (forward and backward) with respect to five types of walking surfaces (flat ground, up stairs, down stairs, uphill road, downhill road). It is determined whether it is a walking scene.
FIG. 8 is a flowchart for explaining a procedure in which the control device 2 performs walking support when the wearer starts a walking cycle.
First, when the wearer starts the walking cycle, the control device 2 detects that the foot mounting portion 24 has moved away from the walking surface based on the detection values from the toe reaction force sensor 10 and the heel reaction force sensor 11, and one foot is It is determined that it has surfaced (step 5).

Next, the control device 2 includes a toe posture sensor 12, a heel posture sensor 13, a waist posture sensor 14, an upper leg posture sensor 15 and a lower leg posture sensor 16 (hereinafter referred to as a toe posture sensor 12 and a heel posture sensor 13). The angle of each joint is calculated from the detected values of the waist posture sensor 14, the upper leg posture sensor 15, and the lower leg posture sensor 16 (abbreviated as posture sensors), and the moving direction of the lifted foot is detected (step 10).
Next, the control device 2 determines whether or not the lifted foot is moving in the forward direction (step 15).
When it is determined that the lifted leg is moving in the forward direction (step 15; Y), the control device 2 determines that the wearer moves forward (step 20).
On the other hand, when determining that it has not moved in the forward direction (step 15; N), the control device 2 performs a backward movement process described later (step 110).

When the control device 2 determines to move forward (step 20), the control device 2 determines whether the front foot is in contact with the walking surface based on the detection values from the front toe reaction force sensor 10 and the heel reaction force sensor 11. Judgment is made (step 25).
When it is determined that the front foot is not in contact with the walking surface (step 25; N), the control device 2 continues monitoring whether the front foot is in contact with the walking surface in step 25, and forward When the control device 2 determines that the foot of the foot touches the walking surface (step 25; Y), the control device 2 calculates the angle of each joint of both feet from the output of the posture sensor of both feet, thereby calculating the height of the front and rear feet. Measure (step 30).
As the height of the foot, for example, the height of an appropriate part such as the height of the heel (heel posture sensor 13) or the height of the ankle joint (ankle joint assist actuator 19) can be used.

Next, the control device 2 determines whether or not there is a difference between the heights of the front and rear legs (step 35). If there is no difference (step 35; N), the control device 2 determines that the walking scene is flat. It is determined that the vehicle is moving forward (step 40), and the process returns to the main routine (FIG. 10).
When there is a difference in the height of the front and rear feet (step 35; Y), the control device 2 tilts the front and rear of the foot mounting portion 24 of the foot that has been put forward based on the detection values of the toe posture sensor 12 and the heel posture sensor 13. Is measured (step 50).

Next, the control device 2 determines whether or not there is a forward / backward inclination (step 55). If there is no forward / backward inclination (step 55; N), the control device 2 determines that the front foot is based on the detection value of the posture sensor. It is determined whether or not it is higher than the rear foot (step 60).
When the front foot is not high, that is, when the front foot is low (step 60; N), the control device 2 determines that the walking scene is forward descent (step 65), and returns to the main routine. .

When the front foot is higher than the rear foot (step 60; Y), the control device 2 determines that the walking scene is an ascending stairs advance (step 75), and returns to the main routine.
In addition, when the front foot mounting portion 24 has a front / rear inclination (step 55; Y), the control device 2 determines that the toes are heels based on the detection values of the front foot toe posture sensor 12 and the heel posture sensor 13. It is judged whether it is lower than (step 85).

When the toe is lower than the heel (step 85; Y), the control device 2 determines that the walking scene is a forward movement on a downhill (step 90), and returns to the main routine.
If the toe is not lower than the toe, that is, if the toe is lower than the toe (step 85; N), the control device 2 determines that the walking scene is an advance on an uphill road (step 100), and enters the main routine. Return.

  As described above, the wearable robot 1 determines whether the wearer walks forward on a flat ground, descending stairs, ascending stairs, descending hill, or ascending hill from the toe, the presence or absence of ground contact, and the posture of the lower limbs. The type of the walking surface can be determined.

FIG. 9 is a flowchart for explaining the backward movement process in step 110 (FIG. 8).
The control apparatus 2 judges that it is a back movement, when the leg which floated moves back (step 120).
Next, the control device 2 determines whether or not the rear foot is in contact with the walking surface based on the detection values of the toe reaction force sensor 10 and the heel reaction force sensor 11 (step 125).
When the back foot is not in contact with the walking surface (step 125; N), the control device 2 monitors whether or not the back is grounded in step 125.

When the rear foot is in contact with the walking surface (step 125; Y), the control device 2 calculates the angle of each joint of both feet based on the detection value of the posture sensor of both feet, thereby calculating the height of the front and rear feet. Measure (Step 130).
When there is no difference in the height of the front and rear legs (step 135; N), the control device 2 determines that the walking scene is backward on the flat ground (step 140), and returns to the main routine (FIG. 10).

When there is a difference in the height of the front and rear feet (step 135; Y), the control device 2 measures the front and rear inclination of the foot mounting portion 24 of the rear foot based on the detection values of the toe posture sensor 12 and the heel posture sensor 13. (Step 150).
Next, the control device 2 determines whether or not there is a forward / backward inclination (step 155). If there is no forward / backward inclination (step 155; N), the control device 2 determines that the front foot is based on the detection value of the posture sensor. It is determined whether it is higher than the rear foot (step 160).
When the front foot is high (step 160; Y), the control device 2 determines that the walking scene is backward going down the stairs (step 165), and returns to the main routine.

When the front foot is not high, that is, when the front foot is low (step 160; N), the control device 2 determines that the walking scene is the backward movement of the upstairs (step 175) and returns to the main routine. .
Further, when the foot mounting portion 24 of the rear foot has a front-back inclination (step 155; Y), the control device 2 determines that the toe is more than the heel from the detection values of the rear foot toe posture sensor 12 and the heel posture sensor 13. It is determined whether it is high (step 185).

When the toe is higher than the heel (step 185; Y), the control device 2 determines that the walking scene is a backward movement on the downhill (step 190), and returns to the main routine.
When the toe is lower than the heel (step 185; N), the control device 2 determines that the walking scene is backward on the uphill road (step 200), and returns to the main routine.

  As described above, the wearable robot 1 determines whether the wearer goes back on a flat ground, descending stairs, ascending stairs, descending hill, or ascending hill according to the presence or absence of toes, heel contact, and lower limb posture. The walking scene can be determined.

Next, a walking assist process using the above-described walking motion determination process and walking scene determination process as a subroutine will be described.
FIG. 10 is a flowchart showing the operation of the walking assist process.
The control device 2 starts to operate when the power is turned on after the walking support device is mounted (step 321).
Then, the walking motion determination process described with reference to FIG. 7 is performed (step 322), and when it is determined that the motion is a walking motion (step 313; Y), the control device 2 performs the walking scene determination processing described with reference to FIGS. (Step 323), the walking scene corresponding to the wearer's walking is determined and acquired.

During or after the walking scene determination process, the control device 2 causes the wearer to end the walking cycle by the target leg from the detection values of the reaction force sensors 10 and 11 and the posture sensors 12 to 16 (step 324). Then, it is detected that the next operation is started (step 325).
Then, the control device 2 determines again whether or not the next started operation (step 325) is a walking operation (step 326).
Since the walking assist process is performed separately for each of the left and right legs, when the right leg walking cycle is completed in step 324, the walking determination process for the next action in step 325 is also performed for the right leg. Is called. The subsequent processing is the same.

When it is determined that the walking motion is determined in the walking motion determination process in step 326 (step 313), the control device 2 determines whether the walking cycle (step 324) ended immediately before is the first walking cycle after the start of the walking motion. Determination is made (step 327).
In this embodiment, the walking assist process determines the assisting power for the next overlapping step from the walking parameter value of the previous walking cycle in the overlapping step, but the walking is accurate from the first walking cycle. This is because the parameter cannot be acquired. For example, if you start walking with your right leg out of a standing state, the walking parameter value obtained in the first walking cycle will be about ½ of the normal, and this value will also start walking in the first walking cycle. This is because the movement is for the purpose of entering the correct value.
Note that the controller 2 determines the assist force for each joint using the default assist rate α with respect to the motion in the first one walking cycle and the walking motion in the second walking cycle.

Therefore, when it is determined that it is the first walking cycle after the start of the walking motion (step 327; Y), the control device 2 returns to step 323 and processes the next one walking cycle.
On the other hand, when it is determined that it is not the first walking cycle (step 327; N), that is, when it is one walking cycle after the second time, the control device 2 determines the walking parameters of one walking cycle of the target leg that has just ended. A value is calculated (step 328).
That is, the control device 2 uses the detected values of the reaction force sensors 10 and 11 and the posture sensors 12 to 16 acquired during one walking cycle ended in step 324 as the walking parameter value in the one walking cycle as the overlapping walking distance y. And the overlapping pace x is calculated.
For example, the control device 2 measures the time from the contact of the heel to the next contact, and when one walking cycle is measured as 2 seconds, the overlapping pace x is x = 60/2 = 30 (number of overlapping steps / minute) Is calculated.

For example, if the target leg is a right leg, the control device 2 determines the distance (step length) of one step of the left leg in one walking cycle and the distance of the next step of the right leg for each step on the left and right. The angle of each joint is calculated from the detection values of the posture sensors 12 to 16 when both feet are in contact with the ground. Then, the control device 2 calculates the angle of each joint at each calculated left and right step, the distance between each joint, the distance between the ankle joint and the sole, and the length of the foot (distance between the reaction force sensors 10 and 11). ) To calculate the overlapping walking distance y in the one walking cycle.
Specifically, the distance from the toe reaction force sensor 10 on the right leg to the toe reaction force sensor 10 on the left leg (the width of one step on the left leg) in the state shown in FIG. The total distance (width of one step of the right leg) in the state where the right and left toe sensors detect the ground contact in the state where the right leg is extended is calculated from the angle of each joint and the distance between each joint in both states. The overlapped distance y is calculated.
In addition, although the grounding of the toe reaction force sensor 10 which can grasp | ascertain the state grounded simultaneously right and left is used, the length of the foot (distance between the toe reaction force sensor 10 and the heel reaction force sensor 11) is constant. Therefore, the overlapping walking distance y can be calculated.

Next, the control device 2 reads the three regression equations for the hip joint, knee joint, and ankle joint stored in the storage unit 7 in advance for the walking scene determined in step 323, and sets the variables X and Y as variables. By substituting the walking parameter values (overlapping distance y, overlapping pacing x) calculated in step 328, the angular impulse product M1 of the target leg 1 walking cycle is calculated for each joint (step 329). .
The calculated angular impulse product of each joint is a total angular impulse required for the next overlapping step determined in steps 325 and 326 (angular impulse product when not assisting).

Next, the control device 2 uses the reference angular impulse product z of one walking cycle stored in the storage unit 7 and the angular impulse product M1 calculated in Step 329, so that the assist rate for each joint in the target leg as a target. β is calculated (step 330).
Here, the assist rate β is calculated by the following equation, where M1 is an angular impulse product of one walking cycle, M2 is an angular impulse product of one walking cycle, and α is an assisting rate of the reference walking cycle.

  Assist rate β = {{M1-M2 × (1-α × 0.01)} / M1} × 100 (%)

  Since the angular impulse product borne by the wearer at the time of assisting the reference walking is the reference angular impulse product z and the reference angular impulse product z = M2 × (1−α × 0.01), the assist rate β for a certain walking cycle Is as follows.

  Assist rate β = {(M1-z) / M1} × 100 (%)

Next, the control device 2 determines an assist force (= joint moment × assist rate β) for each joint of the target leg from the calculated assist rate β for each joint, and becomes an object according to the determined assist force. The hip joint assist actuator 17, the knee joint assist actuator 18, and the ankle joint assist actuator 19 of the target leg are driven (step 331).

  The control device 2 determines whether or not the power of the walking support device is turned off (step 332), and when it is determined that the walking support device is not turned off (step 332; N), returns to step 323 to continue walking support and determines that it is turned off (step 332). Step 332; Y), the walking support process is terminated.

  As described above, according to the walking support device of the present embodiment, even if the angular state product is focused on as the amount of burden on the wearer during one walking cycle and the walking state (the value of the walking parameter) changes, the wearer Since the configuration is such that each joint is assisted so that the angular impulse product in one walking cycle becomes a constant value (reference angular impulse product), the wearer walks with the same angular impulse burden even when the walking state changes. be able to.

As mentioned above, although one embodiment in the walking support device of the present invention was described, the present invention is not limited to the described embodiment, and various modifications can be made within the scope described in each claim. .
For example, in the embodiment described above, the walking method (walking state) in which the angular impulse product in one walking cycle is the minimum value, that is, the walking having the walking parameter value that leads to the optimum solution (minimum angular impulse product) in the derived regression equation. Although the state is defined as the reference walking, another walking state may be defined as the reference walking.
For example, walking with the walking parameter value having the highest frequency in the wearer's walking may be used as the reference walking. In this case, it is possible to obtain the walking parameter value that is the most frequent walking in a predetermined time immediately before (for example, 30 minutes) as the reference walking, or to obtain the walking parameter value that is most frequent in the walking on the previous day and use it as the reference walking.
Alternatively, with the wearer wearing the walking support device, the user walks freely for each walking scene in advance, measures the walking parameter value for each walking cycle, and has the largest number of walking scenes (walking A walking parameter value (which has been long) may be determined, and walking at the walking parameter value may be used as a reference walking. In this case, the walking assistance device may measure without walking assistance or may measure while walking assistance at a predetermined assist rate.

Further, according to the embodiment described above, a regression equation having a walking parameter as a variable and an angular impulse product in one walking cycle as a solution is derived from the walking data, and using this regression equation, walking while actually walking is performed. The case where the angular impulse product with respect to the parameter value is calculated has been described.
On the other hand, by measuring detailed walking data, a walking parameter-angular impulse product correspondence table is created and stored, and by using this correspondence table, the walking parameter value calculated at the time of actual walking is obtained. A corresponding angular impulse may be acquired.
In this case, when the walking parameter value calculated at the time of actual walking is not in the table, the angular impulse is determined using the closest value existing in the correspondence table.
In this case, the angular impulse of the reference gait may be determined using a separately derived regression equation. If the regression equation itself is not created, the angular impulse of the smallest value in the above correspondence table is calculated. It may be an angular impulse.

As the walking parameter, the overlapping distance Y and the overlapping pace X are used, but other parameters may be used.
For example, instead of the overlapping pace X, one walking cycle Q (seconds) may be used.
Further, in addition to the overlapping step distance Y and the overlapping step X (or one step period Q), the floor reaction force may be used as a walking parameter.
Furthermore, the determination of the walking scene may be made unnecessary by adding the difference in height when both feet are in contact with the ground and the angle of the toe joint as walking parameters.

  In the embodiment described above, the assist rate β is determined by using the angular impulse product and the reference angular impulse product. However, instead of the angular impulse product and the reference angular impulse product, the angular impulse / weight and the reference angular impulse / weight are used. It may be.

In the embodiment described above, the walking parameters are not calculated from the first walking cycle after starting the walking motion (step 327; Y), and the walking parameters are calculated from the second and subsequent walking cycles. explained. That is, the walking parameter is calculated from one walking cycle after the second time, and assist is performed for the walking motion in one walking cycle after the third time.
On the other hand, the walking parameter may be calculated from the first walking cycle after the start of the walking motion. In this case, the determination process in step 327 is not necessary.
When the first walking cycle by the target leg is the first motion at the start of walking, that is, when the walking motion is started from the target leg side, the overlapping walking distance calculated from the one walking cycle is calculated. A value obtained by doubling the value and the value of the overlapping pace is calculated as a walking parameter for one walking cycle (step 328).
On the other hand, when the first one walking cycle by the target leg is not the first motion at the start of walking, that is, when the walking motion is started from the leg on the opposite side of the target leg, the overlapping walking calculated from the one walking cycle is performed. The distance value and the overlapping pace value are calculated as walking parameters for one walking cycle (step 328).
According to this modification, it is possible to assist the overlapping steps in the second one walking cycle using the walking parameters calculated from the first one walking cycle.
In this case as well, the assist using the default assist rate is performed for the first walking operation of one walking cycle.

DESCRIPTION OF SYMBOLS 1 Wearable robot 2 Control apparatus 3 Sensor information acquisition part 4 Various parameter calculation part 5 Walking motion determination part 6 Walking scene determination part 7 Memory | storage part 8 Walking assist force determination part 10 Toe reaction force sensor 11 踵 Reaction force sensor 12 Toe posture sensor 13 Waist Posture Sensor 14 Waist Posture Sensor 15 Upper Leg Posture Sensor 16 Lower Leg Posture Sensor 17 Hip Joint Assist Actuator 18 Knee Joint Assist Actuator 19 Ankle Joint Assist Actuator 21 Lumbar Mount Part 22 Upper Thigh Mount Part 23 Lower Thigh Mount Part 24 Leg Mount Part 26 Upper Thigh connecting member 27 Lower thigh connecting member

Claims (7)

Reference angular impulse product storage means for storing a reference angular impulse product z that is a share of the walking support target among the angular impulse products necessary for walking in one walking cycle;
A walking parameter value acquisition means for acquiring a value of a walking parameter for one walking cycle that defines a walking state in the walking support target during walking;
Angular force product calculating means for calculating an angular force product M1 of one walking cycle necessary when walking with the value of the acquired walking parameter from the acquired walking parameter value;
An assist force determining means for determining an assist force so that an angular impulse product of one walking cycle borne by the walking support target person becomes the reference angular impulse product z with respect to the calculated angular impulse product M1;
By applying the determined assist force to the legs of the walking support target person, the walking moment of the walking support target person is reduced by reducing a joint moment generated at the joint of the leg part by the walking support target person. Walking support means;
A walking support device comprising: The walking distance for one walking cycle and the overlapping pace as the walking parameters,
The walking parameter value acquisition means acquires the value of the overlapping walking distance and the overlapping pace for one walking cycle as the value of the walking parameter,
The walking support apparatus according to claim 1. A regression equation storage means for storing a regression equation having a walking parameter as a variable and an angular impulse of one walking cycle as a solution;
As the standard walking is the walking by the value of the walking parameter that leads to the optimal solution in the regression equation, the angular impulse of one walking cycle that the walking support target person should bear in the standard walking is the standard angular impulse,
The angular impulse product M1 is calculated by substituting the value of the acquired walking parameter into the regression equation.
The walking support device according to claim 1 or claim 2, wherein The regression equation storage means uses each walking parameter value and angular impulse for one walking cycle measured by walking in various walking states as walking data, and from the walking data by response surface method or multiple regression analysis. Memorize the derived regression equation,
The walking support device according to claim 3. The walking parameter value acquisition means acquires a walking parameter for each walking cycle of the target leg when one of the left and right legs is the target leg.
The angular impulse product calculating means calculates an angular impulse product M1 of the target leg for each of the acquired walking parameters of each one walking cycle,
The assist force determining means determines the assist force of the next one walking cycle for the target leg from the angular force product M1 of the target leg calculated for the previous one walking cycle.
The walking support device according to any one of claims 1 to 4, wherein the walking support device is characterized in that: A walking surface determination means for determining the type of walking surface on which the walking support target person walks;
The reference angular impulse product storage means stores a reference angular impulse product z for each type of walking surface,
The angular impulse product calculating means calculates an angular impulse product M1 of one walking cycle required when walking with the value of the walking parameter from the acquired walking parameter value and the determined type of walking surface,
The assist force determining means stores, for the calculated angular force product M1, an angular force product of one walking cycle borne by the walking support target person corresponding to the determined type of walking surface stored. Determine the assist force so that z becomes
The walking support device according to any one of claims 1 to 5, wherein A walking parameter value acquisition function for acquiring a walking parameter value for one walking cycle that defines a walking state in a walking support target during walking;
An angular impulse product calculation function for calculating an angular impulse product M1 of one walking cycle required when walking with the value of the acquired walking parameter from the acquired walking parameter value;
For the calculated angular impulse product M1, the angular impulse product of one walking cycle borne by the walking support target person is a share of the walking support target person among the angular impulse products necessary for walking of one walking cycle. An assist force determining function for determining an assist force so as to be a reference angular force product z;
By applying the determined assist force to the legs of the walking support target person, the walking moment of the walking support target person is reduced by reducing a joint moment generated at the joint of the leg part by the walking support target person. Walking support function,
A walking support program that realizes a computer.

Images