JP2013094502A - Assist robot controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To apply a required amount of assist torque for a part requiring the torque, in rehabilitation or muscle boosting use.SOLUTION: An assist robot controller 120 includes: an estimator gain computing unit 121 configured to estimate the estimator gain based on status detection values; a status estimation error computing unit 122 configured to calculate the status estimation error based on the status detection values and estimator gain; a user torque estimator 123 configured to calculate the user torque estimation value based on the status estimation errors; and an assist torque computing unit 124 configured to calculate assist torque required for operation assistance or rehabilitation treatment based on the user torque estimation value and to output a motor current so as to generate assist torque by means of a joint motor.

Description

  The present invention relates to an assist robot controller that assists the operation of a user who wears it or performs rehabilitation treatment.

  The assist robot targeted by the present invention refers to a robot that includes a mechanism composed of a plurality of links and joints and a plurality of motors that drive the joints, and assists the operation of the user who wears it or performs rehabilitation treatment. .

  FIG. 4 shows an assist robot provided with an associated assist robot controller. The assist robot includes an operation assisting device 410, a drive source 411, a joint angle detection unit 412, a biological signal detection unit 413, a relative force detection unit 414, a control device 415, and a power amplification unit 416. Note that description will be made assuming that the wearer 420 uses the assist robot.

  The motion assisting device 410 is the assist robot described above, and inputs the joint angle, myoelectric potential, and relative force from the wearer 420 and outputs assist force to the wearer 420.

  The drive source 411 generates an assist force based on the drive current from the power amplifying unit 416 and outputs it to the wearer 420, and outputs a drive current detection value to the control device 415.

  The joint angle detection means 412 detects the joint angle and outputs it to the control device 415 as a joint angle detection value.

  The biological signal detection unit 413 detects a myoelectric potential corresponding to the muscular force generated in the wearer 420 and outputs it to the control device 415 as a myoelectric potential detection value.

  The relative force detection means 414 detects the relative force acting on the motion assisting device 410 and outputs it to the control device 415 as a relative force detection value.

  The control device 415 includes a drive current detection value from the drive source 411, a joint angle detection value from the joint angle detection unit 412, a myoelectric potential detection value from the biological signal detection unit 413, and a relative force detection unit 414. Based on the force detection value, a dynamic parameter of the wearer 420 is identified, and a control signal is output using an equation of motion substituted with the dynamic parameter.

  The power amplifying unit 416 amplifies the control signal from the control device 415 and outputs a drive current.

  The related assist robot control device assists the operation in consideration of the dynamic parameters of the wearer 420 by the above-described mechanism (for example, Patent Document 1).

JP 2006-204426 A (page 25, FIG. 1)

  When the related assist robot controller is used for rehabilitation treatment that recovers the power of the extremities lost due to injury or illness, there is no mechanism to consider the degree of recovery of torque that can generate a specific joint of the user, so it is suitable for the degree of recovery In some cases, sufficient rehabilitation effects could not be obtained. Further, in general applications including rehabilitation treatment, information of a plurality of detectors such as myoelectric potential is required, and thus there is a problem that an inexpensive, small, and lightweight assist robot cannot be realized.

  The present invention has been made in view of such problems, and the torque generated around each joint by a user wearing an assist robot can be accurately obtained even when only an inexpensive joint angle detector is provided. It is an object of the present invention to provide an assist robot control device that can be estimated and can apply an assist torque as much as necessary to a necessary part without waste in a user's rehabilitation or force amplification application.

An assist robot control device according to the present invention includes a mechanism composed of a plurality of links and joints, and a plurality of motors that drive the joints, and assists the operation of a user who wears the assist robot or performs rehabilitation treatment. , An assist robot control device for controlling based on a state detection value such as a joint angle, an estimator gain calculator for calculating an estimator gain based on the state detection value, the state detection value and the estimator A state estimation error calculator that calculates the state estimation error based on the gain, and a user torque estimation value that is a torque generated by each joint of the user wearing the assist robot based on the state estimation error A user torque estimator and an assist torque necessary for operation assistance or rehabilitation treatment based on the user torque estimate, Serial joint motors and a assist torque calculation unit that outputs a motor current so as to generate the assist torque.
As a result, the torque generated around each joint by the user wearing the assist robot can be accurately estimated even when only the inexpensive joint angle detector is provided.

  In a rehabilitation or force amplification application, as much assist torque as necessary can be applied to a necessary part without waste.

1 is a functional block diagram of an assist robot control device according to a first embodiment; FIG. 3 is a partial view of the assist robot mechanism according to the first exemplary embodiment. It is a figure of the time change of the knee torque in the simulation concerning Embodiment 2. FIG. It is a figure of the assist robot provided with the related assist robot control apparatus.

Embodiment 1
Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a functional block diagram of the assist robot controller. In the present embodiment, first, an outline of a functional configuration of the assist robot control apparatus of the present invention will be described with reference to FIG. In FIG. 1, arrows indicate signal flows, thick solid lines indicate mechanical connections, and dotted lines indicate a group of strongly related functional blocks.

The assist robot 110 includes a joint motor 111, an assist robot mechanism 112, and a state detector 113. The assist robot 110 is a robot that assists the operation of the user 130 when worn by the user 130.
The assist robot control device 120 includes an estimator gain calculator 121, a state estimation error calculator 122, a user torque estimator 123, and an assist torque calculator 124. The assist robot control device 120 is an electronic control system that outputs a motor current to the assist robot 110 based on a state detection value from the assist robot 110.

  The joint motor 111 drives the joint to an assist robot mechanism 112 described later in accordance with a motor current from an assist robot control device 120 described later.

  The assist robot mechanism 112 is composed of a plurality of links connected via a rotatable joint, and has a shape that can be worn by the user 130. The assist robot mechanism 112 is driven by the joint motor 111.

  The state detector 113 detects a state quantity such as a joint angle and outputs it as a state detection value.

  Based on the state detection value from the assist robot 110, the estimator gain calculator 121 calculates and outputs an estimator gain that is a parameter used for calculation of a state estimation error calculator 122 described later.

  The state estimation error calculator 122 is based on the state detection value from the assist robot 110 and the estimator gain from the estimator gain calculator 121, and is a state estimation value that is an estimated value of the state detection value and its time differential value. Is calculated and the state estimation error is calculated and output.

  The user torque estimator 123 estimates the user torque generated by the user 130 based on the state estimation error from the state estimation error calculator 122 and outputs it as an estimated value of the user torque. The assist torque calculator calculates a necessary assist torque according to the use of the assist robot 110 based on the user torque estimated value from the user torque estimator 123, and the assist robot 110 generates the necessary assist torque. Such a motor current is output.

  The user 130 is a vertebrate including humans, dogs, cats, horses, etc., and needs to be amplified by aging or the like, or needs rehabilitation to recover from an injury or the like.

  Here, the assist robot controller 120 can be realized as an ASIC, a programmable electronic system, or the like.

  Hereinafter, the operation principle of each block in FIG. 1 will be derived, and the details of the function will be described.

  The equation of motion of the assist robot mechanism 112 is derived using FIG. FIG. 2 is a partial view of the assist robot mechanism showing the first embodiment. In FIG. 2, a thick straight line represents a link, and a thick circle represents a joint.

  In FIG. 2, the assist robot mechanism 112 includes a k−1 link 201, a k−1th joint 202, a kth link 203, and a kth joint 204. The joint angle, the assist torque, and the user torque are each positive in the counterclockwise direction.

  The equation of motion of the k-th link 203 is derived as equation (1).

However, the meaning of the symbol in Formula (1) is as follows. J _i is the _i- th moment of inertia [kg · m ² ] around the k-th joint 204, D _k is the k-th viscous friction coefficient [N · m · s / rad], and T _ck is the Coulomb friction of the k-th joint 204. The size [N · m], θ _k is the kth joint angle [rad], _Tak is the kth assist torque (assist torque of the kth joint 204) [N · m], and _Tui is the i th user torque. (Torque applied from the user 130 to the i-th link) [N · m].

In equation (1), the i-th link inertia moment J _i around the k-th joint 204 is known from the design value and the joint angle. It is assumed that the viscous friction coefficient D _k of the k-th joint 204 and the Coulomb friction magnitude T _ck of the k-th joint 204 are already known by measurement before shipment.

  The state variable shown in equation (2) is introduced.

  Using Expression (2), Expression (1) can be rewritten as Expression (3) and Expression (4).

In order to estimate the k-th user torque T _uk , assuming that the k-th joint angle x _1k is detected by the state detector 113 and the k-th joint angular velocity x _2k is obtained as a time differential value of the k-th joint angle x _1k . The state estimators of equations (5) and (6) are introduced.

However, the meaning of the symbol in Formula (6) is as follows. L _k is the estimator gain.

  Using Equations (3) to (6), the state estimation error equation can be derived as Equation (7) and Equation (8).

は、第ｋ関節角度推定誤差[ｒａｄ]、

は、第ｋ関節角速度推定誤差[ｒａｄ／ｓ]。 However, the meanings of the symbols in the formulas (7) and (8) are as follows.

Is the kth joint angle estimation error [rad],

Is the k-th joint angular velocity estimation error [rad / s].

  The equilibrium points of Equation (7) and Equation (8) are obtained as Equation (9) and Equation (10) by setting the first-order time derivative of the state estimation error to 0.

  The Jacobian of the system matrix of the system represented by Equation (7) and Equation (8) can be derived as Equation (11).

The eigenvalues at the equilibrium points represented by the expressions (9) and (10) of the Jacobian matrix J _k are expressed as expressions (12) and (13).

Since the k-th estimator gain L _k , the i-th link moment of inertia J _i around the k-th joint 204, and the k-th viscous friction coefficient D _k are positive numbers, eigenvalues s _{k +} shown in Expression (12) and Expression (13) And s _k− can be classified as follows.

Case 1: If the value within the square root of Equation (12) and Equation (13) is 0 or more, s _{k +} and s _k− are negative integers.
Case 2: If the value within the square root of Equation (12) and Equation (13) is less than 0, s _{k +} and s _k− are complex conjugates having a negative real part.

  As described above, since the eigenvalues (12) and (13) have a negative real part, the estimation error converges to the equilibrium point expressed by the equations (9) and (10).

  In case 1 above, the estimation error monotonously converges to the equilibrium point, and in case 2 it oscillates to the equilibrium point. In the use of the assist robot 110 in which monotonous convergence is desirable, the estimator gain calculator 121 calculates the estimator gain so that the condition 1 is satisfied. In applications where vibration convergence is desirable, the estimator gain calculator 121 calculates the estimator gain so as to satisfy the condition of Case 2.

  Solving equation (9) for user torque yields equation (14).

  Equation (15) is obtained from Equation (14).

は、第ｋ使用者トルク推定値（第ｋ使用者トルクの推定値）[Ｎ・ｍ]。 However, the meaning of the symbol in Formula (15) is as follows.

Is the k-th user torque estimated value (the estimated value of the k-th user torque) [N · m].

を演算する。 The user torque estimator 123 uses the equation (15) to estimate the k-th user torque.

Is calculated.

を精度よく得ることができる。 Since equation (15) does not depend on the absolute value of the k-th joint angular velocity, an inexpensive angle detector (for example, a potentiometer) is used as the state detector 113, and the joint angular velocity is not directly detected, but is calculated from the joint angle. The kth user torque estimate

Can be obtained with high accuracy.

  Hereinafter, a configuration example of the assist torque calculator 124 will be shown for each use of the assist robot 110.

When the assist robot 110 is used for rehabilitation treatment that recovers the power of the limbs lost due to injury or illness, the k-th assist torque T _ak shown in Expression (16) can be applied to the k-th joint 204. .

However, the meanings of the symbols in the formula (16) are as follows. T _nk is the torque [N · m] that the k-th joint of the user 130 can generate before injury.

  According to the equation (16), as the recovery of the k-th joint of the user 130 progresses, the ratio of assistance by the assist robot 110 decreases, and the k-th assist torque becomes 0 [N · m] at the time of complete recovery. .

Further, when performing repetitive motions such as walking in the rehabilitation treatment, the kth assist torque _Tak shown in the equation (16) is applied in one cycle, and the kth shown in the equation (17) in the second cycle and thereafter. By applying the assist torque _Tak , the k-th user torque _Tuk becomes a smooth function, so that rehabilitation treatment with less burden on the kth joint can be performed.

  However, the meaning of the symbol in Formula (17) is as follows. T is one cycle [s] of the rehabilitation operation.

When the assist robot 110 is used for the purpose of amplifying the force of the user 130, the kth assist torque _Tak shown in the equation (18) can be applied to the kth joint 204.

However, the meanings of the symbols in the formula (18) are as follows. n _k is the amplification factor of the k-th user torque.

According to Expression (18), it is possible to generate a torque that is _nk times the _k- th user torque T _uk actually generated by the user 130 at the k-th joint. The amplification factor n _k of the _k- th user torque can be arbitrarily set by the user 130 according to how much extra force is required for which part of the user 130. For example, when the muscle that bends and stretches the knee is weakened by aging, _nk may be set to further assist the bending and stretching operation of the knee.

を演算する。アシストトルク演算器１２４は、アシストロボット１１０の用途に応じて、式（１６）、または式（１７）、または式（１８）に基づいて第ｋアシストトルクＴ_ａｋを演算し、関節モータ１１１に第ｋアシストトルクＴ_ａｋを発生させるようなモータ電流を出力する。 As described above, in the assist robot controller 120, the estimator gain calculator 121 determines whether the state estimation error is desirably monotonously converged to the equilibrium point or is desirably converged to the equilibrium point in terms of vibration. Thus, the k-th estimator gain L _k is calculated so as to satisfy the above-described case 1 or 2 condition. The state estimation error calculator 122 calculates a state estimation error based on the estimation error equations of Expressions (7) and (8). The user torque estimator 123 calculates the k-th user torque estimation value based on the equation (15).

Is calculated. The assist torque calculator 124 calculates the k-th assist torque T _ak based on the equation (16), the equation (17), or the equation (18) according to the use of the assist robot 110, A motor current that generates the k assist torque _Tak is output.

  Thus, according to the present invention, the torque generated around each joint by the user wearing the assist robot can be accurately estimated even when only the inexpensive joint angle detector is provided, and the user's rehabilitation or force can be estimated. It is possible to give as much assist torque as necessary to a necessary portion without waste in amplification applications.

Embodiment 2
In this embodiment, a simulation result of the present invention is shown. The numerical values used for the simulation are as follows.
l ₁ = 0.24 [m], l ₂ = 0.38 [m], l ₃ = 0.46 [m], J ₁ = 3.0 × 10 ⁻³ [kg · m ² ], J ₂ = 3.9 × 10 ⁻² [kg · m ² ], J ₃ = 1.2 × 10 ⁻¹ [kg · m ² ], T _2min = −7 [N · m], T _2max = 7 [N · m ].

However, the meanings of the above symbols are as follows. l ₁ is the foot length [m], l ₂ is the shin length [m], l ₃ is the thigh length [m], J ₁ is the moment of inertia around the center of gravity of the foot [kg · m ² ], and J ₂ is the shin Moment of inertia around the center of gravity [kg · m ² ], J ₃ is the moment of inertia around the center of gravity of the thigh [kg · m ² ], and T _2min is the limit value of the knee torque that the user can generate in the direction of bending the knee during rehabilitation [N · m], T _2max is a limit value [N · m] of knee torque that can be generated in the direction in which the user stretches the knee during rehabilitation.

  As the value used for the simulation, the average value of human body measurement values described in DOTHS-801 430, “Investigation Of Inertial Properties Of The Human Body”, U.S. Department Of Transportation, National Highway Traffic Safety Administration, March 1975 was used. In this simulation, rehabilitation treatment that restores knee function through gait training is simulated.

  FIG. 3 shows changes in knee torque over time in the simulation showing the second embodiment. In FIG. 3, the broken line indicates the knee torque generated when the user walks before the injury (hereinafter referred to as “desired knee torque”), and the solid line indicates the estimated value of the knee torque that can be generated by the user during the rehabilitation treatment (hereinafter referred to as “the desired knee torque”). , Referred to as “user torque estimated value”), the alternate long and short dash line is the assist torque. Here, the desired knee torque is based on walking actual measurement data by Dr. Ben Stansfield, University of Strathclyde.

  The user torque estimated value matches the actual user torque, and it can be seen that the user torque estimation using the equation (15) in the user torque estimator 123 is performed with high accuracy. Further, since the added value of the assist torque calculated by the assist torque calculator 124 using the equation (16) and the estimated value of the user torque is a desired knee torque, it is necessary depending on the degree of knee recovery. It can be seen that walking training can be assisted by various assist torques.

  Thus, according to the present invention, the assist torque necessary for the rehabilitation treatment can be applied according to the degree of knee recovery of the user 130.

  Note that the present invention is not limited to the above-described embodiment, and can be changed as appropriate without departing from the spirit of the present invention.

  For example, in the above embodiment, an example of rehabilitation treatment of a user's knee that has become difficult to walk has been shown, but by replacing the knee with another part of the user and walking with another motion It can also be applied to general rehabilitation treatment for recovering the motor function of the user.

  Therefore, according to the present invention, when an increase in force is required in activities of daily life due to aging or the like, and when an increase in force for care or rescue is necessary, an assist provided only with an inexpensive joint angle detector The robot can give as much assist torque as necessary to the necessary part.

  More specifically, when used for rehabilitation treatment for recovering the power of the limbs lost due to injury or illness, the assist torque necessary for recovery of each part can be applied without waste, and the rehabilitation treatment can be effectively performed.

  Moreover, in the rehabilitation treatment for recovering the power of the extremities lost due to injury or illness, when repetitive work such as walking is performed, the rehabilitation treatment with less burden on the joint to be treated can be performed.

  Further, in the application of the user's force amplification, it is possible to give as much assist torque as necessary to a necessary part without waste.

  According to the present invention, a user wearing an assist robot can accurately estimate the torque generated around each joint only with an inexpensive joint angle detector, and the user can be rehabilitated. It can be widely applied to all devices used for medical, nursing, and work that requires power amplification, such as rescue robot suits, walking assist robots, and nursing devices.

DESCRIPTION OF SYMBOLS 110 Assist robot 111 Joint motor 112 Assist robot mechanism 113 State detector 120 Assist robot controller 121 Estimator gain calculator 122 State estimation error calculator 123 User torque estimator 124 Assist torque calculator 130 User 201 Link 202 Joint 203 Link 204 Joint 410 Motion assist device 411 Drive source 412 Joint angle detection means 413 Biosignal detection means 414 Relative force detection means 415 Control device 416 Power amplification means 420 Wearer

Claims (1)

Based on a state detection value such as a joint angle, an operation of an assist robot that includes a mechanism composed of a plurality of links and joints and a plurality of motors that drive the joints, and assists or rehabilitates the operation of the user who wears them. An assist robot control device for controlling
An estimator gain calculator that calculates an estimator gain based on the state detection value;
A state estimation error calculator that calculates the state estimation error based on the state detection value and the estimator gain;
A user torque estimator that calculates a user torque estimation value that is a torque generated by each joint of the user wearing the assist robot based on the state estimation error;
An assist torque calculator that calculates an assist torque necessary for operation assistance or rehabilitation treatment based on the user torque estimation value, and outputs a motor current such that the joint motor generates the assist torque, and
Assist robot controller.

Images