Patent Application
20100234777
The present application is based on and claims the priority benefit of Japanese Patent Application 2007-268304 filed on Oct. 15, 2007, the contents of which are incorporated herein by reference in its entirety.
1. Field of the Invention
The present invention relates to a motion assist device for assisting a periodical motion of a creature by applying a force to the creature.
2. Description of the Related Art
There has been proposed a device for assisting a periodical walking motion of a creature, such as a human whose body part such as a lower limb or the like is suffering from physical hypofunction, by applying a periodically varying force to the creature (refer to Japanese Patent Laid-open No. 2004-073649). There has also been proposed a device for assisting or guiding a periodical motion of a human by adjusting strength of a force applied to the human according to a model (a spring model) representing a behavior feature of a virtual elastic element so as to match a motion scale of the human with a desired motion scale thereof (refer to Japanese Patent Laid-open No. 2007-061217).
According to the spring model, the strength of force applied to the human is adjusted according to a motion rhythm of the periodical motion of the human. Therefore, when an action of the human is performed fast to some extent, the periodical motion of the human can be assisted with a sufficient force from the viewpoint of matching the motion scale of the human to the desired motion scale thereof. However, on the other hand, when the human acts slowly due to remarkable physical hypofunction or the like, it is possible that the strength of force applied to the human by the motion guiding device becomes insufficient from the viewpoint of matching the motion scale of the human to the desired motion scale thereof.
The present invention has been accomplished in view of the aforementioned problems, and it is therefore an object of the present invention to provide a motion assist device for assisting a periodical motion of a creature such as a human by applying a force with appropriate strength to the creature so as to match a motion scale of the creature with a desired motion scale thereof regardless of a motion rhythm of the periodical motion.
A first aspect of a motion assist device is provided with an orthosis mounted on a creature, an actuator connected to the orthosis and a controller configured to control a magnitude and a phase of an output from the actuator, and assists a periodical motion of the creature by applying a periodically varying output from the actuator to the creature through the orthosis, wherein the controller includes a motion variable determination element configured to determine a value of a motion variable serving as a function of the magnitude of the periodical motion of the creature; a motion oscillator determination element configured to determine a first and a second motion oscillators which vary periodically according to the periodical motion of the creature, respectively; a first oscillator generation element configured to generate a first oscillator as an output oscillation signal from a first model, which generates the output oscillation signal varying at a specific angular velocity defined on the basis of a first intrinsic angular velocity by entraining to an input oscillation signal, by inputting the first motion oscillator determined by the motion oscillator determination element as the input oscillation signal to the first model; an intrinsic angular velocity setting element configured to set an angular velocity of a second virtual oscillator as a second intrinsic angular velocity on the basis of a virtual model denoting a first virtual oscillator and a second virtual oscillator which interact with each other and vary periodically with a second phase difference and a first phase difference between the first motion oscillator determined by the motion oscillator determination element and the first oscillator generated by the first oscillator generation element so as to approximate the second phase difference to a desired phase difference; and a second oscillator generation element configured to generate a second oscillator as an output oscillation signal serving as a control basis for the force applied to the creature from a second model, which generates the output oscillation signal varying at a specific angular velocity defined on the basis of the second intrinsic angular velocity set by the intrinsic angular velocity setting element according to an input oscillation signal, by inputting the second motion oscillator determined by the motion oscillator determination element as the input oscillation signal to the second model, and configured to correct the second model so as to approximate the value of the motion variable determined by the motion variable determination element to a desired value.
According to the motion assist device of the first aspect of the present invention, the second model is corrected so as to approximate the value of the motion variable serving as the function of the magnitude (motion scale) of the periodical motion of the creature. The second oscillator is generated according to the corrected second model, and the periodically varying force applied to the creature is controlled according to the second oscillator. According thereto, the periodical motion of the creature can be assisted by applying a force with appropriate strength to the creature so as to match the motion scale of the creature with the desired motion scale thereof regardless of the motion rhythm of the periodical motion.
A second aspect of the motion assist device is dependent on the first aspect of the present invention, wherein the second model is defined by a simultaneous differential equation of a plurality of state variables denoting a behavior state of the creature; the simultaneous differential equation contains a product of the desired value for the motion variable and a coefficient; and the second oscillator generation element generates the second oscillator on the basis of values of the state variables obtained by solving the simultaneous differential equation, and corrects the second model by correcting the coefficient so as to approximate the value of the motion variable determined by the motion variable determination element to the desired value.
According to the motion assist device of the second aspect of the present invention, the second model is defined by the simultaneous differential equation of a plurality of state variables denoting the behavior of a body part of the creature. A product of the desired value for the motion variable and a coefficient are contained in the simultaneous differential equation, which is the function of the magnitude of the periodical motion of the creature. The second model is corrected through adjusting the coefficient. On the basis of the values of the plurality of state variables, namely the solutions of the simultaneous differential equation for defining the corrected second model, the second oscillator is generated. On the basis of the second oscillator, the periodically varying force applied to the creature is controlled. According thereto, the periodical motion of the creature can be assisted by applying a force with appropriate strength to the creature so as to match the motion scale of the creature with the desired motion scale thereof regardless of the motion rhythm of the periodical motion of the creature.
A third aspect of the motion assist device is dependent on the first aspect of the present invention, wherein the orthosis includes a first orthosis mounted on a first body part, a second orthosis mounted on a second body part, and a third orthosis mounted on a third body part of the creature; the actuator is connected directly to the first orthosis and the second orthosis, and is connected indirectly to the third orthosis through the second orthosis; and the motion assist device applies the force to the creature through each of the first orthosis, the second orthosis and the third orthosis by controlling the actuator on the basis of the second oscillator generated by the second oscillator generation element.
According to the motion assist device of the third aspect of the present invention, the second orthosis is actuated with respect to the first orthosis and the third orthosis is actuated to following the second orthosis according to the operations of the actuator. According thereto, in addition to the periodical motion of the second body part with respect to the first body part, the motion of the third body part can be assisted to follow the motion of the second body part.
a) and
An embodiment regarding a motion assist device of the present invention will be described with reference to the drawings. Hereinafter, numerals “L” and “R” are used to differentiate a left side and a right side of legs or the like. If there is not necessary to differentiate the left side and the right side or a vector has both of the left and right components, the numerals can be omitted. In addition, symbols “+” and “−” are used to differentiate a flexion motion (forward motion) and a stretch motion (backward motion) of a leg (in particular, a thigh).
The motion assist device 10 illustrated in
The first orthosis 1100 is provided with a first supporter 1110 and a first link member 1120. The first supporter 1110 is made from a combination of a rigid material such as a rigid resin and a flexible material such as a fiber. The first supporter 1110 is mounted on a back side of the waist. The first link member 1120 is made of a rigid resin, and is fixed at the first supporter 1110 in such a way that when the first supporter 1110 is attached to the waist, the first link member 1120 is located at both sides of the waist laterally. The second orthosis 1200 includes a second supporter 1210, a second link member 1220 and a rib member 1240. Similar to the first supporter 1110, the second supporter 1210 is also made from a combination of a rigid material and a flexible material and is mounted on a front side and a back side of the thigh, respectively. The second link member 1220 is made of a rigid resin, extending vertically along the thigh and is connected to an output shaft of the actuator 15. The rib member 1240 is made of a rigid resin in a way of extending laterally on the front side and back side, respectively, while bending along the thigh and is connected to the second supporter 1210. The third orthosis 1300 includes a third supporter 1310, a third link member 1320. The third supporter 1310 is formed to have a shape of a slipper or a shoe which is mounted to a foot of the human P. The third link member 1320 is made of rigid resin in a way of extending vertically along the lower leg. An upper end of the third link member 1320 is movably connected to a lower end of the first link member 1120, and a lower end thereof is movably connected to or fixed at the third supporter 1310. Note that the third orthosis 1300 may be omitted.
The hip joint angle sensor 11 includes a rotary encoder disposed on a transverse side of the waist of the human P and outputs a signal according to the hip joint angle. The actuator 15 is composed of a motor, including either one or both of a reduction gear and a compliance mechanism where appropriate. The battery 1000 is housed in the first orthosis 1100 (for example, fixed in multiple sheets of cloth constituting the first supporter 1110), which supplies electrical power to the actuator 15, the controller 100 and the like. Note that it is acceptable to attach or house the respective of the controller 100 and the battery 1000 in the second orthosis 1200 and the third orthosis 1300; it is also acceptable to dispose them separately from the motion assist device 10.
The controller 100 includes a computer housed in the first orthosis 1100 and software stored in a memory or a storing device in the computer. The controller 100 controls an operation or an output torque T of the actuator 15 by adjusting an electrical power supplied from the battery 1000 to the actuator 15. The controller 100 as illustrated in
The motion variable determination element 102 determines a motion variable ζ which represents a motion scale of a periodical motion of the human P. The motion oscillator determination element 110 determines an angular velocity of each hip joint as a first motion oscillator φ1 and an angle thereof as a second motion oscillator φ2 on the basis of an output from the hip joint angle sensor 11. The first motion oscillator φ1 and the second motion oscillator φ2 vary periodically according to the periodical motion of the human P, whose variation patterns are defined according to a magnitude and phase (or an angular velocity which is a first order differentiation of phase by time). Determination of an oscillator refers to the determination of a periodical variation pattern of the oscillator. The first oscillator generation element 120 generates a first oscillator ν1 as an output oscillation signal by inputting the first motion oscillator φ1 determined by the motion oscillator determination element 110 to a first model as an input oscillator signal. The generation of an oscillator refers to the definition of the periodical variation pattern of the oscillator. The “first model” is a model which generates the output oscillation signal varying at a specific angular velocity defined according to a first intrinsic angular velocity ω1 by entraining to the input oscillation signal (synchronization phenomenon).
The intrinsic angular velocity setting element 130 includes a first phase difference setting element 131, a second phase difference setting element 132, a correlation coefficient setting element 133, a first angular velocity setting element 134, and a second angular velocity setting element 135. The intrinsic angular velocity setting element 130 sets a second intrinsic angular velocity ω2 on the basis of a first phase difference δθ1 and a virtual model so as to approximate a second phase difference δθ2 to a desired phase difference δθ0. The first phase difference δθ1 is the phase difference between the first motion oscillator φ1 determined by the motion oscillator determination element 110 and the first oscillator ξ1 generated by the first oscillator generation element 120. The virtual model is a model which represents the periodical motion of the human P as a periodical variation of a first virtual oscillator φ1, the periodical operation of the motion assist device 10 as a periodical variation of a second virtual oscillator φ2, and the phase difference between the periodical motion of the human P and the periodical operation of the motion assist device 10 as a phase difference between the first virtual oscillator φ1 and the second virtual oscillator φ2, respectively.
The second oscillator generation element 140 generates a second oscillator ξ2 as an output oscillation signal from a second model by inputting the second motion oscillator φ2 determined by the motion oscillator determination element 110 to the second model as an input oscillator signal. The “second model” is a model which generates the output oscillation signal varying at a specific angular velocity defined according to the second intrinsic: angular velocity ω2 defined by the intrinsic angular velocity setting element 130 on the basis of the input oscillation signal.
The auxiliary oscillator generation element 150, on the basis of the second oscillator ξ2 generated by the second oscillator generation element 140, generates an auxiliary oscillator η for defining a variation pattern of a torque applied to the thigh by the actuator 15 of the motion assist device 10.
Hereinafter, the assisting method for the walking motion of the human P according to the motion assist device 10 with the aforementioned configurations will be explained.
The motion variable determination element 102, on the basis of the output signal from the hip joint angle sensor 11, determines the left hip joint angle and the right hip joint angle at respective finished time of the flexion motion and the finished time of the stretch motion of the thigh in each walking cycle (FIG. 3/S002), in detail the left hip joint angle at the finished time of the flexion motion, the left hip joint angle at the finished time of the stretch motion, the right hip joint angle at the finished time of the flexion motion and the right hip joint angle at the finished time of the stretch motion of the thigh, as the motion variable ζ={ζi|i=L+, L−, R+, R−}. Note that it is acceptable to determine a footstep of the human P as the motion variable ζ. The footstep, for example, may be determined on the basis of a correlation among the hip joint angle of the human P which is determined according to the output signal from the hip joint angle sensor 11, the hip joint angle of the human P stored in memory, and the foot positions in the anteroposterior direction. Herein, it is acceptable to determine a step rate (numbers of steps per unit time) and a walking speed of the human P, and thereafter determine the footstep of the human P on the basis of the step rate and the walking speed. The step rate may be determined on the basis of an output signal from an acceleration sensor attached to the human P, which outputs the output signal according to an acceleration of the human P in the vertical direction. The walking speed may be determined on the basis of an output signal from a speed sensor in a treadmill 30 for outputting the output signal according to the speed of an endless belt 32. Moreover, it is also acceptable to determine the value of a function, namely the motion variable having multiple variables containing at least one of the walking rate (=footstep/step rate), the footstep, the left hip joint angle at the finished time of the flexion motion and the left hip joint angle at the finished time of the stretch motion, the right hip joint angle at the finished time of the flexion motion and the right hip joint angle at the finished time of the stretch motion of the thigh in each walking cycle.
Further, the motion oscillator determination element 110 determines the angular velocity of each of the left and right hip joints of the human P as the first motion oscillator φ1=(φ1L, φ1R) on the basis of the output from the hip joint angle sensor 11 (FIG. 3/S011). Furthermore, the motion oscillator determination element 110 determines the left hip joint angle and the right joint angle of the human P as the second motion oscillator φ2=(φ2L, φ2R) on the basis of the output from the hip joint angle sensor 11 (FIG. 3/S012).
Note that it is acceptable to determine an arbitrary variable varying periodically according to the periodical motion of the human P by using an appropriate sensor as the first motion oscillator φ1 and the second motion oscillator φ2, respectively. For example, the angle or angular velocity of an arbitrary joint, such as the hip joint, knee joint, foot joint, shoulder joint, elbow joint and the like, and the position of the thigh, foot, upper arm, hand and waist (the position or the like in the anteroposterior direction or the vertical direction with reference to the center-of-gravity of the human P), and the variation patterns of velocity and acceleration may be determined as the motion oscillator. The variation patterns of various parameters varying at a rhythm in conjunction with the walking motion rhythm, such as sounds generated when the left or right foot steps on ground, breathing sounds, deliberate phonations or the like, may be determined as one or both of the first motion oscillator φ1 and the second motion oscillator φ2. Moreover, it is acceptable to determine variables representing the periodical motion state of an identical body part, such as the angle and the angular velocity or the like of an identical joint, as each of the first motion oscillator φ1 and the second motion oscillator φ2; it is also acceptable to determine variables representing the periodical motion state of different body parts, such as the respective the angular velocities or the like of different joints, as each of the first motion-oscillator φ1 and the second motion oscillator φ2.
Thereafter, the first oscillator generation element 120 generates the first oscillator ξ1 as the output oscillation signal by inputting the first motion oscillator φ1 determined by the motion oscillator determination element 110 to the first model as the input oscillation signal (FIG. 3/S011). As described above, the first model denotes the correlation between a plurality of the first elements such as the left and right feet or the like, and generates the output oscillation signal which varies at the angular velocity defined according to the first intrinsic angular velocity ω1=(ω1L, ω1R) by entraining to the input oscillation signal. The first model, for example, may be defined by the Van der Pol equation expressed by the equation (10). Moreover, it is possible that the first oscillator generation element 120 sequentially updates the first model by adopting the latest second intrinsic angular velocity ω2 set by the intrinsic angular velocity setting element 130 as the latest first intrinsic angular velocity ω1, and generates a subsequent first oscillator as the output oscillation signal by inputting a subsequent first motion oscillator φ1 as the input oscillation signal into the updated first model.
(d2φ1L/dt2)=A(1−ξ1L2)(dξ1L/dt)−ω1L2ξ1L+g(ξ1L−ξ1R)+K1φ1L,
(d2φ1R/dt2)=A(1−ξ1R2)(dξ1R/dt)−ω1R2ξ1R+g(ξ1R−ξ1L)+K1φ1R (10)
Wherein:
A: a positive coefficient set in such a way that a stable limit cycle may be drawn from the first oscillator ξ1 and the first order temporal differentiation value (dξ1/dt) thereof in a plane of “ξ1−(dξ1/dt)”;
g: a first correlation coefficient for reflecting the correlation of different body parts such as the left and right feet of the human P or the like as a correlation (correlation between the output oscillation signals from the plurality of the first elements) of each of the left and right components of the first oscillator ξ1; and
K1: a feedback coefficient related to the first motion oscillator φ1.
The first oscillator ξ1=(ξ1L, ξ1R) is calculated or generated according to the Runge-Kutta method. The respective angular velocity of the components ξ1L, and ξ1R of the first oscillator ξ1 represents a virtual rhythm which assists the motions of the left foot and the right foot, respectively. Further, the first oscillator x has the property to vary or oscillate periodically with an autonomous angular velocity or rhythm defined on the basis of the first intrinsic angular velocity ω1 while harmonizing with the rhythm of the first motion oscillator φ1 varying at an angular velocity or rhythm substantially same as a rhythm of the actual walking motion, according to the “mutual entrainment” (harmonization effect) which is one of the properties of the Van del Pol equation.
In addition, the first model may be expressed by the Van der Pol equation having a form different from that of the equation (10), or by a certain equation which generates the output oscillation signal varying periodically at an angular velocity defined on the basis of the first intrinsic angular velocity ω1, accompanied by the mutual entrainment to the input oscillation signal. Moreover, it is acceptable to increase the numbers of the first motion oscillator φ1, namely the determination object. The more numbers of the first motion oscillator φ1 are input to the first model, the motion will be more elaborately assisted by considering the motions of various body parts of the human P through the adjustment of the correlation coefficients, although the correlation members in a non-linear differentiation equation corresponding to the generation of the first oscillator ξ1 in the Van der Pol equation for defining the first model will become more accordingly.
The phase difference between the periodical motion of the human P and the periodical motion of the motion assist device 10 is used to define the moving behavior of the human P with respect to the motion of the motion assist device 10. For example, if the phase difference is positive, the human P moves in a way of leading the motion assist device 10. On the other hand, if the phase difference is negative, the human P can move in a way of being led by the motion assist device 10. Therefore, if the phase difference (the first phase difference) δθ1 of the first oscillator ξ1 with respect to the first motion oscillator φ1 deviates from the desired phase difference δθ0, it is likely that the moving behavior of the human P will be made unstable. Consequently, there is a high probability that the motion rhythm of the human P whose relative motions between the waist and the thigh assisted by the torque T varying periodically at an angular velocity in relation to the auxiliary oscillator n would deviate from the desired motion rhythm.
Therefore, from the viewpoint of matching the motion rhythm of the human P with the desired motion rhythm while maintaining the mutual harmonization between the first motion oscillator φ1 and the first oscillator ξi, an appropriate second intrinsic angular velocity ω2 for defining the angular velocity of the second oscillator ξ2 is set by the intrinsic angular velocity setting element 130. In other words, an appropriate second intrinsic angular velocity ω2 is set from the viewpoint of maintaining an appropriate phase difference between an assist rhythm of the motion assist device 10 and the motion rhythm of the human P so that the motion rhythm of the human P is in accordance with the assist rhythm of the motion assist device 10 while harmonizing the assist rhythm of the motion assist device 10 with the motion rhythm of the human P.
Specifically, the first phase difference setting element 131 sets a phase difference between the first motion oscillator φ1 and the first oscillator ξ1 as the first phase difference δθ1 (FIG. 3/S031). The first phase difference δθ1 is calculated or set on the basis of a difference of time between, for example, a time where φ1=0 and (dφ1/dt)>0 and a time where ξ1=0 and (dξ1/dt)>0.
Thereafter, the second phase difference setting element 132 sets the second phase difference δθ2 on a condition that the first phase difference δθ1 over the recent three walking cycles is constant or the variation of the first phase difference δθ1 is within an allowable range (FIG. 3/S032). In detail, a phase difference between the first virtual oscillator φ1=(φ1L, φ1R) and the second virtual oscillator (φ2=(φ2L, φ2R) which are defined in the virtual model denoted by the equations (21) and (22) is set as the second phase difference δθ2 according to the equation (23). The first virtual oscillator φ1 in the virtual model virtually represents the first motion oscillator φ1, and the second virtual oscillator φ2 in the virtual model represents the auxiliary oscillator η virtually.
dφ
1L
/dt=ω
1L+εL sin(φ2L−φ1L), dφ1R/dt=ω1R+εR sin(φ2R−φ1R) (21)
dφ
2L
/dt=ω
2L+εL sin(φ1L−φ2L), dφ2R/dt=ω2R+εR sin(φ1R−φ2R) (22)
εθ2L=arcsin {(ω1/L−ω2/L}, δθ2R=arcsin {(ω1/R−ω2/R)/2εR} (23)
Wherein, each component of “ε=(εL, εR)” stands for a correlation coefficient representing the correlation between each component of the first virtual oscillator φ1 and each component of the second virtual oscillator φ2. “ω1/=(ω1/L, ω1/R)” is the angular velocity for each component of the first virtual oscillator φ1, and “ω2/=(ω2/L, ω2/R)” is the angular velocity for each component of the second virtual oscillator φ2.
Subsequently, the correlation coefficient setting element 133 sets the correlation coefficient ε so that the deviation between the first phase difference δθ1 set by the first phase difference setting element 131 and the second phase difference δθ2 set by the second phase difference setting element 132 will be minimum (FIG. 3/S033).
Specifically, the correlation coefficient ε(ti) at each time tk where the first motion oscillator φ1 for each of the left and right components will be zero is sequentially set according to the equation (24).
εL(tk+1)=εL(tk)−BL{V1L(tk+1)−V1L(tk)}/{εL(tk)−εL(tk−1)},
εR(tk+1)=εR(tk)−BR{V1R(tk+1)−V1R(tk)}/{εR(tk)−εR(tk−1)},
V
1L(tk+1)≡(½){δθ1L(tk+1)−δθ2L(tk)}2,
V
1R(tk+1)≡(½){δθ1R(tk+1)−δθ2R(tk)}2 (24)
Wherein, each component of “B=(BL, BR)” stands for a coefficient representing the stability of a potential V1=(V1L, V1R) for approximating each component of the first phase difference δθ1 to each of the left and right components of the second phase difference δθ2, respectively.
Next, the first angular velocity setting element 134 sets the angular velocity of the first virtual oscillator φ1 as the first angular velocity ω1/ according to the correlation coefficient ε set by the correlation coefficient setting element 133 and the equation (25) so that the deviation between the first phase difference 601 and the second phase difference δθ2 for each component will become minimum on a condition that the angular velocity ω2/ of the second virtual oscillator φ2 is constant (FIG. 3/S034).
ω1/L(tk)=−αL∫dtq1L(t), ω1/R(tk)=−αR∫dtq1R(t)
q
1L(t)=(4εL2(tk)−(ω1/L(t)−ω2/L(tk)))1/2×sin(arcsin [(ω1/L(t)−ω2/L(tk−1))/2εL(tk)]−δθ2L(tk)),
q
1R(t)=(4εR2(tk)−(ω1/R(t)−ω2/R(tk)))1/2×sin(arcsin [(ω1/R(t)−ω2/R(tk−1))/2εR(tk)]−δθ2R(tk)) (25)
Wherein, each component of “α=(αL, αR)” stands for the coefficient representing the stability of the system.
The virtual model is constructed on a condition that the mutual harmonization between the first motion oscillator φ1 and the first oscillator ξ1 is also maintained between the first virtual oscillator φ1 and the second virtual oscillator φ2 through setting the correlation coefficient ε and the angular velocity ω1/. In other words, the virtual model is constructed so that the first virtual oscillator φ1 representing the periodical motion of the human P and the second virtual oscillator φ2 representing the periodical motion of the motion assist device 10 vary periodically with the second phase difference δθ2 while harmonizing with each other.
Thereafter, the second angular velocity setting element 135 sets the angular velocity of the second virtual oscillator φ2 as the second angular velocity ω2/ for each component on the basis of the first angular velocity ω1/ set by the first angular velocity setting element 134 (FIG. 3/S035). The second angular velocity ω2/=(ω2/L, ω2/R) is set according to the equation (26) so that the second phase difference δθ2 for each of the left and right components approximates to the desired phase difference δθ0.
Subsequently, the second angular velocity ω2/ is set as the second intrinsic angular velocity ω2 (FIG. 3/S036).
ω2/L(tk)=βL∫dtq2L(t), ω2/R(tk)=βR∫dtq2R(t)
q
2L(t)=(4εL2(tk)−(ω1/L(t)−ω2/L(tk)))1/2×sin(arcsin [ω1/L(tk)−ω2/L(t))/2εL(tk)]−δθ0),
q
2R(t)=(4εR2(tk)−(ω1/R(t)−ω2/R(tk)))1/2×sin(arcsin [ω1/R(tk)−ω2/R(t))/2εR(tk)]−δθ0) (26)
Wherein, each component of “β=(βL, βR)” stands for the coefficient representing the stability of the system.
Accordingly, the second angular velocity ω2/ is appropriately set from the viewpoint of approximating the phase difference between the periodical motion of the human P represented by the first virtual oscillator φ1 and the periodical motion of the motion assist device 10 represented by the second virtual oscillator φ2 to the desired phase difference δθ0, while the mutual harmonization between the first motion oscillator φ1 and the first oscillator ξ1 is maintained between the periodical motion of the human P and the periodical motion of the motion assist device 10.
The second oscillator generation element 140 corrects the second model by appropriately correcting a coefficient c contained in the simultaneous differentiation equation which represents the second model according to the equation (28) (FIG. 3/S038). “c={ci|i=L+, L−, R+, R−}” is a coefficient to be adjusted so that the motion variable determined by the motion variable determination element 102 will approximate to a desired value ζ0 or a deviation therebetween will become minimum.
c
i(tk+1)=ci(tk)−Ci{Vi(tk+1)−Vi(tk)}/{ci(tk−1)},
V
i(tk+1)≡(½){ζi(tk+1)−ζi(tk)}2 (28)
Each component of “C=CL+, CL−, CR+, CR−}” stands for the coefficient representing the stability of a potential V2=(V2L+, V2L−, V2R+, V2R−) for approximating each component of the determination values of the motion variable to each component of the desired value ζ0 thereof. “ζ0={ζ0i|i=L+, L−, R+, R−}” stands for the desired value for each of the left hip joint angle and the right hip joint angle at the finished time of the flexion motion and the stretch motion of the thigh every walking cycle, respectively. The desired value ζ0 may be calculated on the basis of the desired footstep of the human P which is stored preliminarily in the memory and the correlation stored preliminarily in the memory among the left hip joint angle and the right hip joint angle at the finished time of the flexion motion and the stretch motion of the thigh every walking cycle, respectively, and the footstep.
Thereafter, the second oscillator generation element 140 generates the second oscillator ξ2=(ξ2L+, ξ2L−, ξ2R+, ξ2R−) as an output oscillation signal from the second model by inputting the second motion oscillator φ2 determined by the motion oscillator determination element 110 to the second model as an input oscillation signal (FIG. 3/S040). The second model is a model representing the correlation between a plurality of second elements including the neural elements or the like responsible for the motions to the flexion direction (forward direction) and the stretch direction (backward direction) of each leg. As aforementioned, the second model generates the output oscillation signal varying at an angular velocity defined according to the second intrinsic angular velocity w2 set by the intrinsic angular velocity setting element 130 on the basis of the input oscillation signal.
The second model is defined by a simultaneous differentiation equation represented by, for example, the equation (30). The simultaneous differentiation equation contains therein a state variable u={ui|i=L+, L−, R+, R−} representing the behavior state (specified by magnitude and phase) to each of the flexion direction (forward direction) and the stretch direction (backward direction) of each thigh, and a self-inhibition factor v={vi|i=L+, L−, R+, R−} for representing compliance of each behavior state. Moreover, the simultaneous differentiation equation contains therein the desired value ζ0 for each of the left hip joint angle and the right hip joint angle at the finished time of the flexion motion and the stretch motion of the thigh every walking cycle, respectively, and the coefficient c to be corrected as mentioned above. Moreover, it is acceptable to increase the numbers of the second motion oscillator φ2r namely the determination object. The more numbers of the second motion oscillator φ2 are input to the second model, the more the correlation members in the simultaneous differentiation equation will become, however, it may allow an appropriate assist in the periodical motion of the human P by considering the correlation between motion states of various body parts of the human P through the adjustment of the correlation coefficients.
τ1/L+(duL+/dt)=cL+ζ0L+−uL++wL+/L−ξ2L−+wL+/R+ξ2R+−λLvL++f1(ω2L)+f2(ω2L)K2φ2L,
τ1/L−(duL−/dt)=cL−ζ0L−−uL−+wL−/L+ξ2L++wL−/R−ξ2R−−λLvL−+f1(ω2L)+f2(ω2L)K2φ2L,
τ1/R+(duR+/dt)=cR+ζ0R+−uR++wR+/L−ξ2L−+wR+/R+ξ2R+−λRvR++f1(ω2R)+f2(ω2R)K2φ2R,
τ1/R−(duR−/dt)=cR−ζ0R−−uR−+wR−/L+ξ2L++wR−/R−ξ2R−−λRvL−+f1(ω2R)+f2(ω2R)K2φ2R,
τ2i(dvi/dt)=−vi+ξ2i,
ξ2i=H(ui−uth)=0(ui<uthi) or ui(ui≧uthi), or
ξ2i=fs(ui)=ui/(1+exp(−ui/D)) (30)
“τ1i” is a time constant for defining the variation feature of the state variable ui. τ1i is represented by the equation (31) using a ω-dependant coefficient t(w) and a constant γ=(γL, γR) and varies dependent on the second intrinsic angular velocity ω2.
τ1i=(t(ω2L)/ω2L)−γL(i=L+,L−), (t(ω2R)/ω2R)−γR(i=R+,R−) (31)
“τ2i” is a time constant for defining the variation feature of the self-inhibition factor vi. “wi/j” is a negative second correlation coefficient representing the correlation between the state variables ui and uj which represent the motions of the left and right legs of the human P toward the flexion direction and the stretch direction as the correlation of each component of the second oscillator ξ2 (correlation between the output oscillation signals of the plurality of the second elements). “λL” and “λR” are compliant coefficients. “K2” is a feedback coefficient related to the second motion oscillator φ2.
“f1” is a linear function of the second intrinsic angular velocity ω2 defined according to the equation (32) using the positive coefficient c. “f2” is a quadratic function of the second intrinsic angular velocity ω2 defined according to the equation (33) using the coefficients c0, c1 and c2.
f
1(ω)≡cω (32)
f
2(ω)≡c0+c1ω+c2ω2 (33)
The second oscillator ξ2i equals to zero when the value of the state variable ui is smaller than a threshold value uth; and equals to the value of ui when the value of the state variable ui is not smaller than the threshold value uth. In other words, the second oscillator ξ2i is defined by a sigmoid function fs (refer to equation (30)). According thereto, if the state variable uL+ representing the behavior of the left thigh toward the forward direction increases, the magnitude of the left flexion component ξ2L+ of the second oscillator ξ2 becomes greater than that of the left stretch component ξ2L−; if the state variable uR+ representing the behavior of the right thigh toward the forward direction increases, the magnitude of the right flexion component. ξ2R+ of the second oscillator ξ2 becomes greater than that of the right stretch component ξ2R−. Further, if the state variable uL− representing the behavior of the left thigh toward the backward direction increases, the magnitude of the left stretch component ξ2L− of the second oscillator ξ2 becomes greater than that of the left flexion component ξ2L+; if the state variable uR− representing the behavior of the right thigh toward the backward direction increases, the magnitude of the right stretch component ξ2R− of the second oscillator ξ2 becomes greater than that of the right flexion component ξ2R+. The motion toward the forward or backward direction of the leg (thigh) is recognized by, for example, the polarity of the hip joint angular velocity.
Next, the auxiliary oscillator generation element 150 sets the auxiliary oscillator η=(ηL, ηR) on the basis of the second oscillator ξ2 generated by the second oscillator generation element 140 (FIG. 3/S050). Specifically, the auxiliary oscillator η is generated according to the equation (40). In other words, the left component ηL of the auxiliary oscillator η is calculated as a sum of a product of the left flexion component ξ2L+ of the second oscillator ξ2 and the coefficient χL+, and a product of the left stretch component ξ2L− of the second oscillator ξ2 and the coefficient “−χL−”. The right component ηR of the auxiliary oscillator n is calculated as a sum of a product of the right flexion component ξ2R+ of the second oscillator ξ2 and the coefficient χR+, and a product of the right stretch component ξ2R− of the second oscillator and the coefficient “−χR−”
ηL=χL+ξ2L+−χL−ξ2L−, ηR=χR+ξ2R+−χR−ξ2R− (40)
Thereafter, a current I=(IL, IR) supplied to each of the left and right actuators 15 from the battery 1000 is adjusted by the first controller 100 on the basis of the auxiliary oscillator n. The current I is represented by, for example, I(t)=G1·η(t) (wherein, G1 is a ratio coefficient) on the basis of the auxiliary oscillator η. Thereby, the force for moving each thigh (the second body part) with respect to the waist (the first body part) or the torque T=(TL, TR) around the hip joint, which is applied to the human P from the motion assist device 10 via the first orthosis 1100 and the second orthosis 1200, is 2C adjusted (FIG. 3/S060). The torque T is represented by, for example, T(t)=G2·I(t) (wherein, G2 is a ratio coefficient) on the basis of the current I. Thereafter, the series of the aforementioned processes are performed repeatedly. Note that it is acceptable to control the motion of the motion assist device 10 irrelative to the aforementioned control method on the condition that the thigh is appropriately moved with respect to the waist in a duration from the initiation of the walking motion of the human P to the end of 2-3 steps of the walking motion.
According to the motion assist device 10 in the first embodiment of the present invention which performs the aforementioned functions, the second orthosis 1200 is moved by the actuator 15 with respect to the first orthosis 1100 as illustrated in
Specifically, the motion of the human P can be assisted by the motion assist device 10 so that the motion rhythm of the human P matches the desired motion rhythm due to the following reasons. In other words, as mentioned in the above, the second intrinsic angular velocity ω2 is set appropriately from the viewpoint of approximating the phase difference between the periodical motion of the human P represented by the first virtual oscillator φ1 and the periodical motion of the motion assist device 10 represented by the second virtual oscillator φ2 to the desired phase difference δθ0 while the mutual harmonization between the first motion oscillator φ1 and the first oscillator ξ1 is maintained between the periodical motion of the human P and the periodical motion of the motion assist device 10 (FIG. 3/S031-S036). Since the second oscillator ξ2 varies periodically at the angular velocity defined on the basis of the second intrinsic angular velocity ω2 and the output torque T is controlled on the basis of the second oscillator ξ2, therefore, the output torque T varies periodically at the angular velocity defined on the basis of the second intrinsic angular velocity ω2 (FIG. 3/S040, S050 and S060). According thereto, by applying the torque T to the human P, the motion rhythm of the human P and the motion rhythm of the motion assist device 10 are harmonized, and the periodical walking motion of the human P is assisted so as to match the motion rhythm of the human P with the desired motion rhythm.
Moreover, the motion of the human P can be assisted by the motion assist device 1 so that the motion scale of the human P matches the desired motion scale due to the following reasons. In other words, the second model is corrected so as to approximate the value of the motion variable (the left hip joint angle and the right hip joint angle at the finished time of the flexion motion and the stretch motion of the thigh, respectively, in each walking cycle) representing the motion scale of the periodical walking motion of the human P to the desired value ζ0 (FIG. 3/S038). Thereafter, the second oscillator ξ2 is generated according to the corrected second model and the torque T applied to the human P is controlled on the basis of the second oscillator ξ2 (FIG. 3/S050 and S060). Resultantly, in spite of the rhythm speed of the periodical motion of the human P, the periodical motion can be assisted by applying a force with an appropriate strength to the human P so as to make the motion scale (the footstep, the maximum hip joint angle or the like) thereof match the desired motion scale.
In the aforementioned embodiment, it is described that the motion of the human P is assisted. However, it is also possible to assist the walking motion of a creature other than a human, such as an ape, a dog, a horse, cattle or the like.
In the aforementioned embodiment, the motion assist device 10 is constituted to assist the walking motion of the human P (refer to
The present invention is not limited to the motion assist device 10 to be used in the walk training of the human P as illustrated in
Furthermore, a walking machine 70 as illustrated in
| Number | Date | Country | Kind |
|---|---|---|---|
| 2007-259175 | Oct 2007 | JP | national |
| 2007-268304 | Oct 2007 | JP | national |
| Filing Document | Filing Date | Country | Kind | 371c Date |
|---|---|---|---|---|
| PCT/JP2008/002234 | 8/19/2008 | WO | 00 | 4/1/2010 |
