When a person suffers a stroke, one of the resulting physical impairments is wrist spasticity, which causes excessive wrist stiffness and resistance to stretch, known as “tone.” Studies suggest that this increase in stiffness, or tone, may be due to changes in the wrist flexor and extensor muscles, which result in an increased torque demand to achieve wrist movement.
While devices have been developed that provide assistive force or torque to the wrist in flexion or extension to assist with wrist movement, such devices only provide assistance in one of these directions. Moreover, this assistance decreases as the angular rotation of the wrist increases, which is the opposite of what is needed given that rotation becomes more difficult as the angle of rotation increases. In view of this, it can be appreciated that it would be desirable to have a device that mitigates the intrinsic stiffness of stroke victim wrists and therefore provides assistance to the individual in bending the wrist in both flexion and extension, as well as holding a desired angular position.
The present disclosure may be better understood with reference to the following figures. Matching reference numerals designate corresponding parts throughout the figures, which are not necessarily drawn to scale.
As described above, it would be desirable to have a device that mitigates the intrinsic stiffness of stroke victim wrists and therefore provides assistance to the individual in bending the wrist in both flexion and extension, as well as holding a desired angular position. Disclosed herein are torque-compensating assistive wrist braces that provide assistive torque that counteracts the increased torque that the intrinsic stiffness imposes. In some embodiments, the wrist brace comprises a hand member, a forearm member, and an assistive linkage that provides supporting forces between the hand and forearm members around the wrist in order to balance and cancel out the increased torque. This results in a reduction in effort required by the individual to achieve or hold a desired angular wrist position. In some embodiments, the linkage comprises a spring-actuated six-bar linkage that provides an angle-dependent supporting torque that balances the torque imposed by the wrist stiffness.
In the following disclosure, various specific embodiments are described. It is to be understood that those embodiments are example implementations of the disclosed inventions and that alternative embodiments are possible. All such embodiments are intended to fall within the scope of this disclosure.
As noted above, the disclosed wrist braces can comprise an assistive linkage that provides supporting forces between a hand member and a forearm member. In some embodiments, this linkage comprises a spring-assisted six-bar linkage. Described below is a direct solution of the kinematic synthesis equations for Stephenson III six-bar linkages that yield function generators that can achieve as many as 11 accuracy points. The synthesis equations yield thousands of candidate linkage designs that can be analyzed to identify those that achieve a required task in a single configuration and without passing through a singularity. This requires identification of the function generator cognates among the design candidates and a direct numerical analysis of each candidate design. It can happen that a cognate solution does not appear in the synthesis results due to numerical issues, in which case those cognates are constructed and added to the synthesis results. The result is a set of linkages that achieve the specified function generation task and are free of branch and circuit defects.
A six-bar linkage comprises four binary links, i.e., links having two joints or pivot points, and two ternary links, i.e., links having three joints or pivot points. These systems form two distinct topologies known as Watt and Stephenson six-bar linkages. The Stephenson six-bar linkages have the property that the two ternary links are separated by the binary links, while the Watt topology has the ternary links connected to each other.
The kinematic synthesis of six-bar function generators is an extension of the original work by Freudenstein (1954), in which the loop equations of the system are formulated in each of the configurations specified by the required input-output angles. The result is a set of polynomial equations that are solved to determine the dimensions of the linkage.
The planar kinematics of linkages is conveniently formulated using complex numbers. The scale, orientation, and location of a reference configuration for the linkage is defined in the plane by selecting A=0+0i and D=1+0i. The coordinates of the remaining joints B, C, F, G and H are calculated by solving the synthesis equations.
Introduce (Δϕj, Δψj), j=1, . . . , N−1 as the required input-output angles measured from the reference configuration (
(Δϕj,Δψj)=(ϕj−ϕ0,ψj−ψ0), j=1, . . . ,N−1. (1)
The synthesis equations for the Stephenson III linkage are formed from the loop equations for each set of specified input-output angles.
The coordinates of the moving pivots of the linkage in each task configuration are related to their coordinates in the reference configuration by the equations,
Dj=A+eΔφj(D−A),
Fj=B+eΔψj(F−B),
Gj=A+eΔφj(D−A)+eΔμj(G−D),
Hj=A+eΔφj(D−A)+eΔμj(H−D), j=1, . . . ,N−1. (2)
For convenience, introduce the notation,
Qj=eiΔφj, Rj=eiΔρj, Sj=eiΔψj,
Tj=eΔθj, Uj=eΔμj, j=1, . . . ,N−1. (3)
Notice that Qj and Sj are defined by the required input-output angle pairs. The remaining joint angles are unknowns that satisfy the normalization conditions,
Rj
where the overbar denotes the complex conjugate.
The loop equations for the Stephenson III six-bar linkage are obtained by evaluating Gj−C and Hj−Fj relative to the initial configuration. This yields two sets of complex conjugate loop equations,
Lj:Rj(G−C)=(A+Qj(D−A)+Uj(G−D))−C,
j=1, . . . ,N−1,
Mj:Tj(H−F)=(A+Qj(D−A)+Uj(H−D))−(B+Sj(F−B)),
j=1, . . . ,N−1. (5)
The loop equations Lj, Mj, and the normalization conditions of Equation (4) form 7(N−1) quadratic equations in the 2(5+3(N−1)) unknowns consisting of the pivots locations B, C, F, G, and H and the joint rotations Rj, Tj, and Uj and their complex conjugates. The Stephenson III synthesis equations are similar in form to the Stephenson II and both can be solved for a maximum of N=11 positions to obtain 70 quadratic equations in 70 unknowns that yield a total degree of 270=1.18×1021.
The 70 synthesis equations for the Stephenson III function generator can be reduced to 10 equations in 10 unknowns. This can be achieved by eliminating Rj and
To simplify the presentation of this calculation, introduce the complex numbers,
a=G−D, f=G−C, h=A−C, k=D−A,
c=H−D, g=H−F, m=A−B, o=−(F−B), (6)
so the loop equations take the form
Lj:h+Qjk+Uja−Rjf=0,
Mj:m+Qjk+Ujc+Sjo−Tjg=0,
Eliminate R and
(h+Qjk+Uja)(
(m+Qjk+Ujc+Sjo)(
These 10 pairs of equations are linear in Uj and Ūj, and can be written in the form,
where the complex numbers bj=Dj−C and dj=Dj−Fj, given by
bj=h+Qjk,
dj=m+Qjk+Sjo, j=1, . . . ,10, (10)
are introduced to simplify the presentation of these equations.
Eliminate Uj and Ūj between the pairs of Equation (9) in order to obtain,
where the vertical bars denote the determinant.
The total degree of the polynomial system in Equations (11) is 810=1.07×109, which is a similar case for the Stephenson II linkage. However, the simpler form of Equation (10) allows for an additional reduction before employing a multihomogeneous root count.
In order to reduce the degree of the synthesis Equations (11), introduce the variables,
r1=a
This allows the expansion of the terms,
a
c
And similarly the rest of the terms in Equations (11) can be expanded using the additional identities f=a+h+k and g=c+k+m+o to find,
See that k and
The synthesis Equations (11) can be now be written as,
The result is a set of 10 quartic polynomials, which together with the eight quadratic polynomials (Equation (12)) yields a polynomial system of degree, 41028=268, 435, 456.
However, the 18 unknowns in this polynomial system can be separated into the two homogeneous groups,
C,
The number of roots of this system of equations can be calculated as the coefficient of α16α212 in the expansion of
256α12α26(2α1+2α2)10 (17)
which yields a multihomogeneous degree, 55,050,240. This is a significant reduction in the size of the polynomial homotopy needed to solve these synthesis equations.
The 18 synthesis Equations (12) and (15) were solved on the Gordon cluster at the San Diego Supercomputer Center of the XSEDE supercomputing network using the polynomial homotopy software Bertini. Rather than specify the requirements for a particular task, the input parameters, (Qj, Sj), j=1, . . . , 10, were set to random complex numbers to create a numerically general system. Homotopy paths were tracked over 40 hours on 512×2.6 GHz cores. Nonsingular solutions were sorted by their Jacobian condition number of which 834,441 were found.
The roots for the general system need only be computed once and then can be used as startpoints for a parameter homotopy for any particular set of input parameters. The advantage of parameter homotopy is that nonsingular endpoints of a general system are used as startpoints of a specific system so that only 834,441 paths need to be tracked in order to find all nonsingular solutions of a that specific system.
The solutions of the synthesis equations are examined to determine those that have real values for the linkage dimensions. This is checked by ensuring the joint coordinate pairs (B,
Each design candidate is also evaluated to identify its cognate pair among the solutions to the synthesis equations, or to construct the cognate, if it does not appear among these solutions.
For every Stephenson III function generator, there exists one other Stephenson III function generator with link lengths of different ratios that produces the exact same function. In order to compute this function generator cognate for a Stephenson III linkage ABCDFGH, the linkage can be considered as a four-bar linkage ADGC that controls the motion of the RR dyad BF that is connected at H. The four-bar ADGC has two other path cognates that generate the same coupler curve at H, and one of these cognates has an input link that shares the same angular displacement Δφ as Link AD throughout the motion of H. The Stephenson III function cognate is built from this four-bar path cognate.
The four-bar ADGC and its path cognates AD′G′C′ and C′D″G″C are shown in
However, the synthesis results only contain linkages with specified pivot locations A=0+0i and D=1+0i. So, for the sake of comparison, the cognate linkage must be scaled, rotated, and translates such that pivots C′ and D″ line up with pivots A and D. The transformation which computes this action on a point p is
or equivalently,
Applying ℑ to the cognate linkage, the coordinates of six-bar (ABCDFGH)c are
Therefore, for every linkage solution {B, C, F, G, H} there should exist another solution {Bc, Cc, Fc, Gc, Hc} in the synthesis results. If a missing cognate solution is detected, it is constructed and added to the results.
Once the design candidates have been sorted into cognate pairs, they are analyzed to evaluate the performance of each design. The criteria for a successful design candidate is the same as was used for the kinematic synthesis of Stephenson II function generators, which is that the required accuracy points lie on a single trajectory of configurations without any singularities. This is determined by computing all the configurations of the linkage for a specified range of input angles.
The kinematics equations of the Stephenson III linkage are obtained from the loop equations as,
L=R(G−C)−(A+Q(D−A)+U(G−D))+C,
M=T(H−F)−(A+Q(D−A)+U(H−D))+(B+S(F−B))=0,
which include the now known initial joint locations,
{A,Ā,B,
and the unknown joint angle parameters,
{Q,
In the case that the angle ψ of Link BF is the input parameter, then the input x and output y variables are
x=(S,
and the analysis equations are
These equations have six solutions for a specified input x=(S,
Singular configurations define the bounds of mechanism branches.
A set of input parameters xk, k=1, . . . , n is generated that sweeps around the unit circle,
Equations (26) are solved for each xk to generate n sets of configurations,
Ck={(xk,yk,1), . . . ,(xk,yk,6)} k=1, . . . , n. (29)
The members of Ck for each k appear in no particular order, and the goal of this section is to sort configurations into separate trajectories as k is incremented from 1 to n.
The algorithm initializes by setting the six elements of C1 as the beginning of six trajectories which are built upon by comparing Ck to Ck+1 and deciphering pairs of connecting configurations,
Ck={(xk,yk,p)|p=1, . . . ,6},
Ck+1={(xk+1,yk+1,q)|q=1, . . . ,6}, (30)
where in general configurations (xk, yk,p) and (xk+1, yk+1,q) connect such that p≠q. To decipher connections between Ck and Ck+1, Newton's method is used to solve F(xk+1, y)=0 for y using start points yk,p, for p=1, . . . , 6. These approximate solutions are named {tilde over (y)}k+1,p where,
{tilde over (y)}k+1,p=yk,p−[JF(xk+1, yk,p)]−1F(xk+1, yk,p),
p=1, . . . ,6 (31)
is calculated from a single Newton iteration. Multiple iterations are used for more accuracy. The approximate configuration set {tilde over (C)}k+1 is formed from {tilde over (y)}k+1,p where
{tilde over (C)}k+1={(xk+1,{tilde over (y)}k+1,p)|p−1, . . . ,6|} (32)
Configuration (xk, yk,p) of Ck connects to configuration (xk+1, yk+1,q) of Ck+1 if the following condition evaluates as true,
|{tilde over (y)}k+1,p−yk+1,q|<tol, (33)
where tol is a specified threshold value. For most k, configurations Ck and Ck+1 will connect in a one to one fashion. However, Equation (33) allows the possibility that a configuration of Ck will connect to several or none of the configurations of Ck+1, which is often the case near singularities. In these cases, the following logic can be employed:
1. If a configuration of Ck+1 is not connected to a configuration of Ck, that configuration of Ck+1 begins a new trajectory.
2. If a configuration of Ck connects to multiple configurations of Ck+1, the trajectory associated with the configuration of Ck is duplicated and each duplicate connects to a matching element of Ck+1.
3. If a configuration of Ck does not connect to any configurations of Ck+1, the trajectory associated with the configuration of Ck is concluded.
This procedure is executed for a complete sweep of the unit circle xk, k=1, . . . , n, such that xn=x1. The result of this algorithm is a set of connected sequences of configurations that form separate mechanism trajectories. All combinations of these trajectories are checked for connections from k=n to k=1 configurations. If connections are identified, these trajectories are chained together to form longer trajectories.
Finally, configurations that do not correspond to rigid body movement are removed and the determinant of the Jacobian matrix along each trajectory is evaluated. A sign change indicates a change in trajectory that can arise from numerical error.
Once all trajectories have been assembled for a linkage design candidate, each is checked to see which and how many of the specified accuracy points they contain. A successful design candidate will produce a trajectory that moves through all 11 accuracy points. These designs are referred to herein as 11-point mechanisms.
While linkage designs that contain all 11 accuracy points on a single trajectory is the goal, the above-described design process identifies linkage designs with trajectories that move through less than 11 points as well. It is often the case that these mechanisms only slightly miss some accuracy points and may have other features useful to the designer, such as compact dimensions or reduced link overlap.
As noted above, survivors of strokes often suffer from a muscle control disorder called spasticity which causes increased stiffness in muscles of joints such as the wrist. Measurement data of intrinsic wrist stiffness is illustrated in
The torque profile that the Stephenson III is to reproduce, and then cancel, can be derived from test data that measures the intrinsic stiffness profile in the wrists of stroke survivors. The test data of 21 such survivors was obtained and was least-squares fit with the following fifth degree polynomial,
S(x)=0.3403347740344527x5+2.3767146714792213x4+1.4329074166324411x3−0.21211179259258692x2+0.5381754676253262x+1.903537638831755. (34)
Because stiffness is the rate of change of a spring torque with respect to angular deflection, S(x) is integrated to obtain the torque profile,
T(x)=∫S(x)dx+c0=−0.056722462339075x6+0.475342934295844x5+0.358226854158110x4−0.070703930864196x3+0.269087733812663x2+1.903537638831755x+1.859723104149862. (35)
Equations (34) and (35) are graphed in
The use of a function generator to provide a required input torque profile begins with the assumption that there are no losses from friction, wear, and dynamic effects, which yields the power balance,
Tin{dot over (x)}=Tout{dot over (y)}, (36)
where {dot over (x)} denotes the angular velocity of the input crank and {dot over (y)} is the angular velocity of the output crank.
For this design, the output torque Tout is generated by a torsion spring with stiffness k and equilibrium angle ye, therefore the input torque is given by,
which is a function of the input angle x. Equation (37) can be solved for y=f(x) to obtain the set of input-output angles needed to design a Stephenson III function generator.
Separate variables and integrate to obtain,
and then solve for y to obtain,
The “+” and “−” solutions are two different functions that produce desired torque profile for given spring parameters k and ye.
The input-output function for the synthesis of the Stephenson III function generator is obtained by substituting Equation (35) into Equation (39) with the requirement that k=0.45 N·m/rad and ye=2π rad. The “−” solution was taken to calculate the input-output y=f(x) function shown in
This input-output function was evaluated at 11 positions of x to obtain the coordinated angles shown in Table. 1. Producing this function was investigated using both AD as the input, (x,y)=(Δϕ, Δψ), and BF as the input, (x, y)=(Δψ, Δϕ). The use of Bertini to obtain solutions to the synthesis equations is the same for both cases.
A summary of synthesis results is shown in Table 2. For the cases with φ as the input and ψ as the input, Bertini found 8,341 and 8,583 solutions that corresponded to physical linkages, respectively. Each solution set was then processed to add cognate solutions and remove solutions with very small or large link lengths such that 4,547 and 5,323 solutions were found for each case. The performance of these linkages was analyzed in order to categorize mechanisms by the number of accuracy points they can achieve in a singularity-free trajectory from 6 to 11 points. For φ as the input and ψ as the input, there were 96 and 109 mechanisms, respectively, that passed through all 11 points. The total computation time for each case was 5 hours and 7 hours performed on 64×2.2 GHz nodes of the University of California, Irvine High Performance Computing Cluster.
The 11-point mechanism shown in
With further reference to
The assistive linkage 16 will now be described with reference to
With further reference to
Unlike the other links, the fourth binary link 56 is pivotally mounted on a transverse shaft 72 that passes through the opening 48 provided in the central longitudinal flange 34. This shaft 72 is illustrated in
As noted above, the right side of the assistive linkage 16 (from the perspective of the wearer) also comprises a six-bar linkage. This six-bar linkage, which is visible in
With further reference to
Like the fourth binary link 56 of the left-side six-bar linkage, the fourth binary link 66 of the right-side six-bar linkage is pivotally mounted on the transverse shaft 72. As is further shown in
This application claims priority to U.S. Provisional Application Ser. No. 62/219,957, filed Sep. 17, 2015, which is hereby incorporated by reference herein in its entirety.
Number | Name | Date | Kind |
---|---|---|---|
3707963 | Keropian | Jan 1973 | A |
5002044 | Carter | Mar 1991 | A |
5503619 | Bonutti | Apr 1996 | A |
5853680 | Iijima et al. | Dec 1998 | A |
5891061 | Kaiser | Apr 1999 | A |
6093162 | Fairleigh | Jul 2000 | A |
20100249948 | Kawakami | Sep 2010 | A1 |
20100280423 | Kawakami | Nov 2010 | A1 |
20110282253 | Menon | Nov 2011 | A1 |
20130053741 | Pittaccio | Feb 2013 | A1 |
20140142483 | Jackson, Jr. | May 2014 | A1 |
Entry |
---|
Balli, et al., “Defects in lilnk mechanisms and solution rectification”, Mechanism and Machine Theory 37(0): 2002. |
Bissell, Derek, Wrist (Wrist Resonator for Independent Stroke Training) Thesis, 2014. |
Blackett, Ricardo, “Optimal synthesis of planar five-link mechanisms for the production of nonlinear mechanical advantage”, Thesis; 2001. |
Bulatavic, et al., “Improved cuckoo search (ICS) algorthm for constrained optimization problems”, Mechanism and Machine Theory; 61: 1-13, 2013. |
Chase, et al., “Circuits and branches of single-degree-of-freedom planar linkages”, Journal of Mechanical Design, 115, 1993. |
Dhingra, et al, “Synthesis of six-link, slider-crank and four-link mechanisms for function, path and motion generation using homotopy with m-homogenization”, Journal of mechanical designs, 116(4), 1994. |
Hwang, et al., “Defect-free synthesis of stephenson-II function generators”, Journal of mechanisms and robotics, 2(4), 2010. |
McLarnan, CW., “Synthesis of six-link plane mechanisms by numerical analysis”, Journal of engineering for industry, 85, 1963. |
Mirbagheri, et al., “Neuromuscular properties of different spastic human joints vary systematically”, 32nd Annual International Conference of IEEE EMBS, 2010. |
Plecnik, et al., “Numerical synthesis of six-bar link-ages for mechanical computation”, ASME Journal of mechanisms and robotics, 6(3), 2014. |
Plecnik, et al., “Computational design of stephenson II six-bar function generators for 11 accuracy points”, ASME Journal of Mechanisms and Robotics, Mar. 2015. |
Root, et al., “A survey of optimization methods applied to design of mechanisms”, ASME, 1976. |
Sancibrian, R. “Improved GRG method for the optimal synthesis of linkages in function generation problems”, Mechanism and machine theory, 2011. |
Shiakolas, et al., “On the optimum synthesis of six-bar linkages using differential evolution and the geometric centroid of precision positions technique”, Mechanism and Machine Theory, 2005. |
Simionescu, et al., “Four-and six-bar function cognates and overconstrained mechanisms”, Mechanism and machine theory, 2001. |
Simionescu, et al., “Synthesis of function generators using the method of increasing the degree of freedom of the mechanism”, Proceedins Ninth World congress on the theory of machines and mechanisms, 1995. |
Wampler, CW, “Isotropic coordinates, circularity, and bezout numbers: planar kinematics from a new perspective” 1996 ASME design engineering technical conference and computers, 1996. |
Number | Date | Country | |
---|---|---|---|
20170079825 A1 | Mar 2017 | US |
Number | Date | Country | |
---|---|---|---|
62219957 | Sep 2015 | US |