Recruitment failed

Hi again,

The classical inverse dynamics now works perfectly on my sit-to-stand model.
Thanks again.

But now, when I run the muscle recruitment analysis, I get:

ERROR(OBJ1) : C:\Documents and Settings\Jean-Olivier
Racine\Desktop\LIO\Anybody\Repository.6\ARep\LIO\STS Test
4\MuscleStudy.any(1) : ‘MuscleStudy’ : Muscle recruitment analysis
failed, simplex solver reports that solution does not satisfy all
constraints.

My questions:

  1. From the (Damsgaard, Rasmussen, Christensen, 2001, ASME) article on
    the min/max optimization criteria, I see that the constraints used for the
    min/max criteria are quite classical and are 1) the dynamics equilibrium
    equations and 2) the muscle pull constraint. Is there a way to know which of
    these two constraints is not satisfied by the simplex solver? I know the
    simplex solver, in essence, doesn’t naturally allow it. but we never know.
  2. I tried running the analysis with 4 different trials (data set) and
    in each case, the error appears exactly at seat-off. Knowing that currently
    I do not include the seat force but only the ground force, could that be the
    problem? If it is, what interactions within anybody explain that problem? I
    am surprised that the problem does not appear right from the beginning, when
    the seat force is missing, and not when the seat force disappears
    (seat-off).

I uploaded the model to the group. Filename is JeanOlivier2.rar.

Thanks,

Jean-Olivier

[Non-text portions of this message have been removed]

Hello Jean-Olivier,

> 1. From the (Damsgaard, Rasmussen, Christensen, 2001,
> ASME) article on the min/max optimization criteria, I see
> that the constraints used for the min/max criteria are
> quite classical and are 1) the dynamics equilibrium
> equations and 2) the muscle pull constraint. Is there a
> way to know which of these two constraints is not satisfied
> by the simplex solver? I know the
> simplex solver, in essence, doesn’t naturally allow it.
> but we never know.

No, it is unfortunately very difficult to do. What happens is that
the feasible domain of the problem, i.e. the space of solutions that
fulfil the constraints, is empty. When this happens it is hard to say
whether it is empty because the muscles are not allowed to pull or
because the equilibrium constraints are not fulfilled. In principle a
relaxtion of either of those two might make the problem solvable.

Unfortunately we also sometimes see that the problem is of numerical
nature with the equlibrium equations being ill-conditioned. So even
though they have a solution in priciple, the algorithm is not capable
of finding it for numerical reasons.

I would try with different settings of the RecruitmentSolver and
RecruitmentLpPenalty variables in the study.

> 2. I tried running the analysis with 4 different
> trials (data set) and in each case, the error appears
> exactly at seat-off. Knowing that currently
> I do not include the seat force but only the
> ground force, could that be the
> problem? If it is, what interactions within anybody
> explain that problem? I am surprised that the problem does
> not appear right from the beginning, when
> the seat force is missing, and not when the seat force disappears
> (seat-off).

I understand from what you wrote above that you are using applied
ground reaction forces measured from force platforms. If this is
correct, then the problem might be that you do not have enough actual
reaction forces in the model after it has left the seat.

> I uploaded the model to the group. Filename is JeanOlivier2.rar.

I did not have time to look at the model yet. Sorry. I hope the tips
above might work.

Best regards,
John

Hi Jean-Oliver

I have looked through your model and found a few things.

In the file “leg.root.any” there is an include statement to the 3
element muscle model, but in your main file there is no
specification of any muscle calibration. This means that some
muscles have zero strength and this causes the problem.

So in the file leg.root.any you need do a change so it looks like
this:

// MuscleParameters

AnyFolder MusPar = {

AnyVar StressIndex = .StrengthRef.StrengthIndexLeg;

#include “…/…/…/…/BRep/Aalborg/Leg3D/MusParSimple.any”

};

otherwise you need to include some calibration of the leg. If you
want to do this you need to make some changes to the normal
calibration files because you have removed a lot of muscles from the
model etc.

Best regards
Søren. AnyBody Support

— In anyscript@yahoogroups.com, Jean-Olivier Racine <jean-
olivier.racine.1@…> wrote:
>
> Hi again,
>
>
>
> The classical inverse dynamics now works perfectly on my sit-to-
stand model.
> Thanks again.
>
>
>
> But now, when I run the muscle recruitment analysis, I get:
>
> ERROR(OBJ1) : C:\Documents and Settings\Jean-Olivier
> Racine\Desktop\LIO\Anybody\Repository.6\ARep\LIO\STS Test
> 4\MuscleStudy.any(1) : ‘MuscleStudy’ : Muscle recruitment
analysis
> failed, simplex solver reports that solution does not satisfy all
> constraints.
>
>
>
> My questions:
>
> 1. From the (Damsgaard, Rasmussen, Christensen, 2001, ASME)
article on
> the min/max optimization criteria, I see that the constraints used
for the
> min/max criteria are quite classical and are 1) the dynamics
equilibrium
> equations and 2) the muscle pull constraint. Is there a way to
know which of
> these two constraints is not satisfied by the simplex solver? I
know the
> simplex solver, in essence, doesn’t naturally allow it. but we
never know.
> 2. I tried running the analysis with 4 different trials (data
set) and
> in each case, the error appears exactly at seat-off. Knowing that
currently
> I do not include the seat force but only the ground force, could
that be the
> problem? If it is, what interactions within anybody explain that
problem? I
> am surprised that the problem does not appear right from the
beginning, when
> the seat force is missing, and not when the seat force disappears
> (seat-off).
>
>
>
> I uploaded the model to the group. Filename is JeanOlivier2.rar.
>
>
>
> Thanks,
>
> Jean-Olivier
>
>
>
> [Non-text portions of this message have been removed]
>

Wow, thank you. I guess I removed those during a test and forgot to put them
back! Thanks!

From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of AnyBody Support
Sent: Wednesday, May 23, 2007 3:34 AM
To: anyscript@yahoogroups.com
Subject: [anyscript] Re: Recruitment failed

Hi Jean-Oliver

I have looked through your model and found a few things.

In the file “leg.root.any” there is an include statement to the 3
element muscle model, but in your main file there is no
specification of any muscle calibration. This means that some
muscles have zero strength and this causes the problem.

So in the file leg.root.any you need do a change so it looks like
this:

// MuscleParameters

AnyFolder MusPar = {

AnyVar StressIndex = .StrengthRef.StrengthIndexLeg;

#include “…/…/…/…/BRep/Aalborg/Leg3D/MusParSimple.any”

};

otherwise you need to include some calibration of the leg. If you
want to do this you need to make some changes to the normal
calibration files because you have removed a lot of muscles from the
model etc.

Best regards
Søren. AnyBody Support

— In anyscript@yahoogroups.com <mailto:anyscript%40yahoogroups.com> ,
Jean-Olivier Racine <jean-
olivier.racine.1@…> wrote:
>
> Hi again,
>
>
>
> The classical inverse dynamics now works perfectly on my sit-to-
stand model.
> Thanks again.
>
>
>
> But now, when I run the muscle recruitment analysis, I get:
>
> ERROR(OBJ1) : C:\Documents and Settings\Jean-Olivier
> Racine\Desktop\LIO\Anybody\Repository.6\ARep\LIO\STS Test
> 4\MuscleStudy.any(1) : ‘MuscleStudy’ : Muscle recruitment
analysis
> failed, simplex solver reports that solution does not satisfy all
> constraints.
>
>
>
> My questions:
>
> 1. From the (Damsgaard, Rasmussen, Christensen, 2001, ASME)
article on
> the min/max optimization criteria, I see that the constraints used
for the
> min/max criteria are quite classical and are 1) the dynamics
equilibrium
> equations and 2) the muscle pull constraint. Is there a way to
know which of
> these two constraints is not satisfied by the simplex solver? I
know the
> simplex solver, in essence, doesn’t naturally allow it. but we
never know.
> 2. I tried running the analysis with 4 different trials (data
set) and
> in each case, the error appears exactly at seat-off. Knowing that
currently
> I do not include the seat force but only the ground force, could
that be the
> problem? If it is, what interactions within anybody explain that
problem? I
> am surprised that the problem does not appear right from the
beginning, when
> the seat force is missing, and not when the seat force disappears
> (seat-off).
>
>
>
> I uploaded the model to the group. Filename is JeanOlivier2.rar.
>
>
>
> Thanks,
>
> Jean-Olivier
>
>
>
> [Non-text portions of this message have been removed]
>

[Non-text portions of this message have been removed]