Hi Abra,
The problem lies in the left leg, you can simply see it in the model view:
the left knee is suddenly deviating sideways and the foot takes a strange
orientation trying to keep matching the markers. At the moment the knee
marker is only driven in the anterior-posterior direction, it is probably a
good idea to drive it laterally also to avoid this deviation. Therefore you
need to remove the malleolus driver to keep the balance.
I tried it but it appears that it doesn’t solve the problem completely.
There is a problem now with the ankle eversion, especially on the right
side. Ideally it should be controlled by a marker on the side of the foot.
Such marker was not available in the GaitVaughan data so we had to use the
other markers the best we could. But it seems the same configuration doesn’t
work on your model. In top of that the markers on the pelvis are quite far
(and some others too). Maybe this is not the main problem but it will anyway
introduce some error in the kinematic and that is not nice to have. Maybe
you should verify the position of the body marker of the experiment, they
can differ a little with the current position on AnyBody.
So I’m afraid there is no quick fix, the model is close to running but there
are too many small kinematic inaccuracies in top of each other. I think to
have a clean kinematic you will have to go through the whole process I
described previously.
By the way do you have the exact anthropometric data of the person to input
in the AnyMan file? That way the model will be scaled correctly for sure, it
will be one parameter less to worry about. Then the body marker position
should be checked. With this you will have a good base to start working with
and go through the mentioned process.
Best regards,
Sylvain, AnyBody Support
From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of ens.abra
Sent: 2. marts 2009 15:28
To: anyscript@yahoogroups.com
Subject: [AnyScript] Re: GaitVaughn model
Hi Soeren,
Thanks for the advice. I looked at the constraint violations, and none
seems noticeably larger than the others. I have uploaded the file to
the folder “GaitVaughn “woman””. Any help you can offer is appreciated.
Thanks,
Abra
— In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com,
AnyScript Support <anyscriptsup@…>
wrote:
>
> Hi Abra
>
>
>
> Please upload the model to the file section of the group.
>
>
>
> When the model fails, please try to have a look at which of the
constraints
> violations is largest it can sometimes give you an idea on where the
problem
> is.
>
>
>
> Another trick is to simply scale up the segments a bit can help
solve the
> problem too, other than that it is an iterative process of small
changes to
> the marker locations segments sizes etc. as Sylvain describes.
>
>
>
> It can of course also be drop out maker at this time step that causes
> problems, this is the first thing to check.
>
>
>
>
>
> Best regards
>
> Søren, AnyBody Support
>
>
>
> _____
>
> From: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com
[mailto:anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com]
On Behalf
> Of ens.abra
> Sent: 27 February 2009 19:36
> To: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com
> Subject: [AnyScript] Re: GaitVaughn model
>
>
>
> Hi Sylvain,
> Thanks for your help. I am able to run SetInitialConditions with the
> marker drivers on without encountering any errors. However, The
> kinematic analysis always fails at around step 20. I’ve tried tweaking
> the scaling, marker placements and constraints, but nothing seems to
> make a difference - the kinematic analysis still fails. Is it possible
> for you to take a look at the code?
> Thanks,
> Abra
>
> — In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com,
> AnyScript Support <anyscriptsup@>
> wrote:
> >
> > Hi Abra,
> >
> >
> >
> > Let’s forget the muscles for the moment. There is a more fundamental
> problem
> > with the kinematic, and it has to be solved first.
> >
> > The GaitVaughan model is actually quite difficult kinematically.
> There is
> > two sets of marker, on set is the body markers (belonging to the
> segments)
> > and the other set is the markers recorded by the motion capture
> system. The
> > idea is then to hook up the two sets in a certain way so that the body
> > follows the recorded motion but at the same time we have to avoid over
> > constraining the model. You said that you changed the marker data,
> did you
> > change both sets?
> >
> >
> >
> > As you are modelling a different person then for sure you should
> have a new
> > set of recorded motion markers. Depending on the placement of those
> markers
> > on the body during the record you might have to modify also the body
> markers
> > of the model to fit the configuration. This is done together with the
> > scaling of the model, by those two operations you should end up
> having the
> > two sets of markers matching almost perfectly when you load the
> model. It
> > can sometimes be a quite long operation.
> >
> >
> >
> > Notice that ScalingStandard gives the default size to the model of
a 50
> > percentile European male. You cannot modify the size of the model
> with this
> > scaling law, so you should use one of the three others to model the
> woman
> > size.
> >
> >
> >
> > Now the trick is the following: at the beginning remove the drivers
> that are
> > hooking the two marker sets and use the joint drivers from the
> FreePosture
> > model. With the Mannequin file (controlling the joint drivers) try to
> > position the body in the same initial position as the recorded
> markers. At
> > the same time scale the model and adjust the body markers placement
> so that
> > the two sets of markers match pretty well when you run
> SetInitialConditions.
> > When this is done you should free one DOF at a time and replace the
> joint
> > driver by the corresponding marker hooking driver, beginning
> preferably by
> > the proximal segments like pelvis or thorax. For each new DOF
freed make
> > sure that the system can solve the kinematic constraints by running
> > SetInitialConditions. If this is not the case then some
adjustments are
> > needed in either the initial position or body marker placement or
> even in
> > the driver (direction of the constraint etc).
> >
> > Once all the markers are hooked up and the SetInitialConditions run
> > successfully the model is ready to perform the kinematic analysis.
> >
> >
> >
> > So that is the way to go for the moment. Then you will be able to
> look at
> > the muscles performance.
> >
> >
> >
> > Best regards,
> >
> > Sylvain, AnyBody Support
> >
> >
> >
> >
> >
> > _____
> >
> > From: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com
> [mailto:anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com]
> On Behalf
> > Of ens.abra
> > Sent: 25. februar 2009 16:49
> > To: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com
> > Subject: [AnyScript] Re: GaitVaughn model
> >
> >
> >
> > Hi Sylvain,
> >
> > Even after calibration I get both errors. I tried running the
> > KinematicAnalysis with the muscles turned off. I still got the
> > kinematic error. Could these errors have something to do with scaling?
> > I tried switching from ScalingUniform to ScalingStandard, and while
> > the kinematicAnalysis still failed, it went through about 15 more
> > steps before it did so.
> >
> > The only file that I changed for scaling was the AnyMan file. (I
> > changed mass, thigh length, calf length, and foot length.) Is there
> > something more that I should have changed to reflect the differences
> > between the subjects?
> >
> > Thanks,
> > Abra
> >
> > — In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
ps.com,
> > AnyScript Support <anyscriptsup@>
> > wrote:
> > >
> > > Hi Abra,
> > >
> > >
> > >
> > > Do you mean you still get the warning “Muscle model ‘Main.’ did not
> > > converge. Please check muscle length parameters.”?
> > >
> > > Or just the error “Failed to resolve kinematic constraints”?
> > >
> > > You should not have problems with the muscle model after the
> > calibration, so
> > > it could be a kinematic error due to the markers for example. This
> > is the
> > > first thing to check. Please run a KinematicAnalysis with the body
> model
> > > without muscles. If this doesn’t work then you have to solve it
before
> > > anything else.
> > >
> > >
> > >
> > > Once you are sure the kinematic of the markers is ok you can try to
> > run the
> > > inverse dynamic analysis with the simple muscle and the 3 elements
> > muscle.
> > > See if both are failing or not, it will give a clue if the problem
> > is still
> > > related to the muscle model or something else.
> > >
> > >
> > >
> > > Best regards,
> > >
> > > Sylvain, AnyBody Support
> > >
> > >
> > >
> > >
> > >
> > >
> > >
> > > _____
> > >
> > > From: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
ps.com
> > [mailto:anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
ps.com]
> > On Behalf
> > > Of ens.abra
> > > Sent: 24. februar 2009 15:32
> > > To: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com> ps.com
> > > Subject: [AnyScript] Re: GaitVaughn model
> > >
> > >
> > >
> > > Hi Sylvain,
> > > Thanks for your help. The calibration finishes without any errors,
> > > then the program tries to run the Inverse Dynamic Analysis and I get
> > > the errors I’ve described below. Is it possible that something
else is
> > > the problem?
> > > Thanks,
> > > Abra
> > >
> > > — In anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
> ps.com,
> > > AnyScript Support <anyscriptsup@>
> > > wrote:
> > > >
> > > > Hi Abra,
> > > >
> > > >
> > > >
> > > > From the error message I assume you are using the 3 elements
muscle
> > > model.
> > > > Then it is typically a calibration problem. The 3 elements model
> > has to
> > > > solve an internal kinematic between the tendon and the contractile
> > > element,
> > > > depending on the length of each of them. An initial length is
> > > defined in the
> > > > code for the tendon and the contractile element but it is not
meant
> > > to be
> > > > the optimal one and a calibration is needed to give those elements
> > their
> > > > optimal length in function of the model anthropometry. If the
> > > calibration is
> > > > not run then problems can occur such as, for example, the tendons
> > > being too
> > > > long and not leaving any space for the muscle in between them.
That
> > > would
> > > > make sense in your case as you probably shortened the segments to
> > > model a
> > > > woman.
> > > >
> > > >
> > > >
> > > > So you have to run the calibration to solve this, just select
> > > RunApplication
> > > > in the operation tree and run it, it will automatically calibrate
> > > the model
> > > > and run the inverse dynamic analysis.
> > > >
> > > >
> > > >
> > > > Best regards,
> > > >
> > > > Sylvain, AnyBody Support
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > _____
> > > >
> > > > From: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
> ps.com
> > > [mailto:anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
> ps.com]
> > > On Behalf
> > > > Of ens.abra
> > > > Sent: 23. februar 2009 16:14
> > > > To: anyscript@yahoogrou <mailto:anyscript%40yahoogroups.com>
ps.com
> > > > Subject: [AnyScript] GaitVaughn model
> > > >
> > > >
> > > >
> > > > Hello everyone,
> > > >
> > > > I’m an 3rd year undergrad student working as a research
assistant at
> > > > the University of Waterloo, Canada. My project for this semester
> is to
> > > > find the muscle forces acting on the knee during a lay-up
> (basketball
> > > > jump). So far I have been working with the gait models
> (gaitVaughn and
> > > > gaitMiamiUni) to learn my way around them and to eventually
> input the
> > > > data I will be given for a lay-up.
> > > >
> > > > Since the data I will be using for the lay-up isn’t quite ready,
> I’ve
> > > > been trying to input Vaughn’s “woman” data (as found on the ISB
> > > > website) into the gaitVaughn model. I’ve changed the force plate
> data,
> > > > EMG, anthropometric (in the anyMan.any file), marker data and the
> > > > time.any file, but I can’t seem to get the program to run
> properly. It
> > > > will load, but when I run “set initial conditions” I come up
> with this
> > > > error message:
> > > >
> > > > Set Initial Conditions: (Operation:
> Main.Study.SetInitialConditions):
> > > > 0) Design variables have been updated.
> > > > 1) Load time positions have been re-established.
> > > > 2) Kinematic analysis completed. The kinematic constraints
have been
> > > > resolved.
> > > > Muscle model
> > > >
‘Main.HumanModel.BodyModel.Right.Leg.MusPar.AdductorMagnus1Par’ did
> > > > not converge. Please check muscle length parameters.
> > > > Muscle model
> > > >
‘Main.HumanModel.BodyModel.Right.Leg.MusPar.AdductorMagnus2Par’ did
> > > > not converge. Please check muscle length parameters.
> > > > Muscle model
> > > >
‘Main.HumanModel.BodyModel.Right.Leg.MusPar.QuadratusFemorisPar’ did
> > > > not converge. Please check muscle length parameters.
> > > > Muscle model
> > > >
‘Main.HumanModel.BodyModel.Right.Leg.MusPar.GemellusInferiorPar’ did
> > > > not converge. Please check muscle length parameters.
> > > > 3) Initial conditions are fully updated
> > > >
> > > > When I run the inverse dynamic analysis, the program crashes and
> gives
> > > > me this message:
> > > >
> > > > Failed to resolve kinematic constraints (final kin. error =
> > > 8.868940E-004)
> > > >
> > > > If someone could tell me where I’m going wrong, I would greatly
> > > > appreciate the help.
> > > >
> > > > Thanks,
> > > > Abra
> > > >
> > > >
> > > >
> > > >
> > > >
> > > > [Non-text portions of this message have been removed]
> > > >
> > >
> > >
> > >
> > >
> > >
> > > [Non-text portions of this message have been removed]
> > >
> >
> >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>
>
>
>
>
> [Non-text portions of this message have been removed]
>
[Non-text portions of this message have been removed]