Forums Anyscript.org  

Go Back   Forums Anyscript.org > Archive > AnyScript Message Archive

Notices

 
 
Thread Tools Display Modes
  #1  
Old 02-18-2009, 01:16 PM
olepeterson
Guest
 
Posts: n/a
Default AnyBody and stumbeling humans

Hi there!

I am doing some research on humans that stumble and fall. The focus in
this work are the impact forces when eg. the knee hits the ground and
how these forces spread through the limbs. I recorded motion data with
a Vicon MOCAP system and force data with two forceplates and try to
evaluate them in Anybody afterwards.
I have build a suitable model based on GaitUniMiami and adopted it to
my needs with the much appreciated help of this group. After a look at
the first results of the inverse dynamics analysis, I wonder how
reliable the results of muscle forces in AnyBody are. The problem in
this case, I guess, is that the human body tries to work against the
movement of falling, while AnyBody activates the muscles in order to
conduct the movement that is given to it by the MOCAP data.
Can this perception be correct and, if yes, is there a way to work
around this general issue?

Cheers,
Michael

  #2  
Old 02-20-2009, 12:07 PM
AnyScript Support
Guest
 
Posts: n/a
Default RE: [AnyScript] AnyBody and stumbeling humans

Hi Michael



Sounds as nice model



Usually we say that the recruitment algorithm will work for motions you know
how to do, skilled motions.

In case of stumbling or similar unexpected motions, the recruitment
algorithm will always find the most optimal way of recruiting the muscles,
the analysis might work fine and predict muscles forces, but it may be
another criterion that should be used for determining the muscle forces. The
recruitment in a stumbling situation will most likely not be determined from
a minimum fatigue criterion.



I am curious, how do you model the force acting on the knee, when it hits
the ground?



So what can you to alter the recruitment criteria? The recruitment solver’s
gives you some flexibility to changed the criteria it can add a linear or
quadratic penalty to the objective function, if you are not familiar with
these penalties please have a look at this tutorial
http://www.anybodytech.com/index.php?id=691

If you add a relative large linear penalty it will favour the strong
muscles, I do not know if this is what happens when stumbling.? If you know
that certain muscles are very active you can also prescribe a certain force
on these muscles and then let the recruitment resolve the rest of the
muscles, to get equilibrium. Of course both of these options are “hacks” but
these are the best ideas I have right now.



Best regards

Søren, AnyBody Support





_____

From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of olepeterson
Sent: 18 February 2009 12:16
To: anyscript@yahoogroups.com
Subject: [AnyScript] AnyBody and stumbeling humans



Hi there!

I am doing some research on humans that stumble and fall. The focus in
this work are the impact forces when eg. the knee hits the ground and
how these forces spread through the limbs. I recorded motion data with
a Vicon MOCAP system and force data with two forceplates and try to
evaluate them in Anybody afterwards.
I have build a suitable model based on GaitUniMiami and adopted it to
my needs with the much appreciated help of this group. After a look at
the first results of the inverse dynamics analysis, I wonder how
reliable the results of muscle forces in AnyBody are. The problem in
this case, I guess, is that the human body tries to work against the
movement of falling, while AnyBody activates the muscles in order to
conduct the movement that is given to it by the MOCAP data.
Can this perception be correct and, if yes, is there a way to work
around this general issue?

Cheers,
Michael





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

  #3  
Old 03-05-2009, 12:21 PM
olepeterson
Guest
 
Posts: n/a
Default Re: AnyBody and stumbeling humans

Hi Soren and thanks for your quick and substantial response.

Unfortunately I have not been able to investigate on changing the recruitment
criteria yet, but it is scheduled for beginning of next week.

The measured forces of the impact are applied to the knee by

AnyReacForce PlateFootReaction={
AnyKinLinear Lin={
Ref=1;
AnySeg &ref1=..ForcePlate;
AnyRefNode &ref2=....HumanModel.Left.Leg.Seg.LowerThigh.KneeJ oint;
};
AnyKinRotational Rot={
// Ref=1;
AnySeg &ref1=..ForcePlate;
AnyRefNode &ref2=....HumanModel.Left.Leg.Seg.LowerThigh.KneeJ oint;
Type=RotAxesAngles;
};
};

The Node KneeJoint is just for first investigations, it is on the list for being
changed to a more suitable point near the patella.

Currently I am trying to include a full body marker model in the GaitVaughan
application, in order to compare results from GaitUniMiami, the GaitVaughan with
lower body markers and GaitVaughan with full body markers. While working through
the repository, i came across the definition of the glenohumeral joint:

///Definition of GH joint
AnyKinRotational GHRot = {
AnyRefNode &scapula_gh = ..Seg.Scapula.gh;
AnyRefNode &humerus_gh = ..Seg.Humerus.gh;
Type = RotAxesAngles;
Axis1 = y;
Axis2 = z;
Axis3 = y;

};

and I wondered why there are only 2 independent axis defined. Or am I looking at
the wrong spot for the joint definition?

Cheers,
Michael

--- In anyscript@yahoogroups.com, AnyScript Support <anyscriptsup@...> wrote:
>
> Hi Michael
>
>
>
> Sounds as nice model
>
>
>
> Usually we say that the recruitment algorithm will work for motions you know
> how to do, skilled motions.
>
> In case of stumbling or similar unexpected motions, the recruitment
> algorithm will always find the most optimal way of recruiting the muscles,
> the analysis might work fine and predict muscles forces, but it may be
> another criterion that should be used for determining the muscle forces. The
> recruitment in a stumbling situation will most likely not be determined from
> a minimum fatigue criterion.
>
>
>
> I am curious, how do you model the force acting on the knee, when it hits
> the ground?
>
>
>
> So what can you to alter the recruitment criteria? The recruitment solver's
> gives you some flexibility to changed the criteria it can add a linear or
> quadratic penalty to the objective function, if you are not familiar with
> these penalties please have a look at this tutorial
> http://www.anybodytech.com/index.php?id=691
>
> If you add a relative large linear penalty it will favour the strong
> muscles, I do not know if this is what happens when stumbling.? If you know
> that certain muscles are very active you can also prescribe a certain force
> on these muscles and then let the recruitment resolve the rest of the
> muscles, to get equilibrium. Of course both of these options are "hacks" but
> these are the best ideas I have right now.
>
>
>
> Best regards
>
> Søren, AnyBody Support
>
>
>
>
>
> _____
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
> Of olepeterson
> Sent: 18 February 2009 12:16
> To: anyscript@yahoogroups.com
> Subject: [AnyScript] AnyBody and stumbeling humans
>
>
>
> Hi there!
>
> I am doing some research on humans that stumble and fall. The focus in
> this work are the impact forces when eg. the knee hits the ground and
> how these forces spread through the limbs. I recorded motion data with
> a Vicon MOCAP system and force data with two forceplates and try to
> evaluate them in Anybody afterwards.
> I have build a suitable model based on GaitUniMiami and adopted it to
> my needs with the much appreciated help of this group. After a look at
> the first results of the inverse dynamics analysis, I wonder how
> reliable the results of muscle forces in AnyBody are. The problem in
> this case, I guess, is that the human body tries to work against the
> movement of falling, while AnyBody activates the muscles in order to
> conduct the movement that is given to it by the MOCAP data.
> Can this perception be correct and, if yes, is there a way to work
> around this general issue?
>
> Cheers,
> Michael
>
>
>
>
>
> [Non-text portions of this message have been removed]
>

  #4  
Old 03-10-2009, 01:53 AM
AnyBody Support
Guest
 
Posts: n/a
Default Re: AnyBody and stumbeling humans

Hi Michael

It looks fine the way you apply the force to the knee, as long you ensure that
the plate has no reactions to environment and no force applied when the knee is
not in contact.
Do you have a separate plate for the knee, or do you use the same one for the
left foot? If you use the same the plate it will not work, unless you have
conditional contact defined.

The GH joint is defined according to this article:

Journal of Biomechanics 38 (2005) 981–992
ISB recommendation on definitions of joint coordinate systems of various joints
for the reporting of human Joint motion—Part II: shoulder, elbow, wrist and hand

Here the shoulder joint are listed to have the sequence listed in the model. If
you would like to have another sequence you can just create an additional
rotational measure on the same two node an this for driving the model. Actually
this is what is done in the interface.any file you will find in all body parts.
Here are new rotational measured defined that makes it possible to make the
right and left side kinematics symmetric, it is these measures that we usually
drive the rotations of in the applications, not the rotations of the joint
itself.

Hope it made it clearer

Best regards
Søren, AnyBody Support



--- In anyscript@yahoogroups.com, "olepeterson" <ole.peterson@...> wrote:
>
> Hi Soren and thanks for your quick and substantial response.
>
> Unfortunately I have not been able to investigate on changing the recruitment
criteria yet, but it is scheduled for beginning of next week.
>
> The measured forces of the impact are applied to the knee by
>
> AnyReacForce PlateFootReaction={
> AnyKinLinear Lin={
> Ref=1;
> AnySeg &ref1=..ForcePlate;
> AnyRefNode &ref2=....HumanModel.Left.Leg.Seg.LowerThigh.KneeJ oint;
> };
> AnyKinRotational Rot={
> // Ref=1;
> AnySeg &ref1=..ForcePlate;
> AnyRefNode &ref2=....HumanModel.Left.Leg.Seg.LowerThigh.KneeJ oint;
> Type=RotAxesAngles;
> };
> };
>
> The Node KneeJoint is just for first investigations, it is on the list for
being changed to a more suitable point near the patella.
>
> Currently I am trying to include a full body marker model in the GaitVaughan
application, in order to compare results from GaitUniMiami, the GaitVaughan with
lower body markers and GaitVaughan with full body markers. While working through
the repository, i came across the definition of the glenohumeral joint:
>
> ///Definition of GH joint
> AnyKinRotational GHRot = {
> AnyRefNode &scapula_gh = ..Seg.Scapula.gh;
> AnyRefNode &humerus_gh = ..Seg.Humerus.gh;
> Type = RotAxesAngles;
> Axis1 = y;
> Axis2 = z;
> Axis3 = y;
>
> };
>
> and I wondered why there are only 2 independent axis defined. Or am I looking
at the wrong spot for the joint definition?
>
> Cheers,
> Michael
>
> --- In anyscript@yahoogroups.com, AnyScript Support <anyscriptsup@> wrote:
> >
> > Hi Michael
> >
> >
> >
> > Sounds as nice model
> >
> >
> >
> > Usually we say that the recruitment algorithm will work for motions you know
> > how to do, skilled motions.
> >
> > In case of stumbling or similar unexpected motions, the recruitment
> > algorithm will always find the most optimal way of recruiting the muscles,
> > the analysis might work fine and predict muscles forces, but it may be
> > another criterion that should be used for determining the muscle forces. The
> > recruitment in a stumbling situation will most likely not be determined from
> > a minimum fatigue criterion.
> >
> >
> >
> > I am curious, how do you model the force acting on the knee, when it hits
> > the ground?
> >
> >
> >
> > So what can you to alter the recruitment criteria? The recruitment solver's
> > gives you some flexibility to changed the criteria it can add a linear or
> > quadratic penalty to the objective function, if you are not familiar with
> > these penalties please have a look at this tutorial
> > http://www.anybodytech.com/index.php?id=691
> >
> > If you add a relative large linear penalty it will favour the strong
> > muscles, I do not know if this is what happens when stumbling.? If you know
> > that certain muscles are very active you can also prescribe a certain force
> > on these muscles and then let the recruitment resolve the rest of the
> > muscles, to get equilibrium. Of course both of these options are "hacks" but
> > these are the best ideas I have right now.
> >
> >
> >
> > Best regards
> >
> > Søren, AnyBody Support
> >
> >
> >
> >
> >
> > _____
> >
> > From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
> > Of olepeterson
> > Sent: 18 February 2009 12:16
> > To: anyscript@yahoogroups.com
> > Subject: [AnyScript] AnyBody and stumbeling humans
> >
> >
> >
> > Hi there!
> >
> > I am doing some research on humans that stumble and fall. The focus in
> > this work are the impact forces when eg. the knee hits the ground and
> > how these forces spread through the limbs. I recorded motion data with
> > a Vicon MOCAP system and force data with two forceplates and try to
> > evaluate them in Anybody afterwards.
> > I have build a suitable model based on GaitUniMiami and adopted it to
> > my needs with the much appreciated help of this group. After a look at
> > the first results of the inverse dynamics analysis, I wonder how
> > reliable the results of muscle forces in AnyBody are. The problem in
> > this case, I guess, is that the human body tries to work against the
> > movement of falling, while AnyBody activates the muscles in order to
> > conduct the movement that is given to it by the MOCAP data.
> > Can this perception be correct and, if yes, is there a way to work
> > around this general issue?
> >
> > Cheers,
> > Michael
> >
> >
> >
> >
> >
> > [Non-text portions of this message have been removed]
> >
>

 

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +2. The time now is 02:25 AM.