Question about implementing the ground reaction force and about Reaction.Type

Hi AnyBody-People

In the meantime my AnyBody-model is walking' quite nicely (but perhaps still a bit groggily). My model consists now of one leg and the pelvis. I've modelled all three joints as spherical joints to be able to simulate also the small ab-/adduction and rotation angles in the knee and the ankle. However, only in the hip joint, the muscles contribute to the joint moment in all three directions. In the knee and the ankle, the muscles should only be responsible for the flexion/extension moment, whereas the forces in ab-/adduction and rotation direction should be carried by other structure and therefore be represented by the joint reaction force. Is it correct to do these by means of setting the Reaction.Type (in the AnyKinEq…Driver) to either 0 or 1? I'm not sure if I've understood theReaction.Type’ correctly: 0
means that the muscles have to carry this force, and 1 means that
it’s a force from the outside?
Thus, if I define the drivers for the joint angles, the Reaction.Type
for the hip will be {0,0,0} and for the knee and ankle {1,1,0}. Is
this correctly done?

My other question concerns the implementation of the ground reaction
force.
I’ve created the dummy segment according to the example in the Gait2
file; thus one segment for x-, y- and z-direction:

But the ground reaction force acts always on the same point at the
foot (for example at the node `Foot.Ground.Joint’) what is obviously
not realistic. How can I vary the point at the foot where the force
acts on over time? Do I have to create another dummy segment
connected to the foot via a prismatic joint and to drive this segment
afterwards according to the measured point of load incidence of the
ground reaction? Isn’t there an easier way?

I’ve always got an error message when I wanted to drive a spherical
joint by means of AnyKinEqInterPolDriver. I’ve defined all three
joint angles in the same text file [time abduction rotation flexion].
I’ve solved the problem by creation three drivers and three separate
text files. Isn’t it possible to define the three joint angles in a
single file?

Thanks for your help.
Christine

Hi Christine

It sounds great that your model is walking

I have answered your question below

— In anyscript@yahoogroups.com, “christine_schaerer”
<christine_schaerer@y…> wrote:
>
>
> Hi AnyBody-People
>
> In the meantime my AnyBody-model is `walking’ quite nicely (but

perhaps still a bit groggily). My model consists now of one leg
and
the pelvis. I’ve modelled all three joints as spherical joints to
be
able to simulate also the small ab-/adduction and rotation angles
in
the knee and the ankle. However, only in the hip joint, the
muscles
contribute to the joint moment in all three directions. In the
knee
and the ankle, the muscles should only be responsible for the
flexion/extension moment, whereas the forces in ab-/adduction and
rotation direction should be carried by other structure and
therefore
be represented by the joint reaction force. Is it correct to do
these
by means of setting the Reaction.Type (in the AnyKinEq…Driver) to
either 0 or 1?
I’m not sure if I’ve understood the `Reaction.Type’ correctly: 0
> means that the muscles have to carry this force, and 1 means that
> it’s a force from the outside?

Yes you are right 0 means that something else has to carry this load
ie. muscles or other drivers, 1 mean that the driver will carry the
load, this is equivalent to an infinitely strong engine.

> Thus, if I define the drivers for the joint angles, the
Reaction.Type
> for the hip will be {0,0,0} and for the knee and ankle {1,1,0}. Is
> this correctly done?

In principle this is correct, im not sure if your sequence (1,1,0)
is correct, this will depend on your joint definition

>
> My other question concerns the implementation of the ground
reaction
> force.
> I’ve created the dummy segment according to the example in the
Gait2
> file; thus one segment for x-, y- and z-direction:
>
> But the ground reaction force acts always on the same point at the
> foot (for example at the node `Foot.Ground.Joint’) what is
obviously
> not realistic. How can I vary the point at the foot where the
force
> acts on over time?

We have just created our own gait model, in this model we have a
dummy segment which is driven in the plane of the forceplate. The
position of this segment corresponds to the center of pressure on
the foot. The connection between the foot and the dummy segment is
made using an AnyReacForce which makes is possible to have reactions
between the foot and the dummy segment whitout having any kinematic
constraints. This means that the foot is just driven as always and
below the foot a dummy segment is driven to be loacated at the
center of pressure. This setup will automatically ensure that the
contact point between the foot and forceplate change in time.
This gait model will be part of the next respository release, i have
uploaded a small video of it to the file section.

Do I have to create another dummy segment
> connected to the foot via a prismatic joint and to drive this
segment
> afterwards according to the measured point of load incidence of
the
> ground reaction? Isn’t there an easier way?
>
> I’ve always got an error message when I wanted to drive a
spherical
> joint by means of AnyKinEqInterPolDriver. I’ve defined all three
> joint angles in the same text file [time abduction rotation
flexion].
> I’ve solved the problem by creation three drivers and three
separate
> text files. Isn’t it possible to define the three joint angles in
a
> single file?

Yes this is possible, i have uploaded a small example showing how to
do this to the file section “Driving3dof.zip” . This file also
contains an outcommented part which shows how to add the sliding
force. This part contains some facilities in AnyForce3D which will
be release in the next version of the system, but it can easily be
done in the current version as well. Please see the file “Gait2.Zip”
for details.

Best regards

AnyBody Support

>
> Thanks for your help.
> Christine

Dear Group Members

Those of you who have not yet seen the new gait model developed
recently by the AnyBody Research Group might want to take a look at the
recent newsletter:
http://www.anybodytech.com/37.0.html

and the associatated video animation:
http://www.anybodytech.com/fileadmin/examples/gait-large.avi

Please notice that the latter is 8.3 MB and may take a while to
download. The model very nicely illustrates how gait movement can be
driven accurtely and directly by marker data in three dimensions.

This model will be included in the upcoming update of the AnyScript
Model Repository.

Best regards,
AnyBody Support