pendulum test case

Archive AnyScript Yahoo archive
1

Hello,

I am trying out a test case similar to one of the tutorials where data from an
idealized
mocap marker was attached to a pendulum. The tutorial mentions that the data
for the
marker driving the pendulum was created from another source.

I have done the same thing using a separate piece of software. I created an
AnyBody
model in which I am driving the the pin joint of my pendulum with my outputted
angular
data, and I am keeping the Reaction.Type {On}. My hope is that after running
the inverse
dynamics, that reaction would be 0 (or very close).

Unfortunately, when I bring the position data into AnyBody with an
AnyFunInterpol
function, I can’t seem to get reasonable acceleration data to get the reaction
in the joint
motor to go to 0. I have tried downsampling, increasing the number of data
points, and
increasing the order of the bspline. I was about to start playing around with
some
different filters, but thought I would post the question here first.

I have uploaded my model and the data I have been playing around with in the
file section
(PendulumReactionForce.zip). Any help or guidance would be much appreciated.

Thank you in advance,
David

2

Hi David

Nice little model…

You are right about the acceleration data of the joint, they look rather
strange and not like a sinus shape you would expect from a pendulum.

It looks like that the accelerations has been made linear in sections of
about 0.4 sec or so, I do not think it will be possible to get reasonable
results with this acceleration. I do not think these kinks can come from the
interpolations in AnyBody, it must come from the other simulation tool you
are using.

We have tried to simulate the motion of the pendulum in Excel, and impose
the calculated motion onto the pendulum, this gives zero reaction in the
joint as it should.

The model has been uploaded as PendulumReactionForceV2.zip and it contains a
modified model with two identical pendulums, yours and a new one like yours,
but driven with data from an excel simulation. If you look at the joint
moment you will see that they are zero for the new pendulum driven by the
excel data. The new pendulum is not visible because Jii has been set to
zero, the simulations in excel does not take nonzero inertia into account.

Best regards

Søren, AnyBody Support

From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
Of dww805
Sent: 29 October 2008 23:37
To: anyscript@yahoogroups.com
Subject: [AnyScript] pendulum test case

Hello,

I am trying out a test case similar to one of the tutorials where data from
an idealized
mocap marker was attached to a pendulum. The tutorial mentions that the data
for the
marker driving the pendulum was created from another source.

I have done the same thing using a separate piece of software. I created an
AnyBody
model in which I am driving the the pin joint of my pendulum with my
outputted angular
data, and I am keeping the Reaction.Type {On}. My hope is that after running
the inverse
dynamics, that reaction would be 0 (or very close).

Unfortunately, when I bring the position data into AnyBody with an
AnyFunInterpol
function, I can’t seem to get reasonable acceleration data to get the
reaction in the joint
motor to go to 0. I have tried downsampling, increasing the number of data
points, and
increasing the order of the bspline. I was about to start playing around
with some
different filters, but thought I would post the question here first.

I have uploaded my model and the data I have been playing around with in the
file section
(PendulumReactionForce.zip). Any help or guidance would be much appreciated.

Thank you in advance,
David

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

3

Hi again,

Thanks for your quick reply and suggestions. I appreciate you going to the
trouble to prove and document your
solution. Your solution made me investigate my forward dynamics code a little
more because in that output, I was
getting smooth acceleration profiles. So obviously something didn’t line up.
Upon further investigation, I realized I
wasn’t setting the accuracy of my forward integration to a high enough value
such that when differentiating the position
output, I was getting the noisy acceleration data.

SO with all that said, now the model works with my data. My goal for this model
was/is to produce a very very simple
model that can be used as a homework problem where students are required to
‘find’ the length of the pendulum using
an optimization scheme in AnyBody. Now I can easily create the data I need to
drive the AnyBody model and I know it
will work.

Thanks again for your help!

Best,
David

— In anyscript@yahoogroups.com, AnyScript Support <anyscriptsup@…> wrote:
>
> Hi David
>
>
>
> Nice little model…
>
>
>
> You are right about the acceleration data of the joint, they look rather
> strange and not like a sinus shape you would expect from a pendulum.
>
> It looks like that the accelerations has been made linear in sections of
> about 0.4 sec or so, I do not think it will be possible to get reasonable
> results with this acceleration. I do not think these kinks can come from the
> interpolations in AnyBody, it must come from the other simulation tool you
> are using.
>
>
>
> We have tried to simulate the motion of the pendulum in Excel, and impose
> the calculated motion onto the pendulum, this gives zero reaction in the
> joint as it should.
>
>
>
> The model has been uploaded as PendulumReactionForceV2.zip and it contains a
> modified model with two identical pendulums, yours and a new one like yours,
> but driven with data from an excel simulation. If you look at the joint
> moment you will see that they are zero for the new pendulum driven by the
> excel data. The new pendulum is not visible because Jii has been set to
> zero, the simulations in excel does not take nonzero inertia into account.
>
>
>
> Best regards
>
> Søren, AnyBody Support
>
>
>
>
>
> _____
>
> From: anyscript@yahoogroups.com [mailto:anyscript@yahoogroups.com] On Behalf
> Of dww805
> Sent: 29 October 2008 23:37
> To: anyscript@yahoogroups.com
> Subject: [AnyScript] pendulum test case
>
>
>
> Hello,
>
> I am trying out a test case similar to one of the tutorials where data from
> an idealized
> mocap marker was attached to a pendulum. The tutorial mentions that the data
> for the
> marker driving the pendulum was created from another source.
>
> I have done the same thing using a separate piece of software. I created an
> AnyBody
> model in which I am driving the the pin joint of my pendulum with my
> outputted angular
> data, and I am keeping the Reaction.Type {On}. My hope is that after running
> the inverse
> dynamics, that reaction would be 0 (or very close).
>
> Unfortunately, when I bring the position data into AnyBody with an
> AnyFunInterpol
> function, I can’t seem to get reasonable acceleration data to get the
> reaction in the joint
> motor to go to 0. I have tried downsampling, increasing the number of data
> points, and
> increasing the order of the bspline. I was about to start playing around
> with some
> different filters, but thought I would post the question here first.
>
> I have uploaded my model and the data I have been playing around with in the
> file section
> (PendulumReactionForce.zip). Any help or guidance would be much appreciated.
>
> Thank you in advance,
> David
>
>
>
>
>
> [Non-text portions of this message have been removed]
>