I have been trying to drive the kinematics of the generic wheelchair
model via markers and a motion capture system. Initially, the system
was obviously overconstrained, but after commenting out most of the
initial drivers, I was able to get the model to run through without a
bug. However, I was unable to set the initial conditions and run both
kinematic and inverse dynamics analyses, because of an error message
“Kinematic analysis failed in time step 0.” There is also an
extensive list of constraints that are significantly above the error
tolerance.
Any suggestions as to what I am doing wrong? Any help would be
appreciated.
If you have not already done so, then this will be a really good time
to study the tutorial on kinematic analysis: http://www.anybodytech.com/75.0.html
Kinematics is a deceptively complex subject. Here’s a short
explanation of the root of the difficulties: Kinematics is all about
solving a system of nonlinear equations. This means that the first
thing to ensure is that you have the same number of equations as
degrees of freedom. You can check this with the ModelInformation
operation in the study. When you run this operation you will get a
whole lot of text output. If you scroll up, you will find the
following lines (separated by several other lines):
Total number of rigid-body d.o.f.: 78
Total number of constraints:
Joints: 46
Drivers: 29
Other: 3
Total: 78
The numbers in your model will be different, but the number of rigid-
body d.o.f. must correspond to the total number of constraints. If
this is not the case, then your system is either over- or under-
determinate.
However, it is not done just by having the correct number of
constraints. Since the system is nonlinear, lots of other stuff can
go wrong:
It may still not have a solution
It may have multiple solutions
It may not be able to reach the solution from the initial
configuration.
The best way to avoid all these problems is to start with a model
that works and then gradually replacing the old drivers with new
ones. So my proposal would be the following procedure:
Use the ModelInformation operation to check whether you have the
correct number of constraints. If not, correct the problem and try
again.
If you have the correct number of constraints, then go back to the
old model, replace the drivers a few at a time with your position
markers and make sure that the model works again for every change.
If you dispair completely, zip your model and upload it to the
files secion (just the application part of the model) and post a
request for somebody to have a look at it.
Best regards,
John
— In anyscript@yahoogroups.com, “Sarah R. Sullivan” <sarsulli@e…>
wrote:
>
>
> I have been trying to drive the kinematics of the generic wheelchair
> model via markers and a motion capture system. Initially, the
system
> was obviously overconstrained, but after commenting out most of the
> initial drivers, I was able to get the model to run through without
a
> bug. However, I was unable to set the initial conditions and run
both
> kinematic and inverse dynamics analyses, because of an error message
> “Kinematic analysis failed in time step 0.” There is also an
> extensive list of constraints that are significantly above the error
> tolerance.
>
> Any suggestions as to what I am doing wrong? Any help would be
> appreciated.
>
> Thanks so much!
> Sarah Sullivan