I am interested in using motion capture data of the shoulder and arm
(marker locations from Optotrak) to drive a model in AnyBody. From
there we would like to extract muscle parameters at certain joint
angles, such as muscle lengths and muscle origin and insertions
location and velocities.
Do I need the joint angle position, velocity, and acceleration as a
function of time to drive the motion, or can I define nodes near
segments to simulate skin sensors over bony markers, and use the
kinematic data on these points?
You can indeed use markers for driving the model, however with certain
You can define a node (AnyRefNode) equivalent to your marker on a
segment (AnySeg). By a linear kinematic measure (AnyKinLinear) you can
measure the position of this node and by applying a driver, you can
drive this measure. This is in principle like it was joint angles as
joint angles are also just “kinematic measures” in AnyBody.
However, it may be somewhat more difficult to make the complete system
of constraints determinate. This is one restriction on the marker
approach. Like with joint angle or anything else you might come up
with as driven coordinates, the total number of independent
constraints from joints and drivers must match the degrees of freedom
of all the segments in the model.
Figuring out an independent set of marker drivers may be more much
difficult than joint angles, because you are dealing with points which
are good for defining translation whereas it is more complicated with
When we have used driver coordinates to drive a model, we have defined
an additional segment for each driver. The translation of this is
driven by the measured data and the rotation is driven to a constant,
e.g. zero. This gives you two representations of the marker, the
measured position and segmental-fixed position. You can now make
kinematic constraints between these two.
The power of this is that you can Ã¼se local (segment-fixed)
coordinates in the AnyKinLinear measuring the linear displacement
between the two marker representations. Local coordinates may help you
define the rotations consistently. Moreover it will typically also be
easier to select a suitable subset of the marker coordinates that you
want to drive.
I believe the Gait3D model in the repository uses this approach. There
should also be a presentation of how the marker coordinates have been
selected. (I am sorry tht I am not more precie in my reference here,
it has been a long time since I have looked there myself).
Let me know if you need further assistance.
— In email@example.com, nflieg@… wrote:
> I am interested in using motion capture data of the shoulder and arm
> (marker locations from Optotrak) to drive a model in AnyBody. From
> there we would like to extract muscle parameters at certain joint
> angles, such as muscle lengths and muscle origin and insertions
> location and velocities.
> Do I need the joint angle position, velocity, and acceleration as a
> function of time to drive the motion, or can I define nodes near
> segments to simulate skin sensors over bony markers, and use the
> kinematic data on these points?
> Nick Flieg