There are proâ€™s and conâ€™s for both methods, I will try list them as I see
them. In my answer I assume that the computed angles are also based on the
The advantage of using the markers vs. the joint angles, is that you do not
need to think about the rotations in 3D. If you use the joint angles you
need to ensure compatibility between the way the joint angles are calculated
and the way they are used in the model (type of the rotational measure,
sequence of rotations). Typically I think such joints angles are calculated
using models which calculate three rotations between each segment, and the
joint centers may not be at a fixed position wrt. the segment. So in
principle the segment changes its length in time. When such angles are
imposed on a model which has for example a hinge knee joint and fixed
segment length, it will give some differences in the motion, this is
unavoidable. So for example in a gait model this may be seen as the foot
penetrating or sliding on the floor.
The advantage of using the joint angles are that the model becomes
kinematically robust since it is an open kinematic chain, in contrast to the
marker driven model.
The drawback of the marker driven models as the GaitModel, is that it can be
difficult to drive the model since you have to select which d.o.f. to drive
from each markers, this can be complicated sometimes. It also needs a good
guess on the starting positions of the segments in order to solve
kinematically, this is not needed for a joint driven model.
So it is a matter of taste I guess, which of the two solutions to make use
If you are looking for an example of a marker driven model of the arms, you
may want to look at the wheelchair model named “WheelChairRancho”, it drives
the arms and the upperbody by marker data.
SÃ¸ren, AnyBody Support
From: email@example.com [mailto:firstname.lastname@example.org] On Behalf
Sent: 02 January 2008 17:33
Subject: [AnyScript] Marker vs angle driven models
Are there any advantages of using data from motion capture markers
versus computed angles, or vice versa, for driving simulations?
I would like to be able to drive an arm model using measured data,
either markers or angles. I have looked through many of the posts but
have not found any best or preferred way. I know there is a 3D gait
example for driving a model with marker data, but are there any simple
arm models using an example set of marker data?
Thank you for your time,
[Non-text portions of this message have been removed]