Dear Biomch-L Subscribers:

I am wondering if anyone has done or is aware of any work dealing with the
following problem (I personally call it error propagation problem) in
performing forward kinematics. In movement synthesis or simulation,
suppose Cartesian coordinates of the joints are being derived from segment
angles using recursive kinematics; even if comparable errors were made in
the angles (w.r.t observation), errors in joint coordinates would increase
along the direction of forward kinematics (e.g., proximal to distal). I
would like to find out if there exists a technique to circumvent this
problem, say to evenly distribute the errors.

Thanks for your attention. I will provide a summary of responses.

Xudong Zhang
The University of Michigan