Page 1 of 1

Marker Errors Higher Than Expected

Posted: Fri Jul 30, 2021 2:13 pm
by hicksbrody
Does anyone have any experience performing Inverse Kinematics and reporting marker errors for high-impact motions such as sprinting, cutting i.e. rapid changes of direction, or other athletic movements? If so, do you have a ballpark of what you expect for model marker error?

The simtk documentation (https://simtk-confluence.stanford.edu:8 ... Kinematics) states that RMSE under 2 cm and max marker error of 2-4 cm are achievable, but that these guidelines vary depending on the motion and the model being used.

I am generally within the acceptable range for root mean square error, but I am noticing a couple markers whose max marker error is higher than the generally acceptable 2-4 cm. I am trying to figure out if I need to revisit my model or my experimental data. Any information you could provide would be helpful.

Thank you,

Brody

Re: Marker Errors Higher Than Expected

Posted: Mon Aug 02, 2021 1:27 pm
by aymanh
Hello Brody,

Since Inverse Kinematics is solved frame by frame, and it only uses kinematics (does not use the dynamics), the speed of the motion is not relevant as long as your sampling rate is good enough to capture the motion of interest. The effect of high impact/speed will show, however, in soft tissue artifacts and markers moving/sliding off their bony landmark attachment points so you maybe able to relax the guidelines as they were developed primarily for walking/gait scenarios.

Hope this helps,
-Ayman