AMS 6.1--Operation

#1

Hello,
I am wondering about the difference between the KinematicStudyForParameterIdentification and MotionOptimization In MoCapModel. They have the same contents below, both containing InitialConditons, Kinematics and InverseDynamics. But I can see the difference once I do their Kinematics. In MotionOptimization, the study has 1749 steps while the other has only 249 steps.
I am truely eager to know which I should choose If I want to study the muscles, joint angle and etc.
Thanks!!
Lydian
Operation|311x500

#2

Hi @Lyd

The two studies are more or less identical. Computational speed is the main reason to use a separate kinematic study with fewer time steps for the parameter-identification step.

If we had used the full resolution kinematic study for parameter-identification it would have taken a very long time to run.

Also, the names of the studies are a little misleading in the old repositories. The actual flow looks something like this (from the new AnyMoCap documentation):

The KinematicStudyForParameterIdentification is used as part of the "Parameter-identification` block. The MotionOptimization study is now called “Marker tracking” in the figure.

I am truly eager to know which I should choose If I want to study the muscles, joint angle and etc.
Thanks!!

You could use the MotionOptimization (MarkerTracking in AMMR 2) if you want to investigate joint angles as a function of time. If you also want to look at more stuff, like muscles, forces, accelerations etc. then you also need to run the inverse dynamics step (The last step in the figure).

Best wishes,
Morten

#3

Hello,
As the study in Marker tracking costs a lot of time. May I just run the Kinematics and Inverse Dynamics in KinematicStudyForParameterIdentification instead of Marker tracking? That is to say that I ignore the Marker tracking. Thank you
BW
Lydian

#4

Hi Lydian,

Short answer is no, the parameteridentification study runs on a “downsampled” study, usually it only do analysis of 25% of the frames or something like this. nStep is simply divided by a factor, so if you run kinematics of this study you will not get all the frames, secondly the model is not setup to store the kinematics into the needed files if you did that.

Best regards
Søren

#5

Hello,

Thank you for your reply, I will consider and try to adjust my model to make the study work successfully.

Best Wishes
Lydian