Page 1 of 1

Inverse Kinematics that only uses end-effector position.

Posted: Tue Nov 28, 2017 5:09 pm
by pranavm
Hi Devs!

I am very new to OpenSim, and I was wondering if it is possible to get joint space configuration for the StanfordVA UpperExtrimity model upon specifying just the end-effector (say finger) position trajectories? (i.e, [x,y,z,t] of end-effector to elv_ang, shoulder_rot, shoulder_elv,...)? I see that it is possible in .trc files to specify recorded marker trajectories and optionally coordinates of the joints, but is it possible to do IK given a single marker trajectory ? If yes, how can I get started to generate these fake end-effector marker trajectories?

Any help is much appreciated!

Re: Inverse Kinematics that only uses end-effector position.

Posted: Tue Nov 28, 2017 7:31 pm
by hkimpara
Hello Pranav,

I have a few experience with this. Although you are willing to create arm posture with single end-effector position, you may need a kind of constraints to assume your desired postures because of redundancy.

I have two ideas with OpenSim's IK tool.
1) Virtual markers
Since OpenSim's IK tool has weight ratios for all markers, you can place virtual markers with relatively weaken-weight ratios.
2) Coordinate value
You can directly set a baseline posture using Coordinate value (joint angle). Then you can set relatively weaken-weight ratios for joint constraints.

I hope these would become your help. But if you have more appropriate hypothesis, I think it would be better to seek another method instead of above-mentioned comments.

Thank you, Hide

Re: Inverse Kinematics that only uses end-effector position.

Posted: Thu Nov 30, 2017 3:23 am
by mitkof6
Hi,

You can prescribe the movement in task space (the position of the end effector) and perform a muscle driven simulation or constrained inverse kinematics.

https://simtk.org/projects/task-space
https://github.com/mitkof6/opensim-task-space

Best