I'm relatively new to using OpenSim with experimental data I've collected myself.
I'm trying to use the Rajagopal model to do IK and ID analysis on the lower body during walking under different speed, incline, and exoskeleton conditions.
My question is: can I use the Rajagopal model with a lower-body marker set or do I need a full-body marker set to scale the model properly?
Rajagopal - lower body only?
-
Emily Klinkman
- Posts: 8
- Joined: Mon Sep 26, 2022 10:36 am
-
John Davis
- Posts: 56
- Joined: Mon Aug 26, 2019 7:42 am
Re: Rajagopal - lower body only?
I'm assuming by "lower body marker set" you mean "pelvis + both legs." If that's the case the Rajagopal model should work fine. Unlike the Gait2392 model it has no muscles that span the pelvis/torso joint, so you'll have no trouble there.
You can remove the arms and their torque actuators completely, that won't affect anything. You can also completely remove the torso too: the "residual" forces at the ground/pelvis joint will be very large and totally bogus, but those don't propagate out to the lower legs, so your joint moments in ID will be the same regardless of whether you had a torso or not in your model. This linked thread has some excellent info on why this is the case.
The only issue you may run into is how segment inertial properties are adjusted by the Scale Tool. Scaling can redistribute mass and inertial properties of the model and that might get wonky if the model doesn't have a torso, but you specify the mass of the whole body. You could either (a) leave the torso in for the scaling procedure only and put the "remainder" of the body mass there, scale the torso using the same length/width/depth scale factors as the pelvis, then delete the torso and its actuators before running IK and ID, or (b) adjust the model's inertial properties on your own after scaling. I have some MATLAB code for option (b) here that I can share if you'd like.
You can remove the arms and their torque actuators completely, that won't affect anything. You can also completely remove the torso too: the "residual" forces at the ground/pelvis joint will be very large and totally bogus, but those don't propagate out to the lower legs, so your joint moments in ID will be the same regardless of whether you had a torso or not in your model. This linked thread has some excellent info on why this is the case.
The only issue you may run into is how segment inertial properties are adjusted by the Scale Tool. Scaling can redistribute mass and inertial properties of the model and that might get wonky if the model doesn't have a torso, but you specify the mass of the whole body. You could either (a) leave the torso in for the scaling procedure only and put the "remainder" of the body mass there, scale the torso using the same length/width/depth scale factors as the pelvis, then delete the torso and its actuators before running IK and ID, or (b) adjust the model's inertial properties on your own after scaling. I have some MATLAB code for option (b) here that I can share if you'd like.
-
Mohammadreza Rezaie
- Posts: 379
- Joined: Fri Nov 24, 2017 12:48 am
Re: Rajagopal - lower body only?
Hi, If you already have data of markers attached on just pelvis and lower limb segments, you can remove trunk and upper limb bodies as well as their joints, actuators, and markers (simply edit the model in Notepad++). There is also a model named LaiArnold2017_refined.osim which may fit your current data:
https://simtk.org/frs/?group_id=1302
Regarding the mass distribution, my solution was to multiply the total mass by 0.464, the mass of pelvis + lower limb segments, and use this value in the Scale tool.
https://www.c-motion.com/v3dwiki/index.php/Segment_Mass
https://simtk.org/frs/?group_id=1302
Regarding the mass distribution, my solution was to multiply the total mass by 0.464, the mass of pelvis + lower limb segments, and use this value in the Scale tool.
https://www.c-motion.com/v3dwiki/index.php/Segment_Mass