Hello Everyone -
RRA recommended to decrease masses of all segments, leading to a loss of mass. Is that because gait2392 doesn't have arms? I really trust on total mass of my subject, and GRF data reflect dynamics of that mass. Should I adjust segment masses as RRA recommended, though? If I was RRA, I would increase mass of torso so that it would include masses of arms. Can I run RRa so that it recommends mass adjustments without adjusting total mass?
Thanks -
Rudi
Mass adjustments in RRA
-
Sam Hamner
- Posts: 58
- Joined: Tue Jun 26, 2007 5:03 pm
RE: Mass adjustments in RRA
Hi Rudi,
The mass change recommendation from RRA comes from errors in both kinematics (i.e., joint angles from IK) and modeling errors. Even though you total mass may be accurate, the mass distribution among segments of the model might not exactly represent that of your subject, so it is recommending a change. What is the suggested change in mass? If it is very large, it could also indicate that the external ground reaction forces are not being applied correctly (e.g., the COP location may be slightly off).
With that, gait2392 (and gait 2354) does account for the mass of the upper extremities, but just adds that mass to the torso (imagine wrapping your arms around your chest and shoulders). If you are simulating normal walking, I doubt the mass change recommendation is due to arms. But if you want to see for sure (and you collected data of arm motion) you could try my model with arms here: https://simtk.org/project/xml/downloads ... oup_id=516
Hope that helps,
Sam
The mass change recommendation from RRA comes from errors in both kinematics (i.e., joint angles from IK) and modeling errors. Even though you total mass may be accurate, the mass distribution among segments of the model might not exactly represent that of your subject, so it is recommending a change. What is the suggested change in mass? If it is very large, it could also indicate that the external ground reaction forces are not being applied correctly (e.g., the COP location may be slightly off).
With that, gait2392 (and gait 2354) does account for the mass of the upper extremities, but just adds that mass to the torso (imagine wrapping your arms around your chest and shoulders). If you are simulating normal walking, I doubt the mass change recommendation is due to arms. But if you want to see for sure (and you collected data of arm motion) you could try my model with arms here: https://simtk.org/project/xml/downloads ... oup_id=516
Hope that helps,
Sam
-
Rudolf Klemetti
- Posts: 48
- Joined: Thu Jul 15, 2010 3:09 pm
RE: Mass adjustments in RRA
Hi Sam -
The suggested change of mass was 1.59 kg (for one analyzed time interval). I think that's not too large - and it's less than mass of arms. If I understood right, an average vertical acceleration should be zero for a time interval RRA is employed. For the time interval above, it wasn't zero. Thus, perhaps one half of the recommended mass change can be exmplained by non-zero average vertical acceleration. I will try to find a time interval with exactly zero average acceleration. Once I have found one and adjusted masses, I can run RRA for intervals of non-zero average vertical acceleration to adjust kinematics, right???
Yes, I agree that mass distribution is not necessarily accurate, Nevertheless, total mass of my subject is based on GRF recordings while she was standing on the force plate. So, RRA can't have any better estimate for mass than GRF data. Therefore, it would be desirable to let RRA to adjust mass distribution while keeping total mass as a constant.
Thanks - that helped -
Rudi
The suggested change of mass was 1.59 kg (for one analyzed time interval). I think that's not too large - and it's less than mass of arms. If I understood right, an average vertical acceleration should be zero for a time interval RRA is employed. For the time interval above, it wasn't zero. Thus, perhaps one half of the recommended mass change can be exmplained by non-zero average vertical acceleration. I will try to find a time interval with exactly zero average acceleration. Once I have found one and adjusted masses, I can run RRA for intervals of non-zero average vertical acceleration to adjust kinematics, right???
Yes, I agree that mass distribution is not necessarily accurate, Nevertheless, total mass of my subject is based on GRF recordings while she was standing on the force plate. So, RRA can't have any better estimate for mass than GRF data. Therefore, it would be desirable to let RRA to adjust mass distribution while keeping total mass as a constant.
Thanks - that helped -
Rudi