Body Mass, Inertia & Mass Center Data

[Priyanshu Agarwal]

Hi,

The Upper Extremity Model available for download has no mass, inertia and mass center data which is needed for Inverse Dynamics analysis. From where are the users expected to collect all the data for the bodies. The model has so many small bodies that obtaining the data seems virtually impossible.

Thanks.

[Ayman Habib]

Hi Priyanshu,

Since these models are not part of OpenSim I'd suggest you contact the model author(s) for information regarding this issue.

Good luck,
-Ayman

[Priyanshu Agarwal]

Hi Ayman,

Is there any model offered by OpenSim team that has all the data (including the human upper limbs) available in it?

Thanks.

[Marc Carmichael]

HI,

I had the same problem. I obtained approximate mass/inertia data for the upper limb from two sources. One was David Winter's "Biomechanics and Motor Control of Human Movement". This book has equations for estimating the lengths, mass, and centre of gravity for each segment in the upper body.

For the inertial properties, I found a report "Investigation of Inertial Properties of the Human Body" by the US Department of Transportation. From that I got measured inertial properties of the upper arm, forearm and hand.

I then applied these mass, inertia and COG properties to the humerus, radius, and lunate bodies.

I then tried to perform inverse dynamics, but I was still getting issues with some free bodies not having inertia matrix defined (I'm assuming it was the free bodies in the hand). So I wrote a program to cycle through each body in the model and give it an infinitesimally small mass and inertia, just to satisfy the ID. After this the ID worked.

Hope this helps

[Priyanshu Agarwal]

Thanks Marc. It was really helpful.

[Priyanshu Agarwal]

Hi Marc,

While updating the model for inertia and mass center properties how did you understand the axes orientation of the body. The GUI doesn't explicitly mention the X,Y and Z axis on the coordinate frame attached with the body. Also, the center of mass is not graphically displayed in the GUI which makes it difficult to understand where it is located.

Thanks.

[Marc Carmichael]

Hi Priyanshu,

I can't remember exactly, but something along the lines of clicking on the body in the navigator and selecting "show axes". It will show a color coded x-y-z set of axis. I can't remember which color stands for which axis, but you can easily work this out by showing the axes of the "ground" body and taking note of the colors that represent x, y, and z.

The origin of each body I think is usually at the proximal end of the body, so I just set the COG to be offset along one of the axes at some distance I calculated from "Winter". I also added the Ixx Iyy and Izz mass moments of inertia about the appropriate axes.

I have not actually validated my ID results yet, but so far the results seem realistic. Let me know if you find otherwise.

Cheers

[Priyanshu Agarwal]

Hi Marc,

I too thought the same. However, it comes out that the axes color codes are also not the same. The global axes have RGB color code whereas the body fixed axes have RGY color code. So, did you consider Y color axis as B, because if it stands for a different axis then the whole analysis becomes useless.
Also, did you keep the muscle parameters same. Actually I was trying to cross-verify my results. I previously did an analysis using "Anybody Software". However, it seems that the muscle forces are way too high using OpenSim.

Thanks.

[Marc Carmichael]

Hi Priyanshu,

Yeah I had the same issue with the axis colours. That's why I selected the "Show Axes" of the ground body. Since it has zero rotation in the ground coordinate system (you can verify this by looking at the body's information) then its coloured axes are aligned with the global x,y,z axes. That's how I determined which colours represent which axis. Unfortunately I can't remember them off the top of my head.

I kept the muscle parameters the same, but I am not sure how accurate they are. In the future I am hoping to perform simple simulations and compare the results to published data to verify them.

[Priyanshu Agarwal]

Thanks for the help Marc.