What do exactly getPrincipalMomentsOfInertia() gives?

Anonymous · ‎12-06-2019

In short,

1. getPrincipalMomentsOfInertia() didn't work as I expected. So what exactly it returns?

2. I have a model with "moments of inertia atcenter of mass" with ixy, izy, ixz != 0 (or saying far away). Is it a normal case? I thought it should be a diagonalized matrix.

----

I was working on grabbing diagonalize moments of inertia from the models, saying I want the diagonalized form of the MOI, getting ixx, iyy, izz with all other elements (ixy, iyz, ixz) equals to 0

The Properties UI will display MOI_world and MOI_CoM (Center of Mass), and I can export these two information by using getXYZMomentsOfInertia(), and calculating MOI_CoM from getXYZMomentsOfInertia and physicalProperties.centerOfMass().

However, sometimes the CoM is magically not have ixy, iyz, ixz equals to 0 when my model is complex. And the value not looks like caused by numerical error and it is relatively not small enough to close to 0.

And then I try to explore getPrincipalMomentsOfInertia(). At first from the description I think it will give the ixx, iyy, izz I want but it does not. And I am also not sure what exactly these values are.

# Get the moments of inertia about the principal axes. Unit for returned values is kg/cm^2.
(retVal,i1,i2,i3) = physicalProperties.getPrincipalMomentsOfInertia()

Here is an example with a Ball, Ring and Cylinder. Theoratically the Ball should have the diagonalize MOI with same value as shown in List of moments of Inertia Wiki

MOI_CoM displayed in UI looks reasonable, but the value of getPrincipalMomentsOfInertia() (In my document, it is the "MOI_principal", where I add another three 0.0 b.c. I thought it was the ixx,iyy,izz) is strange and I cannot tell what they are.

----

MoI about CoM with ixy, ixz, iyz != 0. (Sorry I cannot show you my model but only the Properties screenshot)

Is it a normal case?

It looks like a physical problem. Will I always get a diagonalized MoI about center of mass? I know MOI is diagnolized when it align with principal axes, but are they the same for the concept of "MoI at CoM"?

----

Notes:

dict_of_links["center_of_mass"]: Direct output from physicalProperties.centerOfMass

dict_of_links["inertia"] : The moment of inertia about center of mass. It is diplayed in UI but have no direct method to get this from API. I calculated it from the MOI about world frame and CoM.

dict_of_links["MOI_principal"]: Direct output from physicalProperties.getPrincipalMomentsOfInertia() and just add another three 0.0.

dict_of_links["radius_of_gyration"]: Direct output from physicalProperties.getRadiusOfGyration().

Ref
1. Get Physical Properties API Sample API Sample

2. PhysicalProperties.getPrincipalMomentsOfInertia Method

kandennti · ‎12-08-2019

Hi yl1242.

I do not understand the contents, but here physicalProperties_var.getXYZMomentsOfInertia ()
Some people point out that the documentation for is incorrect.
(However, it is Japanese)

http://syuntoku14.hatenablog.com/entry/2018/09/24/085723

(returnValue, xx, yy, zz, xy, yz, xz) = physicalProperties_var.getXYZMomentsOfInertia()

not

(returnValue, xx, yy, zz, xy, xz, yz) = physicalProperties_var.getXYZMomentsOfInertia()

Seems to be correct.

There may be these effects.

微妙な名前のブログ

サービスユーティリティ	エクスポート方法

Anonymous · ‎12-09-2019

Thank you for pointing that out 🙂

I am exactly working with syuntoku14's code. (https://github.com/syuntoku14/fusion2urdf)

I sent a pull request to his repo and thought it was a bug but looks like it's not after reading the post.

After testing, I think the issue you pointed out has already been fixed by Fusion.

So there should be some other staff happening here

kandennti · ‎12-10-2019

On the contrary, I'm sorry to confuse the story.

微妙な名前のブログ

サービスユーティリティ	エクスポート方法

Fusion