Question regarding computeGyroscopicImpulseImplicit_World

[eastsky.kang]

Hi,

I am using bullet physics for my robotics simulator. Since I want to dynamically realistic simulation, I am applying gyroscopic impulse to objects.

I realized explicit gyroscopic impulse (forward euler apprx) is not stable and found great article from Mr. Erin Catto about implicit gyroscopic impulse equation.

http://box2d.org/files/GDC2015/ErinCatt ... ethods.pdf

But my question is, how "computeGyroscopicImpulseImplicit_World" is different with "computeGyroscopicImpulseImplicit_Body"

As I read the code, I guess computeGyroscopicImpulseImplicit_Body is implementation of Mr. Catto's approach,

Besides, computeGyroscopicImpulseImplicit_World seems like uses inertia tensor w.r.t world frame, which can lead to incorrect solution as Mr. Catto mentioned in his slides.


Can someone explain the differences between those two?