Re: [Algorithms] Rigid Body impulses

[<Pau...@sc...>]

On 15/03/2004 21:29:31 gdalgorithms-list-admin wrote: 

> If I dampen the angular momentum things get  considerably worse because 
it 
> relies
> on a very tiny amount of momentum to fall off a  single point in the 
first 
> place (i.e.
> collapse on to it's side), damping it results in it  endless hovering on 
one 
> vertex.
> Unless there is some trick to this? All I'm doing  is AngMom *= 0.999 
each 
> frame.

You don't need to dampen anything imo. I have whiten very stable impulse
based rigid body systems with no damping at all. True you might need to
implement some sort of sleeping algorithm to banish floating point errors
but this doesn't sound like its your problem. You should be able to get a 
system
without any damping or sleeping in which you can stably stack a pyramid of
28 cubes - just like one of the tokomak demos :-) (you have to change a 
compile
flag on the stacking demo to see their pyramid*).

> > When you  say it jumps from corner to corner what are you  describing?
> 
> It lands on one vertex. The collision impulses push  it upwards and it 
starts 
> to rotate,
> but its pushed a little too hard and rotates a  little bit too much so 
that 
> instead of
> falling on its side, it often lands on the next  vertex along, and then 
repeats.

That defiantly sounds like a bad inertia tensor to me. Inertia tensor of a 
box
is (equal sides):

ix = iy = iz = 1/6 * (M*L^2)

M = mass, L = side length (from 
http://maths.dur.ac.uk/~dma0cvj/mathphys/supplements/supplement2/supplement2.html)

At the risk of describing something which is obvious - you need to invert 
this and
store it, then once per frame you need to transform this to world space by

R = rotation matrix
RT = transposed rotation matrix

IW = inertia tensor in world space
IB = inertia tensor in body space

IW = R * IB * RT

Then when you come to compute angular acceleration

aa = IW * (impulse x contact lever)

Cheers, Paul.


*) Annoyingly you can't disable sleeping in tokamak, so it won't be a 
totally fair
comparison between your stuff, but it'll give you an idea.




**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
pos...@sc...

This footnote also confirms that this email message has been checked
for all known viruses.

**********************************************************************
 SCEE 2004