[ODE] Not A Number problems

Megan Fox shalinor at gmail.com
Mon Apr 25 13:43:57 MST 2005


Try increasing your world step size to something less likely to allow
interpenetration - like 100hz, 0.01.  At your step size, it's quite
likely your simulation is going to be unstable, which I suppose could
result in the entire thing detonating and throwing objects to NAN.

(I use a similar force scale, and tried increasing even to 0.016,
60hz, and found the results to be all but unusable - and you're
running at HALF that step size.  Unless you want to greatly scale your
forces down, you'll need to increase the step size to get the sim
stable.)

-Megan Fox

On 4/25/05, j <ode at dynamica.org> wrote:
> Hi all,
> 
> It seems I'm running into apparent limitations 
> to the complexity of my scene where 50 boxes 
> are pressed into the corner of an aquarium by
> the tilted gravity-vector. the boxes measure 
> some 1-10 units, the aquarium is 700 x 500 x 500
> and gravity is some <-50,-50,-50>. The
> boxes weigh all just 1 ( kg ), the CFM and
> ERP values for the dDOUBLE ode are
> dWorldSetCFM( world, 1.0e-9 );
> dWorldSetERP( world, 0.2 );
> and I update by
> dWorldStep( world, 0.03 );
> 
> All other settings are default.
> 
> Ode then comes with the following error
> "vector has zero size" in odemath.cpp,
> specifically in dNormalize4 because it 
> tries to normalize a body's quaternion
> which looks like:
> 
> - q 0x014ed278
>  [0] 1.#QNAN00000000
>  [1] -1.#IND000000000
>  [2] 1.#QNAN00000000
>  [3] -1.#IND000000000
> 
> any ideas on how to resolve this problem ?
> 
> thx,
> 
> J
> 
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode
> 
> 
>



More information about the ODE mailing list