[ODE] composite object split

Guillaume Jouffroy oxymore at tele2.fr
Tue Jul 31 07:30:28 MST 2007

The test_movingtrimesh example works very well at my home (the rabbits 
are falling like hell), even if it is true that the collision of rabbits 
after several of them become a bit odd. So I guess my trimesh should 
work correctly (I create a real trimesh actually with dCreateTrimesh).

The reason that my composite object is dancing (I guess) comes from the 
fact that its base is a sphere (and the body is at the center of this 
sphere). Then the trimesh is offset on the top of that, and finally a 
cylinder offset on top.
As I have read there is no friction for rolling stuffs so the ball is 
rolling, knocking the ground on the trimesh then rolling another way 
etc... I guess there is something to do with the center of masses of the 
composite object which is probably not at the center of the ball. But I 
don't see how (well if you think my guess is good).

Also, the thing is that all the example tests are nearly invisible for 
me, because my computer is too fast. I don't see anything in ode to 
reduce/regulate the framerate of the simulation (I have tried to change 
the quickstep to step, or to increase a lot the precision but the comp 
is still too fast). Thus a little constant move of a problem make things 
go like crazy.

2 more questions:
* Is there a way to fix a geom with a body, in the air to attach things 
to it, for debug purposes ? I can attach things to the ground with the 
fixed joint, but I cannot fix one in the air.
* Is there a way to prevent joints to break when 2 bodies fall on the 
ground ? It seems that when my objects fall on the ground they do it so 
violently !! Why ? Each part is only kg, and the fall from 0.5 height or 
so which is not supposed to be terrible. All my geoms are far less than 
1 in side.

Megan Fox wrote:
> How are you simulating the trimesh?
> Dynamic trimeshes are, in general, extremely unstable.  They're almost
> guaranteed to "dance" as you put it unless you design an extremely
> favorable (which usually means all but useless) world.
> Static trimeshes work fine, of course, but generally speaking, moving
> objects need to be constructed out of the basic primitives.
>> Could you please give me any clue of the reasons that may produce this
>> strange behavior ?
>> _______________________________________________
>> ODE mailing list
>> ODE at ode.org
>> http://ode.org/mailman/listinfo/ode

More information about the ODE mailing list