[ODE] Connected spheres creating instability against static o bjects?

Gopi Prashanth gprashanth at Heavy-Iron.com
Wed Apr 13 15:23:22 MST 2005

This problem is classic, basically upon collision new joints are created to
de-penetrae the spheres, now these joints and the original universal joints
end up competing for stable state(minimum energy configuration - in the
solver). This is what is causing the jitter. The best way to handle this is
to handle ragdoll collisions different from any other collision. Instead of
creating de-penetrating joints upon collision with the world just try and
apply forces. I think this would work, the more experienced gurus would be
able to guide you better and please correct me if I'm wrong.


-----Original Message-----
From: Rafael Horbach [mailto:rafael at southlogic.com]
Sent: Wednesday, April 13, 2005 3:10 PM
To: ODE at q12.org
Subject: [ODE] Connected spheres creating instability against static

Hello there. I'm using ode in a simulation where I have some static 
geometries as the scenery, and with a ragdoll model where there is a 
larger object that is the parent of some (4, on this example) lists of 
connected spheres... it's like a head where you have 4 "hairs" that 
consists, each one, of 5 connected spheres using universal joints. It's 
something like the ascii art below:
|     |
|     |
.     .
@     @
.     .
@     @
.     .
@     @

On the example, the '@'s are the spheres, and the dots are universal 
joints... Pretty simple.

The main object is controlled by the application, and is not involved in 
the physics simulation (I've made some small changes in ODE to allow 
this). Everything works fine, all smooth and so on, as long as the 
spheres have a minimum mass (a small mass causes instability, the 
simulation starts to tremble).

My problem happens when the spheres collide with static geometries, that 
have body 0. When this happens, the system becomes instable, trembling a 
lot, like in the new ascii below, where the left spheres are stable, and 
the right ones react accoarding with a box, but trembling :(
|     |
|     |
.     .
@     @ ________
.     ./       /
@    @/       /
.   ./       /
@  @/_______/

I've tried a lot of things to make it more stable, like changing 
erp/cfm, joints stop erp/cfm, removing joints axis constraints, changing 
the mass, etc... the big object is completely stable, there's no 
movement in the scene at all, except for the trembling on the little 
spheres. Running out of ideas here, so if anyone has a hint :)

I hope that I've made a good description of the problem, if there's more 
info that I can give to make easier for someone to figure out what's the 
problem, please ask. Thanks a lot.

Best Regards,

ODE mailing list
ODE at q12.org

More information about the ODE mailing list