Hinges and mass ratio.

[hsim]

Hi Erwin (and bullet users),

Is there a way to bypass the mass ratio limitation ( 1/10 approx) on rigid bodies connected by a Hinge constraint?
(I'm very happy with bullet but this point is really a limitation, since it is quite easy to have unstable hinges, and I really need to have large mass ratios between my rigid bodies)
One possible work around for my project would be to have this:
Let's imagine a robotic arm with "rigid" but modifiable hinges:
- several rigid bodies would be connected by hinges.
- Physics wouldn't affect hinge angles.
- Hinge angles could be piloted through the API.
- The whole system would be managed as a bigger rigid and articulated assembly. (a kind of rigidbody compound)

(from another point of view, it would be a kind of derformable compound shape )

Would this be possible?
(if somebody has a solution, I'm of course interested !!! )

Best regards,

Gerard,
H-SIM,
Paris.

[hsim]

any idea?

[hsim]

ERWIN, HELPPP !!!

[RBD]

Have you tried using a generic 6DOF constraint instead to see if it works better? (can be configured to behave exactly like a hinge)
I've had success with 10x mass ratio differences at the ends of chains of 6DOF constraints used as hinges and reasonable sim steps (real-time on good system).

Of course, more iterations or lower timestep help.

Otherwise, the current constraint solver may not be ideal for your needs (large mass ratio issue is known). I think there was talk of implementing other constraint solvers (i.e. issue 152), perhaps you could look into that, but that's not something Erwin's going to be able to resolve for you anytime soon. Although there certainly may be other tricks you could try...

[hsim]

Thanks RBD for your answer.
Yes, also tried with 6DOF. The thing is I need ratio much greater than 10. (for instance an aircraft and its articulated landing gear).
Yes I've already heard some stuff about other constraint solvers. Is there something existing?

Best regards.