[ODE] Fwd: Ball and Socket or (Universal +Hinge)

Jaap Stolk jwstolk at gmail.com
Tue Apr 11 00:21:14 MST 2006


CC to list.

---------- Forwarded message ----------
From: Jaap Stolk <jwstolk at gmail.com>
Date: Apr 11, 2006 7:04 AM
Subject: Re: [ODE] Ball and Socket or (Universal +Hinge)
To: Jason Mallios <jmallios at cs.brown.edu>


On 4/10/06, Jason Mallios <jmallios at cs.brown.edu> wrote:
> In user mode, I believe you have to set both the axes and the angles every
> frame.  See section 7.3.8 of the documentation in the table describing
> dAMotorUser.  No one I know of has succesfully implemented reliable (meaning
> predictable) AMotors for 3 rotational degrees of freedom with stops in
> conjunction with a ball joint.  In this forum they have typically been used
> for 2-dof situations like keeping a capsule upright.  Also, the AMotor is
> quite good at keeping a joint fixed when all dParam**Stop* parameters are
> set to 0.0f.
>
> I worked on AMotors for several months and switched to a custom 3-dof joint
> that doesn't rely on ODE at all, so it doesn't have the benefit of seeing
> forward in time as does the constraint equation.  As I recall, my main
> issues with AMotors were maintaining orthogonality of the axes in either
> mode.  I remember a user mode ambiguity regarding which axis should be
> applied to which body (or the "global frame") and the euler mode habit of
> destroying orthogonality when rotating about the mysterious, automatically
> computed, a1 axis.  This included the UNSTABLE branch as of two or three
> months ago.
>
> You should search the archives for more info.

thank you. maybe i will use a combination of joints instead. the 3rd
axes can be anywhere along the second object, it does not need to be
within the same joint. I just thought that a ball joint would be more
efficient to simulate, but as it turns out, ODE is so fast already, i
don't think that will be a problem.



More information about the ODE mailing list