[ODE] Disabling axis of rotations on a body

[Bob Dowland]

> If you truly want to disable an axis of rotation you should 
> just manually
> set it after each step. For car wheels its important that 
> they are always in
> the right rotation, and i found no way other than forcing the 
> rotation to be
> correct.

My goodness! (No offence to Alex) I can see this working okay in the case of wheels but in general it would be a shame if this were the only way to go. In particular in a situ where it is possible to have a number of interactions between objects (dynamic and static) "manually correcting" things can be very dangerous, often leading to (at best) unphysical behaviour and (at worst) producing states that the integrators etc end up throwing up on.

I'm not sure if this is possible in ODE so sorry if this is a completely silly suggestion but... in some systems it is possible to make dynamic objects behave like static or kinematic objects (ie non-physically animated) by just giving them infinite mass (particularly easy when systems store reciprocal mass rather than mass itself). ie, in the linear case, (translational motion) where F/m = a, making 1/m = 0 produces a = 0. I'm wondering whether a similar effect for torques could be got by setting the appropriate coefficent of the rest-frame inertia matrix (which is hopefully diagonal) so that it's inverse say "inv_I" is essentially zero about the required axis, then wdot = inv_I.Ldot giving kth coord of wdot = zero whenever kkth coordinate of inv_I is zero.

Just a thought... :)


> -----Original Message-----
> From: Alex Hetu [mailto:alexhetu@videotron.ca]
> Sent: 23 July 2003 20:54
> To: Erwin de Vries; ode@q12.org
> Subject: RE: [ODE] Disabling axis of rotations on a body
> 
> 
> Hm. yes i was thinking about doing that, I guess it would also require
> reseting the torque too.... it's not hard obviously, and i 
> guess that's what
> i'll use for now, but it sure ain't clean. :)
> 
> Martin C.Martin suggested adding code to allow a AMotor joint 
> to support the
> static environment as body2. It's a very good idea i think. 
> But i wish ODE
> would provide a simple joint to do that though, or maybe just some
> parameters that could be set on the body maybe.
> 
> Thx guys
> 
> 
> -----Original Message-----
> From: ode-admin@q12.org [mailto:ode-admin@q12.org]On Behalf 
> Of Erwin de
> Vries
> Sent: Wednesday, July 09, 2003 3:18 PM
> To: ode@q12.org
> Subject: Re: [ODE] Disabling axis of rotations on a body
> 
> 
> If you truly want to disable an axis of rotation you should 
> just manually
> set it after each step. For car wheels its important that 
> they are always in
> the right rotation, and i found no way other than forcing the 
> rotation to be
> correct.
> 
> Erwin
> 
> ----- Original Message -----
> From: "Alex Hetu" <alexhetu@videotron.ca>
> To: <ode@q12.org>
> Sent: Wednesday, July 23, 2003 20:47
> Subject: [ODE] Disabling axis of rotations on a body
> 
> 
> > Hi,
> >
> > I'd like to know if it is possible to disable certain axis 
> of rotation on
> a
> > body and still move it around in the world and make it 
> collide with other
> > bodies without ever rotating around those disabled axis 
> (ex: a tall box or
> > cylinder that can only rotate on itself around the Y-axis 
> and which can
> > never fall on its sides...). Or maybe it could be possible 
> to actually
> > disable all rotations of a body?
> >
> > I looked for joints or constraints, but joints don't really apply on
> single
> > bodies... and i wish to find a CLEAN solution. Am I missing 
> something?
> >
> > thx!
> > Alex Hetu
> >
> > _______________________________________________
> > ODE mailing list
> > ODE@q12.org
> > http://q12.org/mailman/listinfo/ode
> >
> >
> 
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
> 
> _______________________________________________
> ODE mailing list
> ODE@q12.org
> http://q12.org/mailman/listinfo/ode
>