[ODE] RE: attn: Erin Catto, concerning: fixed path constraint

jnilson_99 at yahoo.com jnilson_99 at yahoo.com
Fri Jul 22 22:54:48 MST 2005


that seems a bit of a kludge. i'm at the point where i
can draw the vector from the upper arm to the plane.
that is i always know how far away from the plane i
am. 

i'm thinking there's got to be some way to use that
vector to find the correct ammount of force to bring
the arm back on plane in one time step. this is really
the same principle as any constraint. the equation to
lock to bodies together via joint assumes the anchors
must occupy the same point and calculates the forces
needed to do this. why can't i do the same thing?

john

--- Erin Catto <erincatto at sbcglobal.net> wrote:

> Since you have complete control over the simulation,
> you could just delete
> the universal and create a ball-and-socket joint at
> that transition point.
> 
> 
> -----Original Message-----
> From: jnilson_99 at yahoo.com
> [mailto:jnilson_99 at yahoo.com] 
> Sent: Friday, July 22, 2005 12:09 AM
> To: Erin Catto; 'ode'
> Subject: RE: [ODE] RE: attn: Erin Catto, concerning:
> fixed path constraint
> 
> 
> actually the more i think about using a universal,
> the
> more i think it won't work.
> 
> here's my problem, i only want the upper arm to
> follow
> a plane for a portion of the throw. around the point
> the ball is released i want to "release" the upper
> arm. that is i just want it to flow freely and
> follow
> it's own momentum. just like in a real pitcher's
> motion, he accelerates, throws and then releases.
> after the release his arm follows whatever path it
> wants.
> 
> if i were to use a universal joint here, after the
> release the arm would still follow a plane, it
> wouldn't look natural. after the release i need the
> upper arm to act like it does in reality, it moves
> based on the fact it's attached to a ball and socket
> joint.
> 
> john
> 
> --- Erin Catto <erincatto at sbcglobal.net> wrote:
> 
> > Yeah, just recreate the joint to get the required
> > plane of motion. The first
> > axis is perpendicular to the plane.
> > 
> > 
> > -----Original Message-----
> > From: jnilson_99 at yahoo.com
> > [mailto:jnilson_99 at yahoo.com] 
> > Sent: Thursday, July 21, 2005 11:15 PM
> > To: Erin Catto; 'ode'
> > Subject: RE: [ODE] RE: attn: Erin Catto,
> concerning:
> > fixed path constraint
> > 
> > 
> > ok, using a universal joint makes sense. my only
> > question would be: how do i allow the user to
> change
> > the plane of the upper arm? this would be
> necessary
> > to
> > change the type of throw,i.e. slider, side arm
> etc.
> > 
> > if "The first axis is the vertical axis", then to
> > change the plane at run time, do i just tweak the
> > direction of this vertical axis? 
> > 
> > thanks,
> > 
> > john
> > 
> > --- Erin Catto <erincatto at sbcglobal.net> wrote:
> > 
> > > Okay, this is clearer.
> > > 
> > > You have just two rotation axes:
> > > 
> > > 1. The first axis is the vertical axis.
> > > 2. The second axis is along the axis of the
> upper
> > > arm.
> > > 
> > > For this constraint you can use a universal
> joint.
> > > No ball joint or planar
> > > joint is needed.
> > > 
> > > Erin
> > > 
> > > -----Original Message-----
> > > From: jnilson_99 at yahoo.com
> > > [mailto:jnilson_99 at yahoo.com] 
> > > Sent: Thursday, July 21, 2005 9:03 PM
> > > To: Erin Catto; 'ode'
> > > Subject: RE: [ODE] RE: attn: Erin Catto,
> > concerning:
> > > fixed path constraint
> > > 
> > > "Perhaps you could explain the specific thing
> you
> > > are
> > > trying to model."
> > > 
> > > i'm trying to model various motions in sports,
> > like
> > > swinging, throwing, etc. all must use the
> > > ball/socket
> > > joint because that's the joint humans are using
> to
> > > accomplish this task.
> > > 
> > > take throwing a baseball for example, the upper
> > arm
> > > moves along a plane roughly parallel to the
> ground
> > > while rotating the lower arm from back to front.
> > > this
> > > motion requires a ball/socket joint because all
> > > three
> > > axis are being used. i could use a hinge to
> > restrict
> > > the upper arm to follow a plane, but then i
> > couldn't
> > > twist the lower arm.
> > > 
> > > i could use Russ's CMAC/Fox controller but for
> > > various
> > > reasons i don't want to do that. the biggest
> > reason
> > > being i want to allow the user to change the
> > motion
> > > of
> > > the shoulder in realtime without requiring
> > thousands
> > > of training steps. using mathematical
> constraints
> > > like
> > > requiring motion only along a plane i would be
> > able
> > > to
> > > change the motion of the shoulder and arm
> > > immediately.
> > > 
> > > 
> > > john
> > > 
> > > --- Erin Catto <erincatto at sbcglobal.net> wrote:
> > > 
> > > > I guess I don't understand what you're after.
> > > > Perhaps you could explain the
> > > > specific thing you are trying to model.
> > > > 
> > > > Erin
> > > > 
> > > > -----Original Message-----
> > > > From: jnilson_99 at yahoo.com
> > > > [mailto:jnilson_99 at yahoo.com] 
> > > > Sent: Wednesday, July 20, 2005 1:32 AM
> > > > To: Erin Catto; 'ode'
> > > > Subject: Re: [ODE] RE: attn: Erin Catto,
> > > concerning:
> > > > fixed path constraint
> > > > 
> > > > "You cannot base a planar joint off a ball
> joint
> > > > because a ball joint has 3 position
> constraints
> > > > while
> > > > a planar joint has only 2. "
> > > > 
> > > > maybe i'm incorrect in using the word
> "planar".
> > > > after
> > > > a few google searches, it seems planar joints
> > are
> > > in
> > > > fact a real joint type. by planar i simply
> mean
> > > > confining either a universal or ball/socket to
> > > > follow
> > > > a plane through the joint anchor. this would
> be
> > > > something like defining a spline and making an
> > arm
> > > > follow the spline's path.
> > > > 
> > > > i'm also a little confused on constraints. yes
> a
> > > > ball
> > > > and socket has 3 constraints per se. but it
> also
> > > has
> > > > "limit" constraints which confine the joint to
> > > > certain
> > > > angle ranges of motion,i.e. hi/lo stops.
> > > > 
> > > > is it possible to add my planar constraint
> like
> > an
> 
=== message truncated ===



More information about the ODE mailing list