Re: CCD

Dave Eberly wrote:
"David Kelly" <DavidKelly@xxxxxxxxxxxx> wrote in message news:vmg2m15t6g36lq4cpgnl0vt9a38aeu3jae@xxxxxxxxxx


Is CCD able to deal with goal orientation also? 


CCD is an iterative algorithm which processes a joint at a time, and
one of six translation or six rotation components at a time.  Each
adjustment of a component is designed to minimize the distance from
the end effector to the goal.  When considering only the positional goal,
each adjustment has a simple mathematical formulation which can be
solved in closed form for the update (ignoring any joint constraints).
You can include aligning the end effector orientation with the goal
orientation, but then the function to optimize must be a combination
of distance to goal and variation from orientation of goal.  I suspect
that you will not have a closed form expression for the optimum value,
so you have yet another iterative numerical optimization step.  Perhaps
you can apply a fixed number of gradient-descent or conjugate-gradient
steps for this phase.

--
Dave Eberly
http://www.geometrictools.com



Optimum solution for the basic CCD step is possible, even with rotational DOF and constraints.
You can have a look at my IK thesis http://www.gameprog.it/?resource=12
sorry just in italian :)
But you can have a look at formulas, i wish it can help you.

Turbo - www.amanith.org
.