Re: Planar Hexapod Kinematics

[Till Franitza <till.franitza-at-isw.uni-stuttgart.de>]

Hallo!
Search for the string EMC_NO_FORWARD_KINEMATICS in emcmot.c.
There is written:
-------------------snip--
*FIXME*
if (EMC_NO_FORWARD_KINEMATICS) {
/*fake it by setting actual pos to commanded pos*/
-----------snip-----------------
Here i left it like it was, though i actuallly had the forward trafo. I just
decided to save computation time by setting EMC_NO_FORWARD_KINEMATICS to 1 some
lines above that. So the genhexkons forward trafo is not called. When using
steppers i have to assume that actual pos=commanded pos.
There are still some fixes on the todo list. Or decisions to be made about how to
handle komplex kinematics. As far as i see through it, what I did is OK even for
reallly controlled drives. Forward trafo only needs to be called when switching
fron joints mode to axis mode. But this is not yet implemented correctly, at least
not by me. There should be a command to be used in the NC-Program and by the user
interface to switch a certain kinematic trafo on and off. It should be implemented
how to control the motors - by axis errors or by joint errors. This depends on how
fast everything runs. In my case it was not possible to run the forward trafo in
each control cycle. In other cases it might be and might be worth it from the
control point of view.

Till


Adriaan Pelzer wrote:

> Hi group and Till,
>
> I'm helping Lukas with the software side of the planar hexapod - we used
> trivkins initially to implement our transforms (cartesian to hexapod), but ran
> into a few troubles along the way. I just need to clear up a few things for
> myself:
>
> On 26-Mar-01 Till Franitza wrote:
> > Hi!
> > What I experienced with my Clavel-style kinematic is that I couldnt afford
> > the kinematics functions being called as often as in the case of trivkins.
> > For The genhexkins there was implementd a NO_FORWARD_TRAFO or something that
>
> Are the genhexkins functions called less often than the Trivkins? Sorry, my
> knowledge doesn't go that far (yet!). Where do you set NO_FORWARD_TRAFO (.run
> file or Makefile?), or is it set already?
>
> > prevents emcmot do calculate the actual TCP-Position out of the Motor
> > Positions in each cycle. I dont remember exactly. You should have a look at
> > that. Instead I think it is sufficient to assume the actual position is the
> > same as the commanded position. So in emcmot this is simply assigned. As far
> > as I understood it, emcmot made the control part in cartesian world
> > coordinates. In their simple case these were the same as the drive
>
> > coordinates. In the case of Parallelkinematics it is very demanding to make
> > the control (PID) in world coordinates and transform back into drive
> > coordinates. Many people would like to do that, because it would result in
> > better behaviour, but it is simply too time consuming if the forward trafo
> > is iterative. In your case, it isnt iterative, but still, methinks, in
> > emcmot there are some trafo things that can simply be omitted in order to
> > speed things up. Maybe your following errors have something to do with it.
>
> Are all of the above accomplished by setting NO_FORWARD_TRAFO, or is it still
> on the to do list?
>
>
> > Lukas du Plessis wrote:
> >
> >> Hi everybody
> >>
> >> We are experiencing problems with implementing the kinematics of our
> >> planar hexapod.  When using only the inverse kinematics, by passing
> >> EMCMOT_NO_FORWARD_KINEMATICS=1 to insmod when insmoding emcfreqmot.o, we
> >> encounter following- errors.  Our .ini file is set up as suggested by
> >> Ray to enforce 200 steps per mm (see below)
> >>
>
> ----------------------------------
> E-Mail: Adriaan Pelzer <adriaan-at-trispen.com>
> Date: 28-Mar-01
> Time: 08:00:10
>
> This message was sent by XFMail
> ----------------------------------

--
Dipl.-Ing. Till Franitza
Institute for Control Engineering of Machine Tools and
Manufacturing Units (ISW)
c.o. FISW
Rosenbergstr. 28
D- 70174 Stuttgart
Germany

phon: ++49 (0)711 22 99 2-28
fax:  ++49 (0)711 22 99 2-22
email: till.franitza-at-isw.uni-stuttgart.de
web:   www.isw.uni-stuttgart.de