Having read the paper, i think it might be possible to use such a framework
for achieving what I want. Such an approach would probably be superior
to the hand made, hard coded counterpart.

I would however need to actually use the code and see examples to be sure
about that. So please let us know when you are ready to leave it out to the
public, I am definitely interested.

I think usable and operating code is enough, even when it would be nice to
have tested and verified code. We all know that's not even the case
with most commercial code. With opensource we have the possibility to improve
and fix by ourselves, if needed.

Thanks

> On Wed, 11 Feb 2009, Bodmer Marc wrote:
> > The more I think about that "motion specification framework", the
> > more I think it might go in the direction we would finally need it.
> >
> > In our process we have several things which influence either the
> > geometry of the toolpath and/or motion parameters like for example
> > feed and/or process parameters like for example cutting power.
> >
> > Depending on the geometry we have to modify the toolpath:
> >  - inside, outside corners need special treatment (e.g. overshoot,
> > tool
> > tilt)
> >    depending on the needed quality. -> adaptive geometry
> >
> > Depending on the geometry we have to modify process parameters:
> >  - stright lines need less cutting power and higher feeds than tight
> > curves -> adaptive feeds and cutting power
> >
> > Depending on user parameters we have to modify the toolpath or other
> > parameters:
> >  - The user can (dynamically) go and choose to add "microjoints".
> > This holds the machined part to the plate it was cut from.
> >
> >  - The user can (dynamically) go and change feeds
> >  - The user can (dynamically) go and change the tool radius
> >
> > The final geometry of a part is fixed, the real toolpath to achieve
> > this part with a certain quality however is/can be process and
> >
> > settings dependent. There are a bunch of rules which describe these
> > dependencies.
> >
> > Would this fit into a scenario for such a motion specification framework?
>
> In principle yes: _if_ you succeed in transforming all your rules into
> "constraints" between several of the tunable parameters. Currently, we
> have an implementation that can deal (only) with linear equality
> constraints; further work is under way for inequality constraints, and
> for special kinds of nonlinear constraints (i.e., the "convex" ones or
> others for which fast solvers exist).
>
> > >It's part of Ruben's PhD work, and he is currently working on an
> >
> > implementation in Orocos. I can send you a paper (off line) that we
> > published about it two years ago
> >
> > It would be nice if you could send me this "paper" of Rubens phd.
> >
> > What is the status of this framework, could it be used by me?
>
> I think it is not yet "production ready", but probably Ruben can
> provide more information...

It's not production ready yet, but it is operational. One of our students is currently using it, and IMHO the framework proves its usability. It does not yet support all the functionality described in the paper (mainly the uncertainty estimation).

I'm currently preparing a paper which explains the higher level usage on top of this framework.

When all functionality (at least the one explained in the paper) is incorporated I would like to write a paper about it and publish all the code
;) (if I'm allowed)

Ruben

Disclaimer: http://www.kuleuven.be/cwis/email_disclaimer.htm