Robotics course notes with concrete kinematic models...

Maybe it is interesting in the ongoing discussion on the kinematics
modelling and class definitions that we can begin to discuss about
concrete models. Therefore, interested people can take a look at the
kinematics chapters of my robotics course:
 <http://people.mech.kuleuven.ac.be/~bruyninc/robotics.ps.gz>
(This is a 2.5MB file...)

The generic structure on p. 100 (which I call the "321" family) covers
many robot arms. Some slight additions are needed for Puma like
robots, i.e., robots that differ from this simple 321 model by some
offsets here and there. I do have some code of some of these
extensions, but not nicely integrated.
The "lrl" library of the current Orocos@KULeuven code contains this
implementation, and some documentation on the various alternative
conventions that can be used. (Look for the "feath6r" directory.)

Another related issue is the efficient numeric calculation of general
kinematic structures. I have put the common algorithms in my course
notes, which could serve as a starting point for a concrete discussion
on this point. These algorithms are certainly also not new and hence
been implemented several times in the past.

Finally, also the mobile platforms should be covered! I describe again
the simplest unicycle and bicycle models, but a generic implementation
is missing. I'm sure some of the mobile robot users in this community
have much more extensive algorithms.

Is there a candidate to coordinate the (hopefully starting)
developments in this kinematics area? The important thing to start is
to define how we will make concrete models. I.e., filling in some more
the interfaces I started to describe in the kinematics document
 <http://www.orocos.org/kindyn-doc.html>

Herman


-- 
  K.U.Leuven, Mechanical Engineering, Robotics Research Group
<http://people.mech.kuleuven.ac.be/~bruyninc> Tel: +32 16 322480