| Newsgroups |
gmane.science.robotics.orocos.user |
| Message-ID |
<Pine.LNX.4.44.0208170853320.6201-100000@srv04.mech.kuleuven.ac.be> |
On Fri, 16 Aug 2002 [email protected] wrote:
> A "snippet" of RIMCP might look like this:
>
> Grasp("Wine glass 7", 0.3, 0.1)
>
> where "Wine glass 7" is already defined as a "detailed 3D drawing",
> 0.3 is the maximum pressure in ounces that can safely be applied to
> the glass,
Orocos wants to be unambiguous in physical units, hence a magic number
like "0.3" won't do :-) And a pressure is force/surface area :-)
And what the hell are ``ounces''? :-)
> and 0.1 is a measure of the rate of change in speed and
> inertia to prevent spilling the liquid in the glass.
Speed and inertia are 6-dimensional entities, so a single number is
ambiguous.
Sorry for this nitpicking, but your examples illustrate one of the
major problems in ``high level'' robotics: there are so many implicit
assumptions made in any high level statement, that exchanging robot
programs is a _very_ tough task.
> In order for the master CPU to understand the "design of a new arm",
> during the learning process, a table of Range of motion values would
> have to be transferred from the arm's CPU to the master CPU.
And many more variables: kinematics, dynamics, available motion
profiles, available sensors + motor, and their performance, ...
> Once we have these "heavy duty theoretical pictures" in place, I
> think it will be easier to encourage mfgs to adopt our
> standards...mostly because they realize that it makes their job so
> much easier.
Do they? Most of them have spent already many millions (in whatever
currency) to get their own stuff together; they are not keen in
spending more in adopting a ``standard'' that they do not control :-)
Herman
--
K.U.Leuven, Mechanical Engineering, Robotics Research Group
<http://people.mech.kuleuven.ac.be/~bruyninc> Tel: +32 16 322480