Minutes of Orocos meeting 18-19 nov, Leuven

I've put the minutes of the Orocos meeting in Leuven, of last 18-10
November, on-line at
 <http://www.orocos.org/meeting-nov2002.html>.
Here is the text of the minutes, but the web page contains links to
the slides of the presentations (and a pictrure of the wokshop).

Herman
========================

This was a two-days meeting:

On the first day we had presentations by the EU project partners,
and external people, about the progress of the project, and
discussions about suggestions and extensions.

27 people attended this first day.


On the second day, Peter Soetens from K.U.Leuven led a workshop in
which he explained the design and implementation of the real-time
motion control core built at K.U.Leuven.

20 people attended this second day.


1st day technical meeting
=========================

Herman Bruyninckx (K.U.Leuven) gave an overview of
the project history and status, and about the efforts taking place at
K.U.Leuven. These are: the real-time motion control core, the robot
kinematics and dynamics library, and a Bayesian probability library.

Anthony Mallet (LAAS Toulouse) explained the status
of their redesign of the Genom ("Generator of Modules") software.
Genom supports both control and data flow between distributed
components during a task execution.  At LAAS, this tool is used in
many different robot systems already, and has proven its usefulness.
Current redesign efforts focus on the components architecture
specification, in order to allow different strategies regarding
communication or execution to be implemented.

Anders Oreback (KTH Stockholm) and Christian Schlegel
(FAW Ulm) presented their work on a CORBA based set of communication
primitives for robot control. The goal of both projects is to get to a
minimalistic, easy to use, and yet feature complete communication
library to connect robotics components together, using standard CORBA
implementations.

Peter Soetens (K.U.Leuven) gave an overview of the design and features
of the real-time core.

Cezary Zielinski (Warsaw) presented his ideas on a complex controller
implementation with inspiration from fuzzy theory.

Daniel Simon (INRIA Grenoble)
gave an overview of the features of Orccad which could be of interest
to Orocos. Orccad is an integrated tool (including the Esterel
language and a GUI for building applications) for the assembly and
formal verification of real-time tasks, with possibly high complexity.


The presentations showed that Orocos has reached a productive
development pace, with interesting functionality being completed right
now. The communication and task execution designs are maturing, and
converging to clearly defined interfaces. However, it's exactly in the
area of interfaces that lots of work still has to be done: no robust
and stable implementations are possible without clean and clear
interfaces.


2nd day workshop
================

Peter Soetens, lead developer of the OROCOS efforts at K.U.Leuven,
introduced the current status of the real-time motion control core
within OROCOS. Due to there being twice as many people than
registered, we had some small delays in getting everybody access to
the net, to download the installation packages. Peter explained the
installation process, and the efforts that went into using automake
and autoconf, to smoothen the installation for the users. 

The larger part of the day was spent on going through the major
packages of the real-time core, such as the Core itself, the motion
control package, the OS interface, etc. 

A demo was given (on the first day) of the motion control software
running on a KUKA 361 industrial robot. The drives and internal analog
control of the industrial robot are used, but the encoders and
setpoint are interfaced to a PC running OROCOS. The current
functionality was illustrated through a demo of position control in
joint and Cartesian space.

Notwithstanding the heavy material, many interesting feedback and
remarks were given by the participants. Some explained they're serious
about trying the beta code out in the very near future.


Herman Bruyninckx
22 November 2002