| Newsgroups |
gmane.science.robotics.orocos.user |
| Message-ID |
<[email protected]> |
Timed Petri Nets are a good modelling tool for systems with concurrency,
synchronisation and timed activities. Analysis is then possible, more
especially if we have only synchronisation or only concurrency in which case
we can perform formal analysis using the underlying mathematics.
Anyway, I consider that PNs are not good specification and programming tools
for our kind of applications.
>I do have the feeling that using time to synchronize two or more
>activities is not a good idea,
Right! This lead to loose any robustness w.r.t. timing uncertainties. Better
schedule activities on events e.g. end_action, exception_raised,
signal_detection... which may occur at unpredictable time. Even execution
times of pieces of control algorithms are subject to timing uncertainties.
But hand design of events/states machines is boring and error prone as soon as
their size becomes big which happens very fast, in particular when we specify
parallel activities of components.
Thus a high level textual and/or graphical FSM specification and generation
tool should be used, e.g. Esterel, Statecharts or so...
Daniel
***********The difference between theory and practice****************
**************is larger in practice than in theory*******************
Daniel SIMON Projet Bip INRIA Rhone-Alpes
ZIRST, 655 avenue de l'Europe, 38330 MONTBONNOT SAINT MARTIN, FRANCE
[email protected] Phone:(33)476615328 Fax:(33)476615252
http://www.inrialpes.fr/bip/people/simon
*********************************************************************