The Model-based Computing Project
The model-based computing project at Xerox PARC is an attempt
to streamline the writing of software for electro-mechanical systems
(e.g., photo-copiers). The central idea behind model-based computing
is to develop compositional, declarative models of the various
components of a system, at different levels of granularity. These
models can then be composed according to the design of the system, to
yield a model of the system.
Together with this, software architectures and generic
algorithms are developed for the tasks that need to be accomplished
(e.g., design evaluation, scheduling, simulation, machine control,
diagnosis). Finally, linking the two are special-purpose
reasoners that produce information of the right kind for the
given task architecture, given the component models and system
configuration.
Here is a longer overview of the
motivation and characteristics of model-based computing.
Please understand that these pages describe an earlier phase of
this project and thus some of the information (e.g., people and
addresses) may be outdated.
Project Groups
- Scheduling
- At the heart of the control software for reprographic
machines is a scheduler. It determines which
operations have to be performed when in order to produce the
desired documents. We have applied model-based computing
techniques to the development of such schedulers, and have
extended this work to the evaluation of the designs of
reprographic machines.
- Language Design
- We are developing a language for compositional modeling of
hybrid systems, which are systems which involve both continuous
and discrete changes. We plan to use this language to model a
photocopier and use it to simulate and diagnose the machine, as well
as generate control code for it.
Here is a list of all publications
concerning the model-based computing projects mentioned above.
The People
Former and current affiliates.
Last updated on January 1, 1997.
