Home | english  | Impressum | Sitemap | KIT

KRASH/ControMAS

KRASH/ControMAS
Ansprechpartner:Prof. P. Lockemann
J. Nimis

KRASH (2000-2004) / ControMAS (2004-2006)

The project KRASH and its follow-on project ControMAS are funded by the Deutsche Forschungsgemeinschaft (DFG - German Research Council) as part of the Priority Programme SPP1083 "Intelligent Agents and Realistic Commercial Application Scenarios".

 

KRASH (2000-2004)

The Karlsruhe Robust Agent Shell

The KRASH project strives to achieve a more robust and predictable operation of multi-agent systems in production planning and control (PPC) applications. KRASH is motivated by the observation that the acceptance of multi-agent-based solutions into the PPC market is slow to non-existent, although there is little debate that, in principle, multi-agent systems are more flexible and better suited to rapidly changing production loads than classical centralized PPC systems. We believe that the market success of multi-agent-based PPC system hinges on their ability to function as robust and with a result quality comparable to established centralized systems.

To achieve this goal, KRASH develops a testbed and runtime environment for multi-agent systems that, on the one hand, allows for an extensive simulation and evaluation of the system in real-world production scenarios and, on the other hand, provides a robust, transaction-based infrastructure for executing agent actions in a resilient way. Both the simulation environment and the transaction layer are designed to fit into the standard FIPA agent platform architecture, thus making their services easily accessible to FIPA-compliant multi-agent systems.

 

ControMAS (2004-2006)

With shop floor schedulling as application domain the ControMAS-project tries to integrate multiagent systems into the existing system and organizational landscape. Therefore, system architectures and interfaces for MAS and business software interaction have to be established. To assist the human scheduler on the shop floor control methods for MAS based on priority rules and constraints are developed.