Anita Raja, Victor Lesser
Embedded systems consisting of collaborating agents capable of interacting with their environment are becoming ubiquitous. It is crucial for these systems to be able to adapt to the dynamic and uncertain characteristics of an open environment. The question of when this adaptation process should be done and how much effort should be invested in the adaptation becomes especially challenging in the context of multi-agent systems. We present a generalized agent framework for meta-level control called GeMEC. We describe GEMEC's decentralized Markov Decision Process (DEC-MDP)-based model for decision making in Netrads, a tornado tracking application. This model will capture interactions where meta-level decisions made in one agent's MDP can affect meta-level MDPs of other agents. The cost of meta-level control can be controlled by constructing and evaluating the DEC-MDPs offline.
Subjects: 7. Distributed AI; 7.1 Multi-Agent Systems
Submitted: May 5, 2008