C O MR O S - A Multi-Agent Robot Architecture

In this paper we propose a refined version of the COMROS multi agent robot architecture. The concepts we use to define the operational principles of our architecture are autonomy of each functional module, task decomposition, and de- centralized control. The structure of our architecture is defined by elementary agents which are connected by two types of networks. Furthermore, we describe how for- mation driving can be realized within the architecture.