Moving from a single agent to multiple agents represents a
"game-changer" in several ways. The at least partly controllable
behaviour of a single agent can, in conjunction with other agents,
lead to so-called "emergent" behaviour. Notwithstanding the current
fashion of attempting to harness emergent dynamics, in the case of
multiagent systems with a well-defined mission, this consists
essentially of unexpected and mostly undesirable behaviour. Unlike
conventional bugs, such dynamics can appear indirectly and can be
difficult to control and debug.
The tutorial will introduce multiagent systems, from more abstract
multiagent to increasingly concrete soccer-oriented scenarios, to give
an overview over the current state-of-the art techniques, phenomena
and methods of analysis.
Universität Bonn, Institute for Computer Science, Autonomous Intelligent Systems