AI Seminar!
-------------------------------
Tuesday Jan. 18, 2000
4:10 pm - 5:00 pm
175 ATL (Large Conference Room)
Exploiting Hierarchical Representations for Efficient
and Flexible Coordination in Multi-Agent Systems
Prof. Edmund H. Durfee
University of Michigan AI Laboratory
ABSTRACT: Because the decisions made by a computational agent can affect
the outcomes of others' decisions, a common concern in multi-agent systems
is to devise techniques for coordinating agents' decisions. A challenge is
that the computation and communication needs for coordinating each and
every specific decision can become overwhelming; it is therefore important
for agents to communicate and coordinate at more abstract levels, knowing
only as much about others as is needed. In this talk, I'll describe our
current efforts in using hierarchical representations of agent activities
to support coordination.