|
(See ISLA for the new IAS website. Most maintenance activities on this old website have stopped.)
 |
Robocup Simulation Soccer |
Objective
The Robot World Cup Initiative (RoboCup) is an attempt to foster AI and
intelligent robotics research by providing a standard problem in which a wide
range of technologies can be integrated and examined. RoboCup's final goal is
a soccer world cup with real robots and in order to realize this goal several
competitions have been set up which all focus on different aspects of the
overall problem. One of these competitions is the Simulation League which
concentrates mainly on RoboCup aspects at a high level of abstraction such as
team strategy and learning capabilities. Our current team UvA Trilearn
participated both in RoboCup-2001 and RoboCup-2002 and reached 4th place on both
occasions. Furthermore we became Champion at the German Open 2002.
Research group members
dhr. J.R. Kok (jellekok@science.uva.nl)
dr. N. Vlassis(vlassis@science.uva.nl)
Prof. dr. ir. Frans Groen (groen@science.uva.nl)
Funding
University of Amsterdam (UvA)
Research Achievements
A soccer game is a specific but very attractive real time multi-agent
environment from the viewpoint of distributed artificial intelligence and
multi-agent research. Several scientific areas of interest which we will address
in our project are the following:
- Dynamic Resource Allocation/Heterogeneous Agents. Given the different
skills agents have, and given that agents grow tired, how should agents best
divide their tasks between them?
- Adjustable Autonomy. How do players decide on interpreting and acting
on a coach's advice or supplied information?
- Multi-Agent Modeling. How can a coach recognize its own team's and an
opponent's strength's and weaknesses? How can it model their behavior?
- Machine Learning: Agent Behavior Models. How can a coach agent build
a model of the behavior of a team from observations of the teams?
- Teamwork and Coordination. How can a group of agents work together,
collaborating and coordinating, as an effective team? How do dynamic small
sub-teams form, and how can they be made effective?
- Machine Learning. How can an agent improve its performance using its
own experiences in interacting with the environment.
- Adversarial Planning. How can agents plan to counter opponents'
behavior?
- Agent Architectures. What architectures are useful for dynamic,
complex, multi-agent settings?
Project Homepage
More detailed information (including sources, binaries, publications and
logfiles) can be found at the
UvA Trilearn
website.
More information about the RoboCup Initiative itself can be found at
http://www.robocup.org.
|
|
