||AI in the Predator/Prey Domain
||Laursen Würtz, Frej
||Bolander, Thomas (Algorithms and Logic, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||This thesis concerns Artificial Intelligence in the Predator/Prey Domain. The Predator/
Prey Domain is in this setup implemented as a game of catch where actual robots,
in the form of Lego MindStorms, will cooperate on capturing a third part robot, either
controlled by a human player or by a computer.
The concept is implemented as a Multi-Agent System (MAS), where each agent represents
different strategies (AIs) on how to solve the problem. Each agent will continuously
during the game negotiate on which strategy to follow for a number of iterations,
and at the end of each run evaluate on the process and establish trust relations based
on the progress. The MAS is further implemented in such a way that new strategies
can be presented at run time.
The entire concept is handled both in physical and simulated environments.
The problem is first solved in a five times five grid with no barriers and is afterwards
considered for an actual maze with internal barriers.
Creation date: 2008-06-04
Update date: 2008-07-09