Risk is a game in which traditional Artificial-Intelligence methods such as for example iterative deepening and Alpha-Beta pruning can not successfully be applied due to the size of the search space. Distributed problem solving in the form of a multi-agent system might be the solution. This needs to be tested before it is possible to tell if a multi-agent system will be successful at playing Risk or not.In this thesis the development of a multi-agent system that plays Risk is explained…
Contents
1 Introduction
1.1 Problem Deļ¬nition
1.2 Scope and Limitations
1.3 Method
1.4 Outline of the thesis
2 Domain
2.1 Risk
2.1.1 Trade in cards
2.1.2 Place armies
2.1.3 Attack
2.1.4 Fortify
2.2 Limitations
3 MARS
3.1 Environment
3.2 Multi-Agent System Architecture
3.2.1 Agent Distribution
3.2.2 Agent Communication
3.2.3 Agent Negotiation
3.3 Design
3.3.1 Mediator Agent
3.3.2 Country Agent
3.3.3 Card Agent
3.3.4 Parameters
3.3.5 Country Evaluation
3.3.6 Dice Odds
3.4 Actions
3.4.1 Trade in cards
3.4.2 Place armies
3.4.3 Attack
3.4.4 Fortify
4 Experiments
4.1 Experiments Setup
4.1.1 Environment
4.1.2 The Opponents
4.1.3 Parameter Optimization
i4.1.4 Performance
5 Experiments Result
5.1 Parameter Optimization
5.2 Performance
5.2.1 Experiment 1
5.2.2 Experiment 2
5.2.3 Experiment 3
6 Discussion
6.1 Ranking
6.2 Runtime
6.3 Stability
6.4 Parameter Optimization
6.5 Multi-Agent System
6.6 Evolutionary Viewpoint
6.7 Methodology
6.8 Strengths and Weaknesses
7 Conclusion
8 Future Work
8.1 Cooperate
8.2 Aware of Different Opponents
8.3 Improved Parameter Optimization
8.4 More Opponents
8.5 Compare to Other MAS Bots
Author: Fredrik Olsson
Source: Blekinge Institute of Technology
Download URL 2: Visit Now