The game of chess has many times been discussed and used for test purpose by science departments of Artificial Intelligence (AI). Although the technique of agent and as well multi-agent systems is quite old, the use of these offspring of AI within chess is limited. This report describes the project performed applying the use of agents to a chess program. To measure the performance of the logic has tests between the developed program main parts been performed. Further tests against a traditional chess program as well to position test suites have been done. The results and the impact of the different logic parts is presented and discussed. The aim of the project is to take the use of agents in chess a step forward.
Contents
1. Introduction
Scope and Limitations
Artificial Intelligence (AI) and Agents
Previous Work in the Area
The Game of Chess
Computer Chess Programs
Traditional Chess Program Development
2. Project Description
Technical Specifications
Agent Logic and Modules
Vocabulary Terms
Agent Logic Levels
Agent Communication
Agent Evaluation
3. Experiments
Test Data to Collect
Crafty
Common Tests
Internal Tests
External Tests
Test Suites
Test Results
Outcome and Material Test Data
Attack Request Test Data
Defense Request Test Data
Move-Attack Test Data
Threat Test Data
Test Suite Data
4. Discussion
Agent Approach vs. Traditional Approach
AgentChess vs. Drogoul’s MARCH
Test Results
5. Conclusions
6. Future of AgentChess
7. References
Author: Henric Fransson
Source: Blekinge Institute of Technology
Reference URL 1: Visit Now