Date of Award

6-2020

Document Type

Thesis

Degree Name

Master of Science

Department

Department of Electrical and Computer Engineering

First Advisor

Kenneth M. Hopkinson, PhD

Abstract

This thesis takes the Scotland Yard board game and modifies its rules to mimic important aspects of space in order to facilitate the creation of artificial intelligence for space asset pursuit/evasion scenarios. Space has become a physical warfighting domain. To combat threats, an understanding of the tactics, techniques, and procedures must be captured and studied. Games and simulations are effective tools to capture data lacking historical context. Artificial intelligence and machine learning models can use simulations to develop proper defensive and offensive tactics, techniques, and procedures capable of protecting systems against potential threats. Monte Carlo Tree Search is a bandit-based reinforcement learning model known for using limited domain knowledge to push favorable results. Monte Carlo agents have been used in a multitude of imperfect domain knowledge games. One such game was in which Monte Carlo agents were produced and studied in an imperfect domain game for pursuit-evasion tactics is Scotland Yard. This thesis continues the Monte Carlo agents previously produced by Mark Winands and Pim Nijssen and applied to Scotland Yard. In the research presented here, the rules for Scotland Yard are analyzed and presented in an expansion that partially accounts for spaceflight dynamics in order to study the agents within a simplified model, while having some foundation for use within space environments. Results show promise for the use of Monte- Carlo agents in pursuit/evasion autonomous space scenarios while also illuminating some major challenges for future work in more realistic three-dimensional space environments.

AFIT Designator

AFIT-ENG-MS-20-J-003

DTIC Accession Number

AD1104463

Download

Share

COinS