Date of Award

7-2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Engineering and Sciences

First Advisor

Thomas C. Eskridge

Second Advisor

Gary Burns

Third Advisor

Terrence O’Connor

Fourth Advisor

William H. Allen

Abstract

Progress in the use of Artificial Intelligence (AI) has been made in many different fields. We are now reaching a point in AI development typical AI implementations are not enough: where we would need humans and AI systems to actively collaborate with each other, basing their actions on the actions and capabilities of each other. This collaboration could be in the form of agents assisting humans processing and analyzing information, assisting humans with smaller physical tasks or working with humans as an equal team member – having the same goals and performing the same tasks – to accomplish a goal. These advancements and new capabilities indicate the need for a teamwork model that can facilitate the creation and efficiency of human-AI teams. There are currently few teamwork models available that can assist AI developers and teamwork theorists in the creation of intelligent, automated teammates. The goal of the research mentioned in this dissertation is to fill this gap by creating a novel goal and information-sharing algorithm which improves and extends the working capabilities of an existing theoretical teamwork model and extends an existing AI planning agent system (a PDDL planner) to share environmental information obtained by humans and agents. The information shared is specifically related to the goals and tasks of each agent. The result of this research is a methodology for extending existing PDDL planners with the capability of mutually beneficial interactions between it and other agents or humans. The results of using the system with ad-hoc teams of existing PDDL planners are improved overall task performance.

Download

Share

COinS