자료유형  

 학위논문

Control Number  

0017160344

International Standard Book Number  

9798383482056

Dewey Decimal Classification Number  

629.8

Main Entry-Personal Name  

Gregory, Jason M.

Publication, Distribution, etc. (Imprint  

[S.l.] : University of Southern California., 2024

Publication, Distribution, etc. (Imprint  

Ann Arbor : ProQuest Dissertations & Theses, 2024

Physical Description  

207 p.

General Note  

Source: Dissertations Abstracts International, Volume: 86-01, Section: B.

General Note  

Advisor: Gupta, Satyandra K.

Dissertation Note  

Thesis (Ph.D.)--University of Southern California, 2024.

Summary, Etc.  

요약The rapid advancement of artificial intelligence, machine learning, human robot interaction, and safe learning and optimization has led to significant leaps in component- and behavior-level capabilities for autonomous robots. However, human capability-enhancing research, development, testing, and evaluation toward understanding and building trust in next-generation autonomous robots still requires additional attention. This is important because autonomous robots cannot be deployed safely alongside humans without sufficient understanding of system performance, limitations, and trustworthiness of capabilities, which necessitates experimentation. The process of constructing experiments, i.e., experimental design, is a supreme step in the concept-to-fielding life cycle of an autonomous robot because it dictates the amount of information gained by the experimenter, the cost of information acquisition, and the rate of building system understanding. Conducting experiments is challenging, though, due to complexity and context. Autonomous robots can be massively complex, multi-disciplinary systems that use artificial intelligence and machine learning across a range of components (e.g., perception, state estimation, localization, mapping, path planning, and control) and experiments are specific to a given scenario, system, experimenter, and set of multi-objective metrics defined by the intended application. To assist with the adaptive, sequential decision-making process of experimental design, a Decision Support System (DSS) can potentially augment the human's abilities to construct more informative, less wasteful experiments. This dissertation aims to provide conceptual and computational foundations for DSSs in the domain of adaptive experimental design for autonomous, off-road ground vehicles. First, I present a six-stage taxonomy of DSSs for experimental design of ground robots, which is informed and inspired by the vast body of literature of DSS development in domains outside of robotics. This taxonomy also serves as a roadmap to guide ongoing development of DSSs for experimental design. Next, I develop and evaluate a Stage 1 DSS that provides design assistance to experimenters in the form of prompts for the purposes of experimental design conceptualization and structured thought analysis. Building on this I propose and evaluate a Stage 2 DSS to provide proactive decision support in the form of alerts so that low value experimental designs might be avoided. Finally, I lay the groundwork for a Stage 3 DSS to provide narrowly-scoped experimental design recommendations for assisting with subsequent experimental selections. I anticipate that this work will help improve human decision-making of experimental design for real-world autonomous ground vehicles.

Subject Added Entry-Topical Term  

Robotics.

Subject Added Entry-Topical Term  

Computer science.

Index Term-Uncontrolled  

Decision support systems

Index Term-Uncontrolled  

Experimental design

Index Term-Uncontrolled  

Field robotics

Index Term-Uncontrolled  

Mobile robotics

Index Term-Uncontrolled  

Off-road ground vehicles

Added Entry-Corporate Name  

University of Southern California Computer Science

Host Item Entry  

Dissertations Abstracts International. 86-01B.

Electronic Location and Access  

로그인을 한후 보실 수 있는 자료입니다.

Control Number  

joongbu:654605