자료유형  

 학위논문

Control Number  

0017161382

International Standard Book Number  

9798382611853

Dewey Decimal Classification Number  

629.8

Main Entry-Personal Name  

Ding, Wenhao.

Publication, Distribution, etc. (Imprint  

[S.l.] : Carnegie Mellon University., 2024

Publication, Distribution, etc. (Imprint  

Ann Arbor : ProQuest Dissertations & Theses, 2024

Physical Description  

247 p.

General Note  

Source: Dissertations Abstracts International, Volume: 85-11, Section: B.

General Note  

Advisor: Zhao, Ding.

Dissertation Note  

Thesis (Ph.D.)--Carnegie Mellon University, 2024.

Summary, Etc.  

요약Training and evaluating autonomous robots within the real world present significant challenges and risks, emanating from unpredictable environments, safety concerns, ethical dilemmas, and limited human oversight. As a mitigation strategy, the use of realistic simulations, also known as digital twins, offers virtual duplication of the actual system or environment, thus fostering the development of trustworthy autonomy.Digital twins enable developers to evaluate the performance of systems in various scenarios to identify potential risks or failure cases. It facilitates the accumulation and subsequent analysis of datasets, which serve to validate and calibrate the autonomous system's perception and decision-making algorithms. By comparing the behavior of the digital twin with real-world data, developers can identify discrepancies, improve accuracy, and enhance the system's safety and reliability.Scenarios that embody dynamic and interactive components reflect the intricacies of digital twins and take precedence in significance. One main value of digital twins is helping us understand how objects interact and behave. For example, in autonomous driving, the behavior of vehicles, pedestrians, and traffic conditions are crucial components of scenarios that need to be accurately modeled. However, not all scenarios in digital twins are created equal. In the pursuit of developing trustworthy autonomy, ordinary scenarios often prove insufficient in subjecting autonomous systems to extreme conditions where safety and robustness are paramount. Although critical scenarios hold the potential to expose model vulnerabilities, their rare occurrence creates a challenge. The process of manually identifying or extrapolating such critical scenarios from normal data or expert design proves not only inefficient but also contains substantial human biases.My doctoral research seeks to harness the potential of generative AI to explore two pivotal questions: (1) Which scenarios are critical in existing data and (2) How to generate such scenarios in digital twins? The proposal begins with the definition of critical scenarios and the corresponding optimization problem and subsequently delves into three distinct categories of scenario generation frameworks: data-driven generative models, adversarial generative models, and knowledge-guided generative models. Concluding this thesis is future directions that effectively combine generation resources from different perspectives and improve the data flywheel by.

Subject Added Entry-Topical Term  

Robotics.

Subject Added Entry-Topical Term  

Computer science.

Subject Added Entry-Topical Term  

Automotive engineering.

Subject Added Entry-Topical Term  

Mechanical engineering.

Index Term-Uncontrolled  

Autonomous driving

Index Term-Uncontrolled  

Digital twins

Index Term-Uncontrolled  

Autonomous robots

Index Term-Uncontrolled  

Accuracy

Added Entry-Corporate Name  

Carnegie Mellon University Mechanical Engineering

Host Item Entry  

Dissertations Abstracts International. 85-11B.

Electronic Location and Access  

로그인을 한후 보실 수 있는 자료입니다.

Control Number  

joongbu:658049