서브메뉴
검색
Towards Robust Eco-Driving: Platform, Algorithms, Scenarios, and Testing
Towards Robust Eco-Driving: Platform, Algorithms, Scenarios, and Testing
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105126
- ISBN
- 9798291551936
- DDC
- 385
- 저자명
- Liang, Zhaohui.
- 서명/저자
- Towards Robust Eco-Driving: Platform, Algorithms, Scenarios, and Testing
- 발행사항
- [Sl] : The University of Wisconsin - Madison, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 117 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-02, Section: B.
- 주기사항
- Advisor: Li, Xiaopeng Shaw.
- 학위논문주기
- Thesis (Ph.D.)--The University of Wisconsin - Madison, 2025.
- 초록/해제
- 요약Eco-driving at signalized intersections is a key strategy for improving fuel efficiency and reducing emissions in urban environments. However, existing eco-driving algorithms often fall short of real-world applicability due to computational complexity, poor disturbance tolerance, or limited integration with real-time vehicle control systems. This dissertation addresses these gaps by developing a unified framework that combines analytical trajectory planning, adversarial scenario generation, and hierarchical field testing to enable robust, scalable, and deployable eco-driving for Connected and Automated Vehicles (CAVs).First, we propose a novel analytical trajectory planning method based on cubic functions, capable of generating closed-form, smooth, and fuel-efficient trajectories in real time. Unlike traditional optimization-based methods, the proposed planner does not require iterative solvers, making it highly suitable for embedded deployment. The method accounts for signal timing constraints, acceleration feasibility, and driver comfort, and includes mechanisms to gracefully handle waiting scenarios through a three-phase planning structure.To evaluate the resilience of eco-driving strategies under real-world uncertainty, we introduce a structured adversarial scenario generation framework. Disturbances are categorized into environmental and systemic types and modeled probabilistically along adversarial directions. A continuous adversarial level parameter, α, enables fine-grained control of disturbance severity, allowing scalable and repeatable stress testing of controller performance. Scenarios include stochastic lead vehicle behavior, adaptive red phase extensions, sensor noise, GPS errors, lateral cut-ins, road surface variations, and control execution delays.We also define two novel evaluation metrics-resilience indicator G and robustness indicator H-to quantify performance degradation under external disturbances and internal variability, respectively. These metrics measure the relative drop in utility between planned and executed trajectories, normalized by the baseline utility gap, providing a unified view of controller stability and adaptability.The proposed methods are implemented and tested on a full-scale Level 3 CAV platform with integrated C-V2X communication, capable of receiving real-time SPaT data from infrastructure-deployed Roadside Units (RSUs). A hierarchical testing pipeline-consisting of software-in-the-loop, hardware-in-the-loop, and closed-road experiments-is used to validate controller behavior and resilience under controlled adversarial conditions. Comparative experiments demonstrate that while both analytical and optimization-based methods achieve similar energy savings, they differ in resilience profiles: optimization-based controllers offer stronger disturbance adaptation through feedback, while analytical methods provide smoother control under low-to-moderate variability.In conclusion, this dissertation contributes a practical and principled approach to robust eco-driving, combining algorithmic simplicity with empirical rigor. By enabling lightweight planning and structured disturbance evaluation, the proposed framework lays the groundwork for certifiable, real-time eco-driving solutions in future intelligent transportation systems. The tools, models, and evaluation metrics introduced here also provide a foundation for further research in resilience-aware autonomous driving.
- 일반주제명
- Transportation
- 기타저자
- The University of Wisconsin - Madison Civil & Environmental Engr
- 기본자료저록
- Dissertations Abstracts International. 87-02B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017359490
■00520260202105126
■006m o d
■007cr#unu||||||||
■020 ▼a9798291551936
■035 ▼a(MiAaPQ)AAI32238850
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a385
■1001 ▼aLiang, Zhaohui.
■24510▼aTowards Robust Eco-Driving: Platform, Algorithms, Scenarios, and Testing
■260 ▼a[Sl]▼bThe University of Wisconsin - Madison▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a117 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-02, Section: B.
■500 ▼aAdvisor: Li, Xiaopeng Shaw.
■5021 ▼aThesis (Ph.D.)--The University of Wisconsin - Madison, 2025.
■520 ▼aEco-driving at signalized intersections is a key strategy for improving fuel efficiency and reducing emissions in urban environments. However, existing eco-driving algorithms often fall short of real-world applicability due to computational complexity, poor disturbance tolerance, or limited integration with real-time vehicle control systems. This dissertation addresses these gaps by developing a unified framework that combines analytical trajectory planning, adversarial scenario generation, and hierarchical field testing to enable robust, scalable, and deployable eco-driving for Connected and Automated Vehicles (CAVs).First, we propose a novel analytical trajectory planning method based on cubic functions, capable of generating closed-form, smooth, and fuel-efficient trajectories in real time. Unlike traditional optimization-based methods, the proposed planner does not require iterative solvers, making it highly suitable for embedded deployment. The method accounts for signal timing constraints, acceleration feasibility, and driver comfort, and includes mechanisms to gracefully handle waiting scenarios through a three-phase planning structure.To evaluate the resilience of eco-driving strategies under real-world uncertainty, we introduce a structured adversarial scenario generation framework. Disturbances are categorized into environmental and systemic types and modeled probabilistically along adversarial directions. A continuous adversarial level parameter, α, enables fine-grained control of disturbance severity, allowing scalable and repeatable stress testing of controller performance. Scenarios include stochastic lead vehicle behavior, adaptive red phase extensions, sensor noise, GPS errors, lateral cut-ins, road surface variations, and control execution delays.We also define two novel evaluation metrics-resilience indicator G and robustness indicator H-to quantify performance degradation under external disturbances and internal variability, respectively. These metrics measure the relative drop in utility between planned and executed trajectories, normalized by the baseline utility gap, providing a unified view of controller stability and adaptability.The proposed methods are implemented and tested on a full-scale Level 3 CAV platform with integrated C-V2X communication, capable of receiving real-time SPaT data from infrastructure-deployed Roadside Units (RSUs). A hierarchical testing pipeline-consisting of software-in-the-loop, hardware-in-the-loop, and closed-road experiments-is used to validate controller behavior and resilience under controlled adversarial conditions. Comparative experiments demonstrate that while both analytical and optimization-based methods achieve similar energy savings, they differ in resilience profiles: optimization-based controllers offer stronger disturbance adaptation through feedback, while analytical methods provide smoother control under low-to-moderate variability.In conclusion, this dissertation contributes a practical and principled approach to robust eco-driving, combining algorithmic simplicity with empirical rigor. By enabling lightweight planning and structured disturbance evaluation, the proposed framework lays the groundwork for certifiable, real-time eco-driving solutions in future intelligent transportation systems. The tools, models, and evaluation metrics introduced here also provide a foundation for further research in resilience-aware autonomous driving.
■590 ▼aSchool code: 0262.
■650 4▼aTransportation
■650 4▼aEnvironmental engineering
■653 ▼aUrban environments
■653 ▼aConnected and Automated Vehicles
■653 ▼aEco-driving strategies
■690 ▼a0709
■690 ▼a0543
■690 ▼a0775
■71020▼aThe University of Wisconsin - Madison▼bCivil & Environmental Engr.
■7730 ▼tDissertations Abstracts International▼g87-02B.
■790 ▼a0262
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359490▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


