서브메뉴
검색
Design, Testing, and Evaluation of Automated Vehicle Control Systems in Transportation
Design, Testing, and Evaluation of Automated Vehicle Control Systems in Transportation
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211153128
- ISBN
- 9798346873976
- DDC
- 385
- 저자명
- Ma, Ke.
- 서명/저자
- Design, Testing, and Evaluation of Automated Vehicle Control Systems in Transportation
- 발행사항
- [Sl] : The University of Wisconsin - Madison, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 146 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-06, Section: B.
- 주기사항
- Includes supplementary digital materials.
- 주기사항
- Advisor: Li, Xiaopeng Shaw.
- 학위논문주기
- Thesis (Ph.D.)--The University of Wisconsin - Madison, 2024.
- 초록/해제
- 요약Automated vehicles (AVs) hold the promise of transforming transportation systems by enhancing safety, mobility, stability, and sustainability. However, current research often falls short of fully capturing the unique characteristics of AV behaviors, particularly in car-following scenarios. Existing models predominantly focus on human-driven vehicles (HVs) and fail to account for the specific control mechanisms and hardware constraints inherent in AVs. This dissertation aims to bridge these gaps by developing a comprehensive framework that includes a uniform trajectory dataset, standardized evaluation metrics, and a unified dynamics structure tailored for AVs.The research is guided by four primary goals: developing a comprehensive and uniform AV trajectory dataset; assessing current commercial AV behaviors using standardized metrics; developing a unified dynamics structure for AV models and controllers; and designing and optimizing improved controllers within the established dynamics structure. To achieve these goals, a large-scale empirical dataset was compiled from 30 AVs across multiple automakers, capturing a wide range of driving scenarios and conditions while considering variability in hardware, models, test sites, and experimental settings.Standardized evaluation metrics focusing on safety, mobility, stability, and sustainability were established, enabling consistent assessment across different studies and applications. Analysis of the dataset revealed that current commercial AVs exhibit certain limitations, such as instability under specific conditions, highlighting the need for improved control strategies.To address the limitations of human-based car-following models, a three-stage unified dynamics structure was developed, accurately modeling AV dynamics and controllers by accounting for AV-specific control mechanisms and hardware constraints. The proposed structure integrates time delays due to sensor detection, data processing, and actuator response, and accommodates both linear and nonlinear controllers. This framework captures the uniform and predictable nature of AV control mechanisms.Within this dynamics structure, both linear and nonlinear car-following models were developed and calibrated using the empirical dataset. An optimization methodology was introduced to identify optimal parameters that maximize mobility while ensuring stability constraints are satisfied. The research methodology involved calibrating the car-following models based on the empirical data, identifying response delays for each AV, and performing stability analysis. For linear models, the system's transfer function was derived, and stability conditions were determined by analyzing the poles of the transfer function. For nonlinear models, local stability was assessed by linearizing the system around equilibrium points and examining the eigenvalues of the Jacobian matrix.Optimization problems were formulated to maximize mobility, such as minimizing average headway, while satisfying the derived stability constraints. Constrained optimization algorithms, including Sequential Quadratic Programming for linear models and Genetic Algorithms for nonlinear models, were employed to solve these problems. The optimized controllers demonstrated improved performance in simulations, achieving a better trade-off between mobility and stability compared to existing commercial models.Key findings of the dissertation include the importance of incorporating time delays in accurately modeling AV behaviors. Including feedback delays (sensor detection and data processing) and parasitic lags (actuator response times) significantly improved the accuracy of the car-following models. While both linear and nonlinear controllers were optimized to enhance performance, linear controllers generally performed better under the tested conditions, suggesting that commercially available AVs likely employ linear or piecewise-linear control mechanisms. The optimized controllers successfully achieved high mobility under stability constraints, mitigating traffic oscillations and promoting smoother traffic flow. This highlights the inherent trade-off between shorter headways, which improve mobility, and potential instability. Additionally, the local fit approach was found to be more suitable for short-term trajectory predictions, whereas the global fit approach was better for long-term simulations.To further advance vehicle automation, the research incorporated connected and automated vehicle (CAV) strategies. By leveraging cooperative driving automation (CDA) technologies, the proposed Visual-Enhanced Cooperative Traffic Operations (VECTOR) system demonstrated significant improvements in energy efficiency and traffic capacity. These enhancements address the limitations of traditional AV systems, by enabling lightweight intention-sharing mechanisms that facilitate more robust and efficient car-following behaviors.This dissertation contributes to the advancement of AV technology by providing a robust and uniform dataset that serves as a foundation for AV behavior analysis, standardized metrics that enable consistent evaluation across different studies and applications, a unified modeling framework that accurately reflects AV dynamics and control mechanisms, and improved control strategies that enhance AV performance in safety, mobility, stability, and sustainability.While significant strides were made, the research acknowledges certain limitations, such as the scope of the dataset and the exclusion of connected and cooperative strategies or end-to-end machine learning approaches. Future research directions include expanding the dataset to encompass more diverse AV models and driving conditions, exploring connected technologies and machine learning methods to further enhance AV performance, and implementing the optimized controllers in real-world AVs or high-fidelity simulators for validation under practical conditions.In conclusion, this dissertation addresses critical gaps in the modeling of AV behaviors, particularly in car-following scenarios. By developing a comprehensive framework that encompasses data collection, evaluation metrics, dynamic modeling, and controller optimization, it provides valuable insights and tools for improving AV performance. The research lays a solid foundation for future explorations into connected technologies and advanced control strategies, ultimately contributing to the effective integration of AVs into traffic systems and the realization of their potential benefits.
- 일반주제명
- Transportation
- 키워드
- Controllers
- 키워드
- Field test
- 키워드
- Trajectory data
- 기타저자
- The University of Wisconsin - Madison Civil & Environmental Engr
- 기본자료저록
- Dissertations Abstracts International. 86-06B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017165142
■00520250211153128
■006m o d
■007cr#unu||||||||
■020 ▼a9798346873976
■035 ▼a(MiAaPQ)AAI31767003
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a385
■1001 ▼aMa, Ke.
■24510▼aDesign, Testing, and Evaluation of Automated Vehicle Control Systems in Transportation
■260 ▼a[Sl]▼bThe University of Wisconsin - Madison▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a146 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-06, Section: B.
■500 ▼aIncludes supplementary digital materials.
■500 ▼aAdvisor: Li, Xiaopeng Shaw.
■5021 ▼aThesis (Ph.D.)--The University of Wisconsin - Madison, 2024.
■520 ▼aAutomated vehicles (AVs) hold the promise of transforming transportation systems by enhancing safety, mobility, stability, and sustainability. However, current research often falls short of fully capturing the unique characteristics of AV behaviors, particularly in car-following scenarios. Existing models predominantly focus on human-driven vehicles (HVs) and fail to account for the specific control mechanisms and hardware constraints inherent in AVs. This dissertation aims to bridge these gaps by developing a comprehensive framework that includes a uniform trajectory dataset, standardized evaluation metrics, and a unified dynamics structure tailored for AVs.The research is guided by four primary goals: developing a comprehensive and uniform AV trajectory dataset; assessing current commercial AV behaviors using standardized metrics; developing a unified dynamics structure for AV models and controllers; and designing and optimizing improved controllers within the established dynamics structure. To achieve these goals, a large-scale empirical dataset was compiled from 30 AVs across multiple automakers, capturing a wide range of driving scenarios and conditions while considering variability in hardware, models, test sites, and experimental settings.Standardized evaluation metrics focusing on safety, mobility, stability, and sustainability were established, enabling consistent assessment across different studies and applications. Analysis of the dataset revealed that current commercial AVs exhibit certain limitations, such as instability under specific conditions, highlighting the need for improved control strategies.To address the limitations of human-based car-following models, a three-stage unified dynamics structure was developed, accurately modeling AV dynamics and controllers by accounting for AV-specific control mechanisms and hardware constraints. The proposed structure integrates time delays due to sensor detection, data processing, and actuator response, and accommodates both linear and nonlinear controllers. This framework captures the uniform and predictable nature of AV control mechanisms.Within this dynamics structure, both linear and nonlinear car-following models were developed and calibrated using the empirical dataset. An optimization methodology was introduced to identify optimal parameters that maximize mobility while ensuring stability constraints are satisfied. The research methodology involved calibrating the car-following models based on the empirical data, identifying response delays for each AV, and performing stability analysis. For linear models, the system's transfer function was derived, and stability conditions were determined by analyzing the poles of the transfer function. For nonlinear models, local stability was assessed by linearizing the system around equilibrium points and examining the eigenvalues of the Jacobian matrix.Optimization problems were formulated to maximize mobility, such as minimizing average headway, while satisfying the derived stability constraints. Constrained optimization algorithms, including Sequential Quadratic Programming for linear models and Genetic Algorithms for nonlinear models, were employed to solve these problems. The optimized controllers demonstrated improved performance in simulations, achieving a better trade-off between mobility and stability compared to existing commercial models.Key findings of the dissertation include the importance of incorporating time delays in accurately modeling AV behaviors. Including feedback delays (sensor detection and data processing) and parasitic lags (actuator response times) significantly improved the accuracy of the car-following models. While both linear and nonlinear controllers were optimized to enhance performance, linear controllers generally performed better under the tested conditions, suggesting that commercially available AVs likely employ linear or piecewise-linear control mechanisms. The optimized controllers successfully achieved high mobility under stability constraints, mitigating traffic oscillations and promoting smoother traffic flow. This highlights the inherent trade-off between shorter headways, which improve mobility, and potential instability. Additionally, the local fit approach was found to be more suitable for short-term trajectory predictions, whereas the global fit approach was better for long-term simulations.To further advance vehicle automation, the research incorporated connected and automated vehicle (CAV) strategies. By leveraging cooperative driving automation (CDA) technologies, the proposed Visual-Enhanced Cooperative Traffic Operations (VECTOR) system demonstrated significant improvements in energy efficiency and traffic capacity. These enhancements address the limitations of traditional AV systems, by enabling lightweight intention-sharing mechanisms that facilitate more robust and efficient car-following behaviors.This dissertation contributes to the advancement of AV technology by providing a robust and uniform dataset that serves as a foundation for AV behavior analysis, standardized metrics that enable consistent evaluation across different studies and applications, a unified modeling framework that accurately reflects AV dynamics and control mechanisms, and improved control strategies that enhance AV performance in safety, mobility, stability, and sustainability.While significant strides were made, the research acknowledges certain limitations, such as the scope of the dataset and the exclusion of connected and cooperative strategies or end-to-end machine learning approaches. Future research directions include expanding the dataset to encompass more diverse AV models and driving conditions, exploring connected technologies and machine learning methods to further enhance AV performance, and implementing the optimized controllers in real-world AVs or high-fidelity simulators for validation under practical conditions.In conclusion, this dissertation addresses critical gaps in the modeling of AV behaviors, particularly in car-following scenarios. By developing a comprehensive framework that encompasses data collection, evaluation metrics, dynamic modeling, and controller optimization, it provides valuable insights and tools for improving AV performance. The research lays a solid foundation for future explorations into connected technologies and advanced control strategies, ultimately contributing to the effective integration of AVs into traffic systems and the realization of their potential benefits.
■590 ▼aSchool code: 0262.
■650 4▼aTransportation
■650 4▼aEnvironmental engineering
■653 ▼aAutomated vehicle
■653 ▼aConnected technology
■653 ▼aControllers
■653 ▼aField test
■653 ▼aPerformance metrics
■653 ▼aTrajectory data
■690 ▼a0709
■690 ▼a0543
■690 ▼a0775
■71020▼aThe University of Wisconsin - Madison▼bCivil & Environmental Engr.
■7730 ▼tDissertations Abstracts International▼g86-06B.
■790 ▼a0262
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17165142▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


