서브메뉴
검색
A Haptic Shared Control Framework Leveraging High-Performance Trajectory Planning and Control to Teach Advanced Driving Skills
A Haptic Shared Control Framework Leveraging High-Performance Trajectory Planning and Control to Teach Advanced Driving Skills
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105240
- ISBN
- 9798291568712
- DDC
- 621
- 저자명
- Yu, Siyuan.
- 서명/저자
- A Haptic Shared Control Framework Leveraging High-Performance Trajectory Planning and Control to Teach Advanced Driving Skills
- 발행사항
- [Sl] : University of Michigan, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 188 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-02, Section: B.
- 주기사항
- Advisor: Ersal, Tulga.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2025.
- 초록/해제
- 요약Recent advancements in autonomous driving technologies have showcased remarkable success in performing routine driving tasks, yet human drivers remain the primary operators behind the wheel and bear full responsibility for vehicle operations. When faced with safety-critical situations or scenarios demanding advanced driving skills, the responsibility predominantly falls on human operators, whose expertise in everyday driving often does not translate to high-performance or complex maneuvers. To address this critical need, this dissertation aims to develop a haptic shared control framework that leverages high-performance trajectory planning and control to teach advanced driving skills to human drivers. For such a framework to effectively teach advanced skills, the autonomy system must itself possess such skills. This requires of a real-time planning and control framework that integrates key vehicle dynamics and operates effectively at the limits of Handling, even in highly dynamic and complex scenarios. However, state-of-the-art trajectory planning algorithms face significant challenges, including scalability limitations, a lack of generality for complex scenarios, and an over-reliance on predefined paths, all of which hinder their practical applicability in this problem domain. On the other hand, while literature demonstrates the potential of haptic shared control frameworks, their applications have been limited to achieving short-term benefits or teaching long-term skills in relatively simple tasks. This dissertation seeks to bridge this gap by developing a haptic shared control framework designed to teach human drivers advanced driving skills in high-performance scenarios.To accomplish the overarching objective, this dissertation pursues four key goals. First, it addresses the limitations of current vehicle dynamics models used in optimization-based control, which often struggle to balance model fidelity, computational efficiency, and continuous differentiability. A tailored vehicle model is developed to accurately capture the key dynamics excited near the limits of handling while remaining well-suited for optimization algorithms. Second, it overcomes the scalability challenges and restrictive assumptions-such as constant-speed operation-commonly found in existing drivable tube based algorithms. A twice continuously differentiable drivable tube constraint is formulated to represent collision-free regions without being overly-conservative to shrink the vehicle's operational space. Third, a unified reference-free trajectory planning and control algorithm is developed that integrates the proposed vehicle dynamics model and collision avoidance formulation. This algorithm is enable advanced vehicle planning and control in both simulation and real-world environments. Finally, the autonomy system is incorporated into a haptic shared control framework aimed at training human drivers in advanced driving skills. A set of performance metrics and a structured training scheme is also developed to evaluate its effectiveness.Therefore, this dissertation makes the following original contributions:1. Development of an efficient vehicle dynamics model that captures critical dynamics under extreme conditions, specifically tailored for optimization-based control techniques.2. Development of conservative twice continuous differentiable safety constraints using envelopes to mathematically describe the drivable regions.3. Development of a single-level real-time scalable trajectory planning and control algorithm using envelope-based model predictive control.4. Development of a haptic shared control framework for teaching advanced driving skills by smoothly transitioning control authority based on human performance.
- 일반주제명
- Mechanical engineering
- 일반주제명
- Engineering
- 일반주제명
- Automotive engineering
- 일반주제명
- Robotics
- 키워드
- Driving skills
- 기타저자
- University of Michigan Mechanical Engineering
- 기본자료저록
- Dissertations Abstracts International. 87-02B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017359949
■00520260202105240
■006m o d
■007cr#unu||||||||
■020 ▼a9798291568712
■035 ▼a(MiAaPQ)AAI32271993
■035 ▼a(MiAaPQ)umichrackham006348
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a621
■1001 ▼aYu, Siyuan.
■24512▼aA Haptic Shared Control Framework Leveraging High-Performance Trajectory Planning and Control to Teach Advanced Driving Skills
■260 ▼a[Sl]▼bUniversity of Michigan▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a188 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-02, Section: B.
■500 ▼aAdvisor: Ersal, Tulga.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2025.
■520 ▼aRecent advancements in autonomous driving technologies have showcased remarkable success in performing routine driving tasks, yet human drivers remain the primary operators behind the wheel and bear full responsibility for vehicle operations. When faced with safety-critical situations or scenarios demanding advanced driving skills, the responsibility predominantly falls on human operators, whose expertise in everyday driving often does not translate to high-performance or complex maneuvers. To address this critical need, this dissertation aims to develop a haptic shared control framework that leverages high-performance trajectory planning and control to teach advanced driving skills to human drivers. For such a framework to effectively teach advanced skills, the autonomy system must itself possess such skills. This requires of a real-time planning and control framework that integrates key vehicle dynamics and operates effectively at the limits of Handling, even in highly dynamic and complex scenarios. However, state-of-the-art trajectory planning algorithms face significant challenges, including scalability limitations, a lack of generality for complex scenarios, and an over-reliance on predefined paths, all of which hinder their practical applicability in this problem domain. On the other hand, while literature demonstrates the potential of haptic shared control frameworks, their applications have been limited to achieving short-term benefits or teaching long-term skills in relatively simple tasks. This dissertation seeks to bridge this gap by developing a haptic shared control framework designed to teach human drivers advanced driving skills in high-performance scenarios.To accomplish the overarching objective, this dissertation pursues four key goals. First, it addresses the limitations of current vehicle dynamics models used in optimization-based control, which often struggle to balance model fidelity, computational efficiency, and continuous differentiability. A tailored vehicle model is developed to accurately capture the key dynamics excited near the limits of handling while remaining well-suited for optimization algorithms. Second, it overcomes the scalability challenges and restrictive assumptions-such as constant-speed operation-commonly found in existing drivable tube based algorithms. A twice continuously differentiable drivable tube constraint is formulated to represent collision-free regions without being overly-conservative to shrink the vehicle's operational space. Third, a unified reference-free trajectory planning and control algorithm is developed that integrates the proposed vehicle dynamics model and collision avoidance formulation. This algorithm is enable advanced vehicle planning and control in both simulation and real-world environments. Finally, the autonomy system is incorporated into a haptic shared control framework aimed at training human drivers in advanced driving skills. A set of performance metrics and a structured training scheme is also developed to evaluate its effectiveness.Therefore, this dissertation makes the following original contributions:1. Development of an efficient vehicle dynamics model that captures critical dynamics under extreme conditions, specifically tailored for optimization-based control techniques.2. Development of conservative twice continuous differentiable safety constraints using envelopes to mathematically describe the drivable regions.3. Development of a single-level real-time scalable trajectory planning and control algorithm using envelope-based model predictive control.4. Development of a haptic shared control framework for teaching advanced driving skills by smoothly transitioning control authority based on human performance.
■590 ▼aSchool code: 0127.
■650 4▼aMechanical engineering
■650 4▼aEngineering
■650 4▼aAutomotive engineering
■650 4▼aRobotics
■653 ▼aAutonomous racing
■653 ▼aNonlinear model predictive control
■653 ▼aHigh performance driving
■653 ▼aHaptic shared control
■653 ▼aDriving skills
■690 ▼a0548
■690 ▼a0537
■690 ▼a0540
■690 ▼a0771
■71020▼aUniversity of Michigan▼bMechanical Engineering.
■7730 ▼tDissertations Abstracts International▼g87-02B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17359949▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


