서브메뉴
검색
Safe And Efficient Variational Inference Model Predictive Control with Application to Aggressive Autonomous Driving
Safe And Efficient Variational Inference Model Predictive Control with Application to Aggressive Autonomous Driving
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105512
- ISBN
- 9798263330705
- DDC
- 796.72
- 저자명
- Yin, Ji.
- 서명/저자
- Safe And Efficient Variational Inference Model Predictive Control with Application to Aggressive Autonomous Driving
- 발행사항
- [Sl] : Georgia Institute of Technology, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 162 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: A.
- 주기사항
- Advisor: Tsiotras, Panagiotis.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2025.
- 초록/해제
- 요약Autonomous driving in aggressive, dynamic environments presents a unique challenge: balancing optimal performance with stringent safety requirements. Model Predictive Control (MPC) methods, including deterministic and stochastic variants, have been widely used for trajectory optimization and control. While many MPC approaches can handle nonlinear dynamics and constraints, they often rely on simplified models, assume specific forms of noise, or suffer from high computational costs. Furthermore, Variational Inference MPC (VIMPC) recasts control as a probabilistic inference problem, enabling a more flexible and scalable approach to trajectory optimization. For instance, Model Predictive Path Integral (MPPI) control, a type of VIMPC, focuses on achieving optimal trajectories through extensive forward simulations using general nonlinear dynamics to reduce simulation-to-reality gaps. However, existing VIMPC approaches demand substantial computational resources for real-time implementation and often lack both risk-awareness and formal safety guarantees. These limitations pose significant challenges in safety-critical applications, where robustness and reliability are essential.The key research question of this dissertation is: How can we design high-performing VIMPC controllers that ensure safety while maintaining computational feasibility? The significance of this research lies in its potential to advance the field of autonomous driving using robust controllers that can operate efficiently on limited computational resources. By incorporating techniques such as covariance steering to shape trajectory sampling distributions, integrating metrics like Conditional Value-at-Risk (CVaR) to account for risk, and employing formal methods such as control barrier functions (CBFs) to guarantee safety, the proposed VIMPC approaches enable real-time computation of optimal control in complex environments without compromising safety or performance. This research bridges the gap between theory and practice in the realm of MPC methods, and paves the way for safer, more reliable autonomous systems in high-risk scenarios.
- 일반주제명
- Automobile racing
- 일반주제명
- Robust control
- 일반주제명
- Failure
- 일반주제명
- Control algorithms
- 일반주제명
- Planning
- 일반주제명
- Autonomous vehicles
- 일반주제명
- Robots
- 일반주제명
- Design
- 일반주제명
- Visualization
- 일반주제명
- Robotics
- 일반주제명
- Transportation
- 기본자료저록
- Dissertations Abstracts International. 87-05A.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017360354
■00520260202105512
■006m o d
■007cr#unu||||||||
■020 ▼a9798263330705
■035 ▼a(MiAaPQ)AAI32308359
■035 ▼a(MiAaPQ)GeorgiaTech77825
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a796.72
■1001 ▼aYin, Ji.
■24510▼aSafe And Efficient Variational Inference Model Predictive Control with Application to Aggressive Autonomous Driving
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a162 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: A.
■500 ▼aAdvisor: Tsiotras, Panagiotis.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2025.
■520 ▼aAutonomous driving in aggressive, dynamic environments presents a unique challenge: balancing optimal performance with stringent safety requirements. Model Predictive Control (MPC) methods, including deterministic and stochastic variants, have been widely used for trajectory optimization and control. While many MPC approaches can handle nonlinear dynamics and constraints, they often rely on simplified models, assume specific forms of noise, or suffer from high computational costs. Furthermore, Variational Inference MPC (VIMPC) recasts control as a probabilistic inference problem, enabling a more flexible and scalable approach to trajectory optimization. For instance, Model Predictive Path Integral (MPPI) control, a type of VIMPC, focuses on achieving optimal trajectories through extensive forward simulations using general nonlinear dynamics to reduce simulation-to-reality gaps. However, existing VIMPC approaches demand substantial computational resources for real-time implementation and often lack both risk-awareness and formal safety guarantees. These limitations pose significant challenges in safety-critical applications, where robustness and reliability are essential.The key research question of this dissertation is: How can we design high-performing VIMPC controllers that ensure safety while maintaining computational feasibility? The significance of this research lies in its potential to advance the field of autonomous driving using robust controllers that can operate efficiently on limited computational resources. By incorporating techniques such as covariance steering to shape trajectory sampling distributions, integrating metrics like Conditional Value-at-Risk (CVaR) to account for risk, and employing formal methods such as control barrier functions (CBFs) to guarantee safety, the proposed VIMPC approaches enable real-time computation of optimal control in complex environments without compromising safety or performance. This research bridges the gap between theory and practice in the realm of MPC methods, and paves the way for safer, more reliable autonomous systems in high-risk scenarios.
■590 ▼aSchool code: 0078.
■650 4▼aAutomobile racing
■650 4▼aRobust control
■650 4▼aFailure
■650 4▼aControl algorithms
■650 4▼aPlanning
■650 4▼aAutonomous vehicles
■650 4▼aRobots
■650 4▼aDesign
■650 4▼aVisualization
■650 4▼aRobotics
■650 4▼aTransportation
■690 ▼a0771
■690 ▼a0389
■690 ▼a0709
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05A.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360354▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


