서브메뉴
검색
Learning and Inference for Adaptable Manipulation Planning
Learning and Inference for Adaptable Manipulation Planning
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20250211153016
- ISBN
- 9798384045885
- DDC
- 629.8
- 저자명
- Power, Thomas J.
- 서명/저자
- Learning and Inference for Adaptable Manipulation Planning
- 발행사항
- [Sl] : University of Michigan, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 190 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 86-04, Section: B.
- 주기사항
- Advisor: Berenson, Dmitry.
- 학위논문주기
- Thesis (Ph.D.)--University of Michigan, 2024.
- 초록/해제
- 요약A central challenge for developing general-purpose robot assistants is the development of algorithms for robot manipulation that can perform a wide range of tasks across a diverse set of environments. In this thesis, I develop planning and trajectory optimization methods that can adapt to new and unforeseen systems. The key to these methods is the ability of robots to learn from experience and reason about related uncertainty. Using modern machine learning and approximate probabilistic inference techniques, the work I present in this thesis improves the ability of planning methods to do so. Probabilistic inference is useful in two ways. First, by using a probabilistic framing, probabilities can be used as a way of expressing confidence in our current models. I develop a method that learns to predict the uncertainty of a given dynamics model with a small amount of data collected online and avoids areas where the model is uncertain. I also propose an approach that learns a generative model of control sequences to complete a given task. I demonstrate that we can detect and adapt this generative model to situations where the environment differs from the training environments.Second, I incorporate probabilistic inference into the proposed methods by viewing planning itself as an inference problem. By framing planning as inference, we construct probability distributions over trajectories. This framework allows me to develop a method that views constrained trajectory optimization as inference, generating diverse sets of constraint-satisfying trajectories for completing manipulation tasks. This allows improved adaptation to online disturbances, since at any given time, there is a set of trajectories to select from. I demonstrate the effectiveness of this method on several different tasks, including a 7DoF manipulator turning a wrench and a 16DoF multi-fingered hand turning a precision screwdriver.The methods I present in this thesis contribute to the development of adaptable algorithms for robotic manipulation for the next generation of general-purpose robot assistants.
- 일반주제명
- Robotics
- 일반주제명
- Computer engineering
- 일반주제명
- Information technology
- 키워드
- Machine learning
- 기타저자
- University of Michigan Robotics
- 기본자료저록
- Dissertations Abstracts International. 86-04B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008250123s2024 us c eng d■001000017164554
■00520250211153016
■006m o d
■007cr#unu||||||||
■020 ▼a9798384045885
■035 ▼a(MiAaPQ)AAI31631529
■035 ▼a(MiAaPQ)umichrackham005840
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a629.8
■1001 ▼aPower, Thomas J.
■24510▼aLearning and Inference for Adaptable Manipulation Planning
■260 ▼a[Sl]▼bUniversity of Michigan▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a190 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 86-04, Section: B.
■500 ▼aAdvisor: Berenson, Dmitry.
■5021 ▼aThesis (Ph.D.)--University of Michigan, 2024.
■520 ▼aA central challenge for developing general-purpose robot assistants is the development of algorithms for robot manipulation that can perform a wide range of tasks across a diverse set of environments. In this thesis, I develop planning and trajectory optimization methods that can adapt to new and unforeseen systems. The key to these methods is the ability of robots to learn from experience and reason about related uncertainty. Using modern machine learning and approximate probabilistic inference techniques, the work I present in this thesis improves the ability of planning methods to do so. Probabilistic inference is useful in two ways. First, by using a probabilistic framing, probabilities can be used as a way of expressing confidence in our current models. I develop a method that learns to predict the uncertainty of a given dynamics model with a small amount of data collected online and avoids areas where the model is uncertain. I also propose an approach that learns a generative model of control sequences to complete a given task. I demonstrate that we can detect and adapt this generative model to situations where the environment differs from the training environments.Second, I incorporate probabilistic inference into the proposed methods by viewing planning itself as an inference problem. By framing planning as inference, we construct probability distributions over trajectories. This framework allows me to develop a method that views constrained trajectory optimization as inference, generating diverse sets of constraint-satisfying trajectories for completing manipulation tasks. This allows improved adaptation to online disturbances, since at any given time, there is a set of trajectories to select from. I demonstrate the effectiveness of this method on several different tasks, including a 7DoF manipulator turning a wrench and a 16DoF multi-fingered hand turning a precision screwdriver.The methods I present in this thesis contribute to the development of adaptable algorithms for robotic manipulation for the next generation of general-purpose robot assistants.
■590 ▼aSchool code: 0127.
■650 4▼aRobotics
■650 4▼aComputer engineering
■650 4▼aInformation technology
■653 ▼aMachine learning
■653 ▼aTrajectory optimization
■653 ▼aRobot manipulation
■653 ▼aProbability distributions
■653 ▼aControl sequences
■690 ▼a0771
■690 ▼a0800
■690 ▼a0489
■690 ▼a0464
■71020▼aUniversity of Michigan▼bRobotics.
■7730 ▼tDissertations Abstracts International▼g86-04B.
■790 ▼a0127
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17164554▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
