서브메뉴
검색
Robust Autonomous Navigation Framework for Exploration in GPS-Absent and Challenging Environment
Robust Autonomous Navigation Framework for Exploration in GPS-Absent and Challenging Environment
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105515
- ISBN
- 9798263340797
- DDC
- 629.4
- 저자명
- Chen, Mengzhen.
- 서명/저자
- Robust Autonomous Navigation Framework for Exploration in GPS-Absent and Challenging Environment
- 발행사항
- [Sl] : Georgia Institute of Technology, 2024
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2024
- 형태사항
- 347 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Mavris, Dimitri N.
- 학위논문주기
- Thesis (Ph.D.)--Georgia Institute of Technology, 2024.
- 초록/해제
- 요약The benefits of autonomous systems have attracted the industry's attention during thepast decade. Different kinds of autonomous systems have been applied to various fieldssuch as transportation, agriculture, healthcare, etc. Tasks unable or risky to be completedby humans alone can now be handled by autonomous systems efficiently, and the laborcost has been greatly reduced. Among various kinds of tasks that an autonomous systemcan perform, the capability of an autonomous system to understand its surrounding environment is of great importance. Either using an Unmanned Aircraft System (UAS) forpackage delivery or self-driving vehicles requires the autonomous system to be more robust during operation under different scenarios. This work will improve the robustness ofautonomous systems under challenging and GPS-absent environments.When exploring an unknown environment, if external information such as a GPS signalis unavailable, mapping and localization are equally important and complementary. Therefore, simultaneously creating a map and localizing itself is essential. Under such conditions, Simultaneous Localization and Mapping (SLAM) was created in the robotics community to provide the capability of building a map for the surroundings of an autonomoussystem and localizing itself during operation. SLAM architecture has been designed fordifferent kinds of sensors and scenarios during the past several decades. Among differentSLAM categories, visual SLAM, which uses cameras as the sensors, outperforms others.It has the advantage of extracting rich information from images while other sensors aloneare incapable. Since the images captured by the camera are treated as the inputs, therefore,the accuracy of the results will heavily depend on their quality. Most SLAM architecturecan easily handle high-quality images or video streams, while poor-quality ones are stillchallenging. The first challenging scenario that the visual SLAM is facing is the motionblur scenario in which the performance of the visual SLAM will be severely downgraded.The other challenging scenario that the visual SLAM is facing is the low-light environment.Since the poor illumination condition has less information shared with the camera, it alsodowngrades the accuracy of the visual SLAM system. Furthermore, the visual SLAM addsan extra requirement for computational efficiency since the operation needs to be real-time.Based on these observations, the research objective of this dissertation has been formedwhich is improving the visual SLAM performance under these two challenging conditions.In this dissertation, three research areas have been defined to achieve the overarching research objective. The first research area focuses on developing the capabilities of recovering these poor-quality images captured under these challenging scenarios in real-time. Twohighly efficient deep learning models, a single image deblurring model, and a low-lightimage enhancement model, have been developed and evaluated in this dissertation. Thesecond research area focuses on the uncertainty quantification for the results generated bythe visual SLAM systems. Since some of the visual SLAM systems have nondeterministicbehaviors, a statistical approach has been developed in this dissertation to reduce and factor out the uncertainties in the results and provide a quantitative method for performanceevaluation. The third research area focuses on creating a visual SLAM validation datasetthat can be utilized for testing the performance under motion blur scenarios since the majority of the existing dataset does not have enough blurriness or is limited to the indoorenvironment. In this dissertation, a synthetic blurry SLAM dataset has been created withthe help of utilizing a physics-based virtual simulation environment. From a combinationof the three research areas, a visual SLAM framework is proposed and tested with severalvisual SLAM datasets captured under the two challenging scenarios. Based on the experiment results, for the proposed visual SLAM framework, accuracy improvements have beenobserved through a statistical approach for all the use cases when compared with the benchmark visual SLAM system. Therefore, the proposed visual SLAM framework in which theimage enhancement modules have been added does improve the visual SLAM performanceunder challenging conditions.Two key contributions have been made through work:A visual SLAM framework that is designed for tackling real-world challenging conditions such as motion blur and low-light environment.A novel pipeline that utilizes the physics-based simulation environment to generatea realistic synthetic blurry visual SLAM dataset.
- 일반주제명
- Space exploration
- 일반주제명
- Deep learning
- 일반주제명
- Navigation systems
- 일반주제명
- Neural networks
- 일반주제명
- Aerospace engineering
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2024 us c eng d■001000017360375
■00520260202105515
■006m o d
■007cr#unu||||||||
■020 ▼a9798263340797
■035 ▼a(MiAaPQ)AAI32309285
■035 ▼a(MiAaPQ)GeorgiaTech75250
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a629.4
■1001 ▼aChen, Mengzhen.
■24510▼aRobust Autonomous Navigation Framework for Exploration in GPS-Absent and Challenging Environment
■260 ▼a[Sl]▼bGeorgia Institute of Technology▼c2024
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2024
■300 ▼a347 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Mavris, Dimitri N.
■5021 ▼aThesis (Ph.D.)--Georgia Institute of Technology, 2024.
■520 ▼aThe benefits of autonomous systems have attracted the industry's attention during thepast decade. Different kinds of autonomous systems have been applied to various fieldssuch as transportation, agriculture, healthcare, etc. Tasks unable or risky to be completedby humans alone can now be handled by autonomous systems efficiently, and the laborcost has been greatly reduced. Among various kinds of tasks that an autonomous systemcan perform, the capability of an autonomous system to understand its surrounding environment is of great importance. Either using an Unmanned Aircraft System (UAS) forpackage delivery or self-driving vehicles requires the autonomous system to be more robust during operation under different scenarios. This work will improve the robustness ofautonomous systems under challenging and GPS-absent environments.When exploring an unknown environment, if external information such as a GPS signalis unavailable, mapping and localization are equally important and complementary. Therefore, simultaneously creating a map and localizing itself is essential. Under such conditions, Simultaneous Localization and Mapping (SLAM) was created in the robotics community to provide the capability of building a map for the surroundings of an autonomoussystem and localizing itself during operation. SLAM architecture has been designed fordifferent kinds of sensors and scenarios during the past several decades. Among differentSLAM categories, visual SLAM, which uses cameras as the sensors, outperforms others.It has the advantage of extracting rich information from images while other sensors aloneare incapable. Since the images captured by the camera are treated as the inputs, therefore,the accuracy of the results will heavily depend on their quality. Most SLAM architecturecan easily handle high-quality images or video streams, while poor-quality ones are stillchallenging. The first challenging scenario that the visual SLAM is facing is the motionblur scenario in which the performance of the visual SLAM will be severely downgraded.The other challenging scenario that the visual SLAM is facing is the low-light environment.Since the poor illumination condition has less information shared with the camera, it alsodowngrades the accuracy of the visual SLAM system. Furthermore, the visual SLAM addsan extra requirement for computational efficiency since the operation needs to be real-time.Based on these observations, the research objective of this dissertation has been formedwhich is improving the visual SLAM performance under these two challenging conditions.In this dissertation, three research areas have been defined to achieve the overarching research objective. The first research area focuses on developing the capabilities of recovering these poor-quality images captured under these challenging scenarios in real-time. Twohighly efficient deep learning models, a single image deblurring model, and a low-lightimage enhancement model, have been developed and evaluated in this dissertation. Thesecond research area focuses on the uncertainty quantification for the results generated bythe visual SLAM systems. Since some of the visual SLAM systems have nondeterministicbehaviors, a statistical approach has been developed in this dissertation to reduce and factor out the uncertainties in the results and provide a quantitative method for performanceevaluation. The third research area focuses on creating a visual SLAM validation datasetthat can be utilized for testing the performance under motion blur scenarios since the majority of the existing dataset does not have enough blurriness or is limited to the indoorenvironment. In this dissertation, a synthetic blurry SLAM dataset has been created withthe help of utilizing a physics-based virtual simulation environment. From a combinationof the three research areas, a visual SLAM framework is proposed and tested with severalvisual SLAM datasets captured under the two challenging scenarios. Based on the experiment results, for the proposed visual SLAM framework, accuracy improvements have beenobserved through a statistical approach for all the use cases when compared with the benchmark visual SLAM system. Therefore, the proposed visual SLAM framework in which theimage enhancement modules have been added does improve the visual SLAM performanceunder challenging conditions.Two key contributions have been made through work:A visual SLAM framework that is designed for tackling real-world challenging conditions such as motion blur and low-light environment.A novel pipeline that utilizes the physics-based simulation environment to generatea realistic synthetic blurry visual SLAM dataset.
■590 ▼aSchool code: 0078.
■650 4▼aSpace exploration
■650 4▼aDeep learning
■650 4▼aNavigation systems
■650 4▼aNeural networks
■650 4▼aAerospace engineering
■690 ▼a0538
■690 ▼a0800
■71020▼aGeorgia Institute of Technology.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0078
■791 ▼aPh.D.
■792 ▼a2024
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360375▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.


