서브메뉴
검색
상세정보
Spatiotemporal Occupancy Prediction for Autonomous Driving- [electronic resource]
Spatiotemporal Occupancy Prediction for Autonomous Driving- [electronic resource]
상세정보
- 자료유형
- 학위논문(국외)
- 자관 청구기호
- 기본표목-개인명
- 표제와 책임표시사항
- Spatiotemporal Occupancy Prediction for Autonomous Driving - [electronic resource] / Maneekwan Toyungyernsub
- 발행, 배포, 간사 사항
- 발행, 배포, 간사 사항
- 형태사항
- 1 online resource(p.89 )
- 일반주기
- Source: Dissertations Abstracts International, Volume: 85-03, Section: B.
- 일반주기
- Advisor: Kochenderfer, Mykel J.;Gao, Grace X.;Kennedy, Monroe.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2023.
- 이용제한주기
- This item must not be sold to any third party vendors.
- 요약 등 주기
- 요약Advancements in robotics, computer vision, machine learning and hardware have contributed to impressive developments of autonomous vehicles. However, there still exist challenges that must be tackled in order for the autonomous vehicles to be safely and seamlessly integrated into human environments. This is particularly the case in dense and cluttered urban settings. Autonomous vehicles must be able to understand and anticipate how their surroundings will evolve in both time and space. This capability will allow the autonomous vehicles to proactively plan safe trajectories and avoid other traffic agents.A common prediction approach is an agent-centric method (e.g., pedestrian or vehicle trajectory prediction). These methods require detection and tracking of all agents in the environment since trajectory prediction is performed on each agent. An alternative approach is a map-based (e.g., occupancy grid map) prediction method where the entire environment is discretized into grid cells and the collective occupancy probabilities for each grid cell are predicted. Hence, object detection and tracking capability is generally not needed. This makes a map-based occupancy prediction approach more robust to partial object occlusions and is capable of handling any arbitrary number of agents in the environments. However, a common problem with occupancy grid map prediction is the vanishing of objects from the predictions, especially at longer time horizons.In this thesis, we consider the problem of spatiotemporal environment prediction in urban environments. We propose an occupancy prediction framework that lever- ages meaningful environment information. In our first contribution, we develop a prediction framework that leverages environment dynamics. Our proposed model learns to predict the spatiotemporal evolution of the static and dynamic parts of the environment separately, and outputs the final OGM predictions of the entire envi- ronment. Compared to other repurposed state-of-the-art video-frame-prediction models, our method shows an improvement in prediction accuracy and a reduction in the vanishing object issue in the predictions.In our second contribution, we develop an occupancy prediction framework to address the limitation of previous work that requires static and dynamic parts of the environment to be known in advance. Our proposed modular framework consists of two modules, which are static-dynamic segmentation and occupancy prediction. The upstream static-dynamic segmentation module operates on the proposed map- centric occupancy-based representations to allow for direct integration between the two modules. Experiments conducted show a superior performance of our dynamics-aware occupancy prediction framework compared to baseline methods.Lastly, we develop an environment prediction model that leverages environment semantic information. Our prediction model consists of two sub-modules, which are the semantic grid map prediction and occupancy map prediction. Inspired by the success of incorporating environment dynamics in prior work, we propose to represent environment semantics in the form of semantic gird maps that are similar to the occupancy grid representation. This allows a direct flow of semantic information to the occupancy prediction sub-module. We compare our framework with two other methods, which are the dynamics-aware occupancy prediction, and a vanilla occupancy prediction. We find that our semantics-aware occupancy prediction framework outperforms the other two models.
- 주제명부출표목-일반주제명
- 주제명부출표목-일반주제명
- 주제명부출표목-일반주제명
- 부출표목-단체명
- 기본자료저록
- Dissertations Abstracts International. 85-03B.
- 기본자료저록
- Dissertation Abstract International
- 전자적 위치 및 접속
- 원문정보보기
- 소장사항
-
202402 2024
MARC
008240306s2023 s | s 0000c| eng d■001000016933798
■00520240214101338
■006m o d
■007cr
■020 ▼a9798380265041
■035 ▼a(MiAaPQ)AAI30561731
■035 ▼a(MiAaPQ)STANFORDrp366yd5112
■040 ▼aMiAaPQ▼cMiAaPQ
■08204▼a621.3
■090 ▼a전자도서(박사논문)
■1001 ▼aToyungyernsub, Maneekwan.
■24510▼aSpatiotemporal Occupancy Prediction for Autonomous Driving▼h[electronic resource]▼cManeekwan Toyungyernsub
■260 ▼a[S.l.]▼bStanford University. ▼c2023
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2023
■300 ▼a1 online resource(p.89 )
■500 ▼aSource: Dissertations Abstracts International, Volume: 85-03, Section: B.
■500 ▼aAdvisor: Kochenderfer, Mykel J.;Gao, Grace X.;Kennedy, Monroe.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2023.
■506 ▼aThis item must not be sold to any third party vendors.
■520 ▼aAdvancements in robotics, computer vision, machine learning and hardware have contributed to impressive developments of autonomous vehicles. However, there still exist challenges that must be tackled in order for the autonomous vehicles to be safely and seamlessly integrated into human environments. This is particularly the case in dense and cluttered urban settings. Autonomous vehicles must be able to understand and anticipate how their surroundings will evolve in both time and space. This capability will allow the autonomous vehicles to proactively plan safe trajectories and avoid other traffic agents.A common prediction approach is an agent-centric method (e.g., pedestrian or vehicle trajectory prediction). These methods require detection and tracking of all agents in the environment since trajectory prediction is performed on each agent. An alternative approach is a map-based (e.g., occupancy grid map) prediction method where the entire environment is discretized into grid cells and the collective occupancy probabilities for each grid cell are predicted. Hence, object detection and tracking capability is generally not needed. This makes a map-based occupancy prediction approach more robust to partial object occlusions and is capable of handling any arbitrary number of agents in the environments. However, a common problem with occupancy grid map prediction is the vanishing of objects from the predictions, especially at longer time horizons.In this thesis, we consider the problem of spatiotemporal environment prediction in urban environments. We propose an occupancy prediction framework that lever- ages meaningful environment information. In our first contribution, we develop a prediction framework that leverages environment dynamics. Our proposed model learns to predict the spatiotemporal evolution of the static and dynamic parts of the environment separately, and outputs the final OGM predictions of the entire envi- ronment. Compared to other repurposed state-of-the-art video-frame-prediction models, our method shows an improvement in prediction accuracy and a reduction in the vanishing object issue in the predictions.In our second contribution, we develop an occupancy prediction framework to address the limitation of previous work that requires static and dynamic parts of the environment to be known in advance. Our proposed modular framework consists of two modules, which are static-dynamic segmentation and occupancy prediction. The upstream static-dynamic segmentation module operates on the proposed map- centric occupancy-based representations to allow for direct integration between the two modules. Experiments conducted show a superior performance of our dynamics-aware occupancy prediction framework compared to baseline methods.Lastly, we develop an environment prediction model that leverages environment semantic information. Our prediction model consists of two sub-modules, which are the semantic grid map prediction and occupancy map prediction. Inspired by the success of incorporating environment dynamics in prior work, we propose to represent environment semantics in the form of semantic gird maps that are similar to the occupancy grid representation. This allows a direct flow of semantic information to the occupancy prediction sub-module. We compare our framework with two other methods, which are the dynamics-aware occupancy prediction, and a vanilla occupancy prediction. We find that our semantics-aware occupancy prediction framework outperforms the other two models.
■590 ▼aSchool code: 0212.
■650 4▼aComputer engineering.
■650 4▼aAutomotive engineering.
■650 4▼aCivil engineering.
■690 ▼a0543
■690 ▼a0464
■690 ▼a0540
■71020▼aStanford University.
■7730 ▼tDissertations Abstracts International▼g85-03B.
■773 ▼tDissertation Abstract International
■790 ▼a0212
■791 ▼aPh.D.
■792 ▼a2023
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T16933798▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
■980 ▼a202402▼f2024
![Spatiotemporal Occupancy Prediction for Autonomous Driving - [electronic resource] / Manee...](/Sponge/Images/bookDefaults/DDbookdefaultsmall.png)


한글
ENG
日本
中文
Việt Nam