Call number:
S 99.0139(371)
In:
Wissenschaftliche Arbeiten der Fachrichtung Geodäsie und Geoinformatik der Leibniz Universität Hannover
Type of Medium:
Series available for loan
Pages:
xi, 127 Seiten
,
Illustrationen, Diagramme
ISBN:
978-3-7696-5284-0
,
9783769652840
ISSN:
0174-1454
Series Statement:
Wissenschaftliche Arbeiten der Fachrichtung Geodäsie und Geoinformatik der Universität Hannover Nr. 371
Language:
English
Note:
Contents
1 Introduction
1.1 Shared Space
1.2 Motivation and Research Objectives
1.3 Proposals and Contributions
1.3.1 Interaction Detection
1.3.2 Trajectory Prediction
1.4 Structure of the Thesis
2 Concepts of Deep Learning and Fundamental Methods for Behavior Modeling
2.1 Introduction to Deep Learning
2.1.1 Feed-Forward Network with Backpropagation
2.1.2 Convolutional Neural Network
2.1.3 Recurrent Neural Network
2.2 Approaches for Object Detection and Classification
2.2.1 You Only Look Once
2.2.2 Multi-Level Feature Pyramid Network
2.3 Optical Flow
2.4 Spatial Clustering
2.5 Transformer Encoder with Self-Attention
2.6 Conditional Generative Model
2.6.1 Variational Auto-Encoder
2.6.2 Conditional Variational Auto-Encoder
3 Related Work
3.1 Interaction Detectio
3.1.1 Collisions, Conflicts, and Interactions
3.1.2 Automated Detection Using Computer Vision Methods
3.2 Trajectory Prediction
3.2.1 Expert vs. Data Driven
3.2.2 State-of-the-Art Deep Learning Approaches
4 Methodological Contributions
4.1 Interaction Detection
4.1.1 Problem Formulation and the Proposed Model
4.1.2 Sequence-to-Sequence Processing
4.1.3 Estimation of Uncertainty
4.1.4 Feature Extraction
4.2 Trajectory Prediction
4.2.1 Problem Formulation and the Proposed Model
4.2.2 Trajectory Ranking
4.2.3 Feature Extraction
5 Interaction Detection
5.1 Data Acquisition and Pre-processing
5.2 Experiments
5.2.1 Pipeline
5.2.2 CVAE Model for Interaction Detection
5.2.3 Baseline Model
5.2.4 Ablation Studies
5.2.5 Evaluation Metrics
5.3 Results
5.3.1 Quantitative Results
5.3.2 Qualitative Results
5.3.3 Analysis of the Results
5.4 Discussion
5.4.1 Failed Detection
5.4.2 Challenges of Cross-Dataset Generalization
5.5 Summary
6 Trajectory Prediction
6.1 Multi-Context Encoder Network
6.1.1 Framework
6.1.2 Experiments
6.1.3 Results
6.1.4 Discussion
6.1.5 Summary
6.2 Attentive Maps Encoder Network
6.2.1 Framework
6.2.2 Experiments
6.2.3 Results
6.2.4 Discussion
6.2.5 Summary
6.3 Dynamic Context Encoder Network
6.3.1 Framework
6.3.2 Experiments
6.3.3 Results
6.3.4 Discussion
6.3.5 Summary
7 Conclusion and Outlook
7.1 Conclusion
7.2 Outlook
List of Figures
List of Tables
Acronyms
Bibliography
Acknowledgements
Curriculum Vitae
,
Sprache der Kurzfassungen: Deutsch, Englisch
Location:
Lower compact magazine
Branch Library:
GFZ Library
Permalink