Metka, Benjamin: Robust Visual Self-localization and Navigation in Outdoor Environments Using Slow Feature Analysis. 2019
Inhalt
- Introduction
- Localization, Mapping and Navigation
- Unsupervised Learning of Spatial Representations
- Principle of Slowness Learning
- Slow Feature Analysis
- Model for the Formation of Place and Head-Direction Cells
- Model Architecture and Training
- Analysis of the Learned Representations
- Data Recording and Ground Truth Acquisition
- Self-localization
- Validation of the Approach
- Localization in a Simulated Environment
- Localization in a Real World Environment
- The Impact of the Window Size
- Discussion
- Comparison to Visual Simultaneous Localization and Mapping Methods
- Image Acquisition and Preprocessing
- Experiments in an Indoor Environment
- Experiments in an Outdoor Environment
- Discussion
- Odometry Integration
- Unsupervised Metric Learning
- Fusion of SFA Estimates and Odometry in a Probabilistic Filter
- Discussion
- Landmark Based SFA-localization
- Experiments
- Localization With a Single Landmark
- Localization With Two Landmarks
- Localization With Two Landmarks and Occlusions
- Discussion
- Conclusion
- Robust Environmental Representations
- Robustness of Local Visual Features
- Learning Robust Representations with SFA
- Learning Short-term Invariant Representations
- Loop Closure Detection
- Training Using Feedback
- Experiments
- Experimental Setup
- Localization in a Static Environment
- Results
- Localization with Changing Light
- Results
- Localization with a Dynamic Object
- Results
- Localization Using Feedback from BoW Loop Closures
- Results
- Discussion
- Learning Long-term Invariant Representations
- Conclusion
- Navigation Using Slow Feature Gradients
- Navigation with Slow Feature Gradients
- Future Perspectives for Navigation in Slow Feature Space
- Navigation with Weighted Slow Feature Representations
- Weighting the Slow Feature Representations
- Navigation Experiments with Weighted Slow Feature Representations
- Discussion
- Implicit Optimization of Traveling Time
- Conclusion
- Summary and Conclusion
- Bibliography