This book introduces several appearance-based place recognition pipelines based on different mapping techniques for addressing loop-closure detection in mobile platforms with limited computational resources. The motivation behind this book has been the prospect that in many contemporary applications efficient methods are needed that can provide high performance under run-time and memory constraints. Thus, three different mapping techniques for addressing the task of place recognition for simultaneous localization and mapping (SLAM) are presented. The book at hand follows a tutorial-based structure describing each of the main parts needed for a loop-closure detection pipeline to facilitate the newcomers. It mainly goes through a historical review of the problem, focusing on how it was addressed during the years reaching the current age. This way, the reader is initially familiarized with each part while the place recognition paradigms follow.
Author(s): Konstantinos A. Tsintotas, Loukas Bampis, Antonios Gasteratos
Series: Springer Tracts in Advanced Robotics, 133
Publisher: Springer
Year: 2022
Language: English
Pages: 124
City: Cham
Series Editor’s Foreword
Preface
Acknowledgements
Contents
1 The Revisiting Problem in Simultaneous Localization and Mapping
1.1 Foundation of Loop-Closure Detection
1.2 Simultaneous Localization and Mapping
1.2.1 Localization
1.2.2 Mapping
1.2.3 Sensing
1.3 Loop-Closure Detection Structure
1.3.1 Feature Extraction
1.3.2 Looking Behind
1.3.3 Decision Making
1.4 Placing the Presented Contribution Within the State-of-the-Art
References
2 Benchmarking
2.1 Evaluation Metrics
2.2 Datasets
2.2.1 Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) Vision Suite
2.2.2 Lip 6 Outdoor
2.2.3 EuRoC Machine Hall 05
2.2.4 Malaga 2009 Parking 6L
2.2.5 New College Vision Suite
2.2.6 The Oxford Dataset: City Centre
2.3 Reference Solutions
References
3 Probabilistic Appearance-Based Place Recognition Through Hierarchical Mapping
3.1 Methodology
3.1.1 Defining Sub-Maps
3.1.2 Assigning Visual Words to Places
3.1.3 Sub-Map Indexing
3.1.4 Images' Correspondence
3.2 Experimental Results
3.2.1 Performance Evaluation
3.2.2 System's Response
3.2.3 Comparative Results
References
4 Dynamic Places' Definition for Sequence-Based Visual Place Recognition
4.1 Methodology
4.1.1 Efficient Places' Definition Through Point Tracking
4.1.2 Images' Modulation
4.1.3 Place-to-Place Association
4.1.4 Local Best Match
4.2 Experimental Results
4.2.1 DOSeqSLAM
4.2.2 Parameters' Discussion
4.2.3 Performance Evaluation
4.2.4 System's Response
4.2.5 Comparative Results
References
5 Modest-Vocabulary Loop-Closure Detection with Incremental Bag of Tracked Words
5.1 Methodology
5.1.1 Bag of Tracked Words
5.1.2 Probabilistic Loop-Closure Detection Pipeline
5.2 Experimental Results
5.2.1 Parameters' Discussion
5.2.2 Performance Evaluation
5.2.3 System's Response
5.2.4 Comparative Results
References
6 Open Challenges and Conclusion
6.1 Dynamic Environments
6.1.1 Robust Visual Representations
6.1.2 Learning and Predicting the Appearance Changes
6.2 Viewpoint Variations
6.3 Map Management and Storage Requirements
6.3.1 Key-Frame Selection
6.3.2 Representing Each Node in a Sparse Map by a Group of Sequential and Visually Similar Images
6.3.3 Short-Memory Scale Discretization
6.4 Computational Complexity
6.5 Conclusion
References
