Publication: Comparison of ROS-Based Monocular Visual SLAM Methods: DSO, LDSO, ORB-SLAM2 and DynaSLAM
Issued Date
2020-01-01
Resource Type
ISSN
16113349
03029743
03029743
Other identifier(s)
2-s2.0-85092904826
Rights
Mahidol University
Rights Holder(s)
SCOPUS
Bibliographic Citation
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol.12336 LNAI, (2020), 222-233
Suggested Citation
Eldar Mingachev, Roman Lavrenov, Tatyana Tsoy, Fumitoshi Matsuno, Mikhail Svinin, Jackrit Suthakorn, Evgeni Magid Comparison of ROS-Based Monocular Visual SLAM Methods: DSO, LDSO, ORB-SLAM2 and DynaSLAM. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol.12336 LNAI, (2020), 222-233. doi:10.1007/978-3-030-60337-3_22 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/59952
Research Projects
Organizational Units
Authors
Journal Issue
Thesis
Title
Comparison of ROS-Based Monocular Visual SLAM Methods: DSO, LDSO, ORB-SLAM2 and DynaSLAM
Abstract
© 2020, Springer Nature Switzerland AG. Stable and robust path planning of a ground mobile robot requires a combination of accuracy and low latency in its state estimation. Yet, state estimation algorithms should provide these under computational and power constraints of a robot embedded hardware. The presented study offers a comparative analysis of four cutting edge publicly available within robot operating system (ROS) monocular simultaneous localization and mapping methods: DSO, LDSO, ORB-SLAM2, and DynaSLAM. The analysis considers pose estimation accuracy (alignment, absolute trajectory, and relative pose root mean square error) and trajectory precision of the four methods at TUM-Mono and EuRoC datasets.