DeepSLAM: A Robust Monocular SLAM System with Unsupervised Deep Learning
- Submitting institution
-
The University of Essex
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 1461
- Type
- D - Journal article
- DOI
-
10.1109/tie.2020.2982096
- Title of journal
- IEEE Transactions on Industrial Electronics
- Article number
- -
- First page
- 3577
- Volume
- 68
- Issue
- 4
- ISSN
- 0278-0046
- Open access status
- Compliant
- Month of publication
- March
- Year of publication
- 2020
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
2
- Research group(s)
-
D - Robotics and Embedded Systems (RES)
- Citation count
- 0
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Published in IEEE-TIE, one of the top journals in the field, this paper reported a new visual SLAM technique based on a deep learning approach. It is significant for its novel use of deep learning in the topic and the work demonstrated a significant improvement on robustness over the state-of-the-art geometric based methods in some challenging real-world environments. The early version of the paper was presented at an IEEE conference and led to an invitation as a speaker in International Summer School of Deep Leaning in Poland 2019. A patent (WO/2019/180414) has resulted and commercialisation is ongoing.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -