Design of Robot SLAM System Based on Deep Learning Feature Point Extraction
International Journal of Scientific Research in Science, Engineering and Technology · 09.08.2026 · 13 (4) · 278-283
DOI: 10.32628/ijsrset26134218 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Scientific Research in Science, Engineering and Technology, online erschienen am 09.08.2026. Verfasst von GU zhe-xiong. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
In response to the problem of ORB-SLAM2 system mistakenly matching two points with similar appearances but not corresponding points during ORB feature matching. This article optimizes the SLAM system algorithm through deep neural networks, replacing the ORB feature detection part in the SLAM system with this learning scheme. Geometric correspondences are generated based on the transformation of feature points in adjacent frames before and after, which are used to create visual odometry. Combining Convolutional Neural Network (CNN) with Recurrent Neural Network (RNN) for training, detecting the positions of key points and generating corresponding descriptors. The experimental results demonstrate. This algorithm can improve the accuracy of tracking pose (position and attitude) by 49.16% compared to ORB-SLAM2, and the improved SLAM system has better robustness. Compared to ORB-SLAM2 system, it can cope with more environmental features and is less likely to lose targets.
Zitieren
APA
zhe-xiong, G. (2026). Design of Robot SLAM System Based on Deep Learning Feature Point Extraction. International Journal of Scientific Research in Science, Engineering and Technology, 13(4), 278-283. https://doi.org/10.32628/ijsrset26134218
Vancouver
zhe-xiong G. Design of Robot SLAM System Based on Deep Learning Feature Point Extraction. International Journal of Scientific Research in Science, Engineering and Technology. 2026;13(4):278-283. doi:10.32628/ijsrset26134218
BibTeX
@article{zhexiong2026,
author = {zhe-xiong, GU},
title = {Design of Robot SLAM System Based on Deep Learning Feature Point Extraction},
journal = {International Journal of Scientific Research in Science, Engineering and Technology},
year = {2026},
volume = {13},
number = {4},
pages = {278--283},
doi = {10.32628/ijsrset26134218}
}
EUVIMED