Location and Map Reconstruction based on Monocular SLAM with CNN Learned Information

DOI

10.2991/iccia-19.2019.47How to use a DOI?

Keywords

monocular sensor; simultaneous localization and mapping; Convolutional Neural Network; Deep learning; feature matching; linear optimization.

Abstract

a method based on Convolutional Neural Network for depth prediction and monocular SLAM (simultaneous localization and mapping) is proposed for the problem of time-consuming and scale uncertainty. Firstly, the RGB image is extracted and matched, and the 3D information of SLAM feature points is obtained by the depth prediction of the convolutional neural network. Then the camera position is solved by the linear optimization. Finally, the motion trajectory and the three–dimensional dense point cloud are potted by loop closure and optimized global pose. Experimental results based on standard test set show that the method of information fusion based on convolutional neural network depth prediction and monocular SLAM can improve the accuracy of SLAM system mapping.

Copyright

© 2019, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Di Cui
AU  - Jianjun Fang
AU  - Dawei Luo
PY  - 2019/07
DA  - 2019/07
TI  - Location and Map Reconstruction based on Monocular SLAM with CNN Learned Information
BT  - Proceedings of the 3rd International Conference on Computer Engineering, Information Science & Application Technology (ICCIA 2019)
PB  - Atlantis Press
SP  - 308
EP  - 312
SN  - 2352-538X
UR  - https://doi.org/10.2991/iccia-19.2019.47
DO  - 10.2991/iccia-19.2019.47
ID  - Cui2019/07
ER  -