Optimizing Visual SLAM for Robust and Scalable Mobile Robot Navigation

Kaiyu Xu

2025

Abstract

The application of vision Simultaneous Localization and Map Building (SLAM) to robot navigation is an important field. But the main challenges faced in this field are how to improve robustness as well as scalability. This thesis aims to optimize the robustness and scalability of vision SLAM in robot navigation. In this paper, it mprove robustness and scalability from both deep learning and image processing. In deep learning, this paper uses Convolutional Neural Network (CNN) algorithm for feature matching and constructs Proximal Policy Optimization (PPO) algorithm model for training. In image processing, this paper mainly uses Oriented FAST and Rotated BRIEF (ORB) technique for feature extraction and loopback detection. The result is that all four experiments are well optimized for robustness and scalability. This allows the robot to operate in complex environments without interference from other external factors, such as lighting conditions and dynamic environments. It also allows the robot to adapt to larger environments and to take on the computational load required to handle larger environments. Ultimately, it is hoped that these methods can be applied to real-life robot navigation.

Paper Citation

in Harvard Style

Xu K. (2025). Optimizing Visual SLAM for Robust and Scalable Mobile Robot Navigation. In Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI; ISBN 978-989-758-792-4, SciTePress, pages 43-48. DOI: 10.5220/0014307300004718

in Bibtex Style

@conference{emiti25,
author={Kaiyu Xu},
title={Optimizing Visual SLAM for Robust and Scalable Mobile Robot Navigation},
booktitle={Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI},
year={2025},
pages={43-48},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0014307300004718},
isbn={978-989-758-792-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 2nd International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI
TI - Optimizing Visual SLAM for Robust and Scalable Mobile Robot Navigation
SN - 978-989-758-792-4
AU - Xu K.
PY - 2025
SP - 43
EP - 48
DO - 10.5220/0014307300004718
PB - SciTePress