Design exploration of visual localization algorithms

• Main Author: Siow, Keith Kee Tat
• Other Authors: Lam Siew Kei
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2023
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/166139
• Institution: Nanyang Technological University

SLAM can be defined as two parts, navigating in an unknown environment (Localisation) and tracking the position of the sensors or camera through different positions and different times in that environment. Currently, the most popular method for localization is through the use of a Global-Positioning-System (GPS) sensor. However, the usage of GPS is only effective when there is a clear view of the sky and navigation indoors causes huge amounts of error, proving it to be ineffective. Visual-SLAM provides a means for localization and mapping to be done by utilizing visual information through the usage of a camera, attached existingly or added on. This project proposes the design and development of a modular Visual-SLAM Graphical User Interface to visualize the different Visual-SLAM approaches that can be implemented in real-time for the localization applications for robots or evaluate the simulation. The project is designed on QT framework and a modified ORB-SLAM 2 Visual-SLAM system. Preliminary and post development user study was conducted to find out expectations and evaluate if the expectations have been met. The GUI proved to provide a better learning experience for users after a comparison with the original Visual-SLAM and is defined as successful after meeting its primary goals.