Real-time visual SLAM on embedded platforms

Real-time visual SLAM on embedded platforms

ORB-SLAM2 is a visual based SLAM (Simultaneous Localization and Mapping) application that estimates the trajectory of a moving object and at the same time, creates a map of the environment. This allows for good indoor spatial tracking as well as movement tracking. However, the robustness of ORB-SLAM...

Full description

Saved in:
Bibliographic Details
Main Author: Chai, Boon Hui
Other Authors: Lam Siew Kei
Format: Final Year Project
Language:English
Published: Nanyang Technological University 2020
Subjects:
Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
Online Access:https://hdl.handle.net/10356/137898
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:ORB-SLAM2 is a visual based SLAM (Simultaneous Localization and Mapping) application that estimates the trajectory of a moving object and at the same time, creates a map of the environment. This allows for good indoor spatial tracking as well as movement tracking. However, the robustness of ORB-SLAM2 is degraded in the presence of moving objects in the scene as the pixels that are selected for referencing between image frames may belong to these dynamic objects. This problem can be mitigated by using semantic segmentation to distinguish between static and dynamic objects in the scene and allow only the pixels of static objects to be used for referencing between image frames. The purpose of this report is to evaluate the performance of combining ORB-SLAM2 and semantic segmentation on an embedded platform.

Similar Items