GPS-SLAM: An Augmentation of the ORB-SLAM Algorithm.

Dániel Kiss-Illés,Esther Salamí,Cristina Barrado

doi:10.3390/s19224973

Abstract

This work presents Global Positioning System-Simultaneous Localization and Mapping (GPS-SLAM), an augmented version of Oriented FAST (Features from accelerated segment test) and Rotated BRIEF (Binary Robust Independent Elementary Features) feature detector (ORB)-SLAM using GPS and inertial data to make the algorithm capable of dealing with low frame rate datasets. In general, SLAM systems are successful in case of datasets with a high frame rate. This work was motivated by a scarce dataset where ORB-SLAM often loses track because of the lack of continuity. The main work includes the determination of the next frame’s pose based on the GPS and inertial data. The results show that this additional information makes the algorithm more robust. As many large, outdoor unmanned aerial vehicle (UAV) flights save the GPS and inertial measurement unit (IMU) data of the capturing of images, this program gives an option to use the SLAM algorithm successfully even if the dataset has a low frame rate.

Highlights

There are a lot of algorithms using the basic idea of Simultaneous Localization and Mapping (SLAM) [1,2,3]
Summarizing the results, we can state that the goals of the augmentation of the ORB-SLAM algorithm were reached
Starting from a partly successful tracking of the olive tree dataset by the original version, we successfully changed the functionalities of the code in a way that the tracking by Global Positioning System (GPS)-SLAM became more robust and of better quality

Summary

Introduction

There are a lot of algorithms using the basic idea of Simultaneous Localization and Mapping (SLAM) [1,2,3]. In these methods, both the localization and the map creation are made by the algorithm itself. There is additional information which was taken during the capture of the images. Satellite System (GNSS) data about the position of the camera and the inertial measurement unit (IMU). The modified algorithm got the name GPS-SLAM since the most famous system of GNSS is the Global Positioning System (GPS).

Methods

Results

Conclusion