GPU-accelerated Localization in Confined Spaces using Deep Geometric Features

Rune Y. Brogaard; Ole Ravn; Evangelos Boukas

doi:10.1109/IST50367.2021.9651425

GPU-accelerated Localization in Confined Spaces using Deep Geometric Features

Rune Y. Brogaard, Ole Ravn, Evangelos Boukas

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

Navigating within dark and confined spaces require robotic platforms to utilize accurate and reliable localization systems to operate safely and unattended. This paper presents an absolute localization system, for known confined spaces, using state of the art 3D pointcloud descriptors. Local geometric features are extracted from a known map and registered to matching features visible in the robots field of view. The 3D registrations are motion-filtered and fused with a visual inertial odometry estimate in an Extended Kalman filter, which return a fast and accurate absolute pose estimate. The proposed localization system is tested with different deep learning feature descriptors in a structured confined space, and our results indicate greater accuracy and lower processing time when compared to mainstream 3D registration approaches.

Original language	English
Title of host publication	Proceedings of 2021 IEEE International Conference on Imaging Systems and Techniques
Number of pages	6
Publisher	IEEE
Publication date	26 Aug 2021
Pages	1-6
Article number	9651425
ISBN (Print)	978-1-7281-7372-6
DOIs	https://doi.org/10.1109/IST50367.2021.9651425
Publication status	Published - 26 Aug 2021
Event	2021 IEEE International Conference on Imaging Systems and Techniques - Virtual Conference, New York, United States Duration: 24 Aug 2021 → 26 Aug 2021

Conference

Conference	2021 IEEE International Conference on Imaging Systems and Techniques
Location	Virtual Conference
Country/Territory	United States
City	New York
Period	24/08/2021 → 26/08/2021

Keywords

Location awareness
Training
Visualization
Three-dimensional displays
Navigation
Training data
Feature extraction

Access to Document

10.1109/IST50367.2021.9651425

OpenUrl availability

Full text

Cite this

@inproceedings{53c7c24bd8c343c7bb04ae43064aa867,

title = "GPU-accelerated Localization in Confined Spaces using Deep Geometric Features",

abstract = "Navigating within dark and confined spaces require robotic platforms to utilize accurate and reliable localization systems to operate safely and unattended. This paper presents an absolute localization system, for known confined spaces, using state of the art 3D pointcloud descriptors. Local geometric features are extracted from a known map and registered to matching features visible in the robots field of view. The 3D registrations are motion-filtered and fused with a visual inertial odometry estimate in an Extended Kalman filter, which return a fast and accurate absolute pose estimate. The proposed localization system is tested with different deep learning feature descriptors in a structured confined space, and our results indicate greater accuracy and lower processing time when compared to mainstream 3D registration approaches.",

keywords = "Location awareness, Training, Visualization, Three-dimensional displays, Navigation, Training data, Feature extraction",

author = "Brogaard, {Rune Y.} and Ole Ravn and Evangelos Boukas",

year = "2021",

month = aug,

day = "26",

doi = "10.1109/IST50367.2021.9651425",

language = "English",

isbn = "978-1-7281-7372-6",

pages = "1--6",

booktitle = "Proceedings of 2021 IEEE International Conference on Imaging Systems and Techniques",

publisher = "IEEE",

address = "United States",

note = "2021 IEEE International Conference on Imaging Systems and Techniques, IST 2021 ; Conference date: 24-08-2021 Through 26-08-2021",

}

Brogaard, RY , Ravn, O & Boukas, E 2021, GPU-accelerated Localization in Confined Spaces using Deep Geometric Features. in Proceedings of 2021 IEEE International Conference on Imaging Systems and Techniques ., 9651425, IEEE, pp. 1-6, 2021 IEEE International Conference on Imaging Systems and Techniques, New York, New York, United States, 24/08/2021. https://doi.org/10.1109/IST50367.2021.9651425

TY - GEN

T1 - GPU-accelerated Localization in Confined Spaces using Deep Geometric Features

AU - Brogaard, Rune Y.

AU - Ravn, Ole

AU - Boukas, Evangelos

PY - 2021/8/26

Y1 - 2021/8/26

N2 - Navigating within dark and confined spaces require robotic platforms to utilize accurate and reliable localization systems to operate safely and unattended. This paper presents an absolute localization system, for known confined spaces, using state of the art 3D pointcloud descriptors. Local geometric features are extracted from a known map and registered to matching features visible in the robots field of view. The 3D registrations are motion-filtered and fused with a visual inertial odometry estimate in an Extended Kalman filter, which return a fast and accurate absolute pose estimate. The proposed localization system is tested with different deep learning feature descriptors in a structured confined space, and our results indicate greater accuracy and lower processing time when compared to mainstream 3D registration approaches.

AB - Navigating within dark and confined spaces require robotic platforms to utilize accurate and reliable localization systems to operate safely and unattended. This paper presents an absolute localization system, for known confined spaces, using state of the art 3D pointcloud descriptors. Local geometric features are extracted from a known map and registered to matching features visible in the robots field of view. The 3D registrations are motion-filtered and fused with a visual inertial odometry estimate in an Extended Kalman filter, which return a fast and accurate absolute pose estimate. The proposed localization system is tested with different deep learning feature descriptors in a structured confined space, and our results indicate greater accuracy and lower processing time when compared to mainstream 3D registration approaches.

KW - Location awareness

KW - Training

KW - Visualization

KW - Three-dimensional displays

KW - Navigation

KW - Training data

KW - Feature extraction

U2 - 10.1109/IST50367.2021.9651425

DO - 10.1109/IST50367.2021.9651425

M3 - Article in proceedings

SN - 978-1-7281-7372-6

SP - 1

EP - 6

BT - Proceedings of 2021 IEEE International Conference on Imaging Systems and Techniques

PB - IEEE

T2 - 2021 IEEE International Conference on Imaging Systems and Techniques

Y2 - 24 August 2021 through 26 August 2021

ER -

GPU-accelerated Localization in Confined Spaces using Deep Geometric Features

Abstract

Conference

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this