Exploiting the accuracy-acceleration tradeoff: VINS-assisted real-time object detection on moving systems

Betty Le Dem, Kazuo Nakazawa

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In recent years, Convolutional Neural Networks (CNNs) have repeatedly shown state-of-the-art performance for their accuracy in the task of object detection, but their heavy computational costs impede their ability for real-time detection when the supporting system is moving, particulary when it is accelerating. At the same time, recent progress on visual inertial systems takes great advantage of movement information to robustly estimate the robot state and its surrounding. This paper proposes to exploit the advantages of inertial odometry research for the purpose of real-time object detection system on mobile robots. We combine a CNN detector with VINS-Mono, a moving visual odometry system, and show reliable improvement in the detection process, especially when the robot accelerates or decelerates. Our system is ready-to-use in that it has very low deployment cost and requires no calibration. The resulting system allows for simultaneous robot state estimation and object detection, as well as object tracking. Lastly, this architecture proves to be flexible because not restrained to a specific object type or detector.

Original languageEnglish
Title of host publicationProceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages483-488
Number of pages6
ISBN (Electronic)9781728124933
DOIs
Publication statusPublished - 2019 Jul
Event2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019 - Hong Kong, China
Duration: 2019 Jul 82019 Jul 12

Publication series

NameIEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM
Volume2019-July

Conference

Conference2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019
CountryChina
CityHong Kong
Period19/7/819/7/12

Fingerprint

Robots
Detectors
Neural networks
State estimation
Mobile robots
Costs
Calibration
Object detection

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Control and Systems Engineering
  • Computer Science Applications
  • Software

Cite this

Dem, B. L., & Nakazawa, K. (2019). Exploiting the accuracy-acceleration tradeoff: VINS-assisted real-time object detection on moving systems. In Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019 (pp. 483-488). [8868536] (IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM; Vol. 2019-July). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/AIM.2019.8868536

Exploiting the accuracy-acceleration tradeoff : VINS-assisted real-time object detection on moving systems. / Dem, Betty Le; Nakazawa, Kazuo.

Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 483-488 8868536 (IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM; Vol. 2019-July).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Dem, BL & Nakazawa, K 2019, Exploiting the accuracy-acceleration tradeoff: VINS-assisted real-time object detection on moving systems. in Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019., 8868536, IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM, vol. 2019-July, Institute of Electrical and Electronics Engineers Inc., pp. 483-488, 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019, Hong Kong, China, 19/7/8. https://doi.org/10.1109/AIM.2019.8868536
Dem BL, Nakazawa K. Exploiting the accuracy-acceleration tradeoff: VINS-assisted real-time object detection on moving systems. In Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 483-488. 8868536. (IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM). https://doi.org/10.1109/AIM.2019.8868536
Dem, Betty Le ; Nakazawa, Kazuo. / Exploiting the accuracy-acceleration tradeoff : VINS-assisted real-time object detection on moving systems. Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 483-488 (IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM).
@inproceedings{10fa7fa63c9b4c75b328cf43ddef515e,
title = "Exploiting the accuracy-acceleration tradeoff: VINS-assisted real-time object detection on moving systems",
abstract = "In recent years, Convolutional Neural Networks (CNNs) have repeatedly shown state-of-the-art performance for their accuracy in the task of object detection, but their heavy computational costs impede their ability for real-time detection when the supporting system is moving, particulary when it is accelerating. At the same time, recent progress on visual inertial systems takes great advantage of movement information to robustly estimate the robot state and its surrounding. This paper proposes to exploit the advantages of inertial odometry research for the purpose of real-time object detection system on mobile robots. We combine a CNN detector with VINS-Mono, a moving visual odometry system, and show reliable improvement in the detection process, especially when the robot accelerates or decelerates. Our system is ready-to-use in that it has very low deployment cost and requires no calibration. The resulting system allows for simultaneous robot state estimation and object detection, as well as object tracking. Lastly, this architecture proves to be flexible because not restrained to a specific object type or detector.",
author = "Dem, {Betty Le} and Kazuo Nakazawa",
year = "2019",
month = "7",
doi = "10.1109/AIM.2019.8868536",
language = "English",
series = "IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "483--488",
booktitle = "Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019",

}

TY - GEN

T1 - Exploiting the accuracy-acceleration tradeoff

T2 - VINS-assisted real-time object detection on moving systems

AU - Dem, Betty Le

AU - Nakazawa, Kazuo

PY - 2019/7

Y1 - 2019/7

N2 - In recent years, Convolutional Neural Networks (CNNs) have repeatedly shown state-of-the-art performance for their accuracy in the task of object detection, but their heavy computational costs impede their ability for real-time detection when the supporting system is moving, particulary when it is accelerating. At the same time, recent progress on visual inertial systems takes great advantage of movement information to robustly estimate the robot state and its surrounding. This paper proposes to exploit the advantages of inertial odometry research for the purpose of real-time object detection system on mobile robots. We combine a CNN detector with VINS-Mono, a moving visual odometry system, and show reliable improvement in the detection process, especially when the robot accelerates or decelerates. Our system is ready-to-use in that it has very low deployment cost and requires no calibration. The resulting system allows for simultaneous robot state estimation and object detection, as well as object tracking. Lastly, this architecture proves to be flexible because not restrained to a specific object type or detector.

AB - In recent years, Convolutional Neural Networks (CNNs) have repeatedly shown state-of-the-art performance for their accuracy in the task of object detection, but their heavy computational costs impede their ability for real-time detection when the supporting system is moving, particulary when it is accelerating. At the same time, recent progress on visual inertial systems takes great advantage of movement information to robustly estimate the robot state and its surrounding. This paper proposes to exploit the advantages of inertial odometry research for the purpose of real-time object detection system on mobile robots. We combine a CNN detector with VINS-Mono, a moving visual odometry system, and show reliable improvement in the detection process, especially when the robot accelerates or decelerates. Our system is ready-to-use in that it has very low deployment cost and requires no calibration. The resulting system allows for simultaneous robot state estimation and object detection, as well as object tracking. Lastly, this architecture proves to be flexible because not restrained to a specific object type or detector.

UR - http://www.scopus.com/inward/record.url?scp=85074256828&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85074256828&partnerID=8YFLogxK

U2 - 10.1109/AIM.2019.8868536

DO - 10.1109/AIM.2019.8868536

M3 - Conference contribution

AN - SCOPUS:85074256828

T3 - IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM

SP - 483

EP - 488

BT - Proceedings of the 2019 IEEE/ASME International Conference on Advanced Intelligent Mechatronics, AIM 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -