An examination of feature detection for real-time visual odometry in untextured natural terrain

Kyohei Otsu, Masatsugu Otsuki, Genya Ishigami, Takashi Kubota

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

Abstract

Estimating the position of a robot is an essential requirement for autonomous mobile robots. Visual Odometry is a promising localization method in slippery natural terrain, which drastically degrades the accuracy of Wheel Odometry, while relying neither on other infrastructure nor any prior knowledge. Visual Odometry, however, suffers from the instability of feature extraction from the untextured natural terrain. To date, a number of feature detectors have been proposed for stable feature detection. This paper compares commonly used detectors in terms of robustness, localization accuracy and computational efficiency, and points out their trade-off problems among those criteria. To solve the problem, a hybrid algorithm is proposed which dynamically switches between multiple detectors according to the texture of terrain. Validity of the algorithm is proved by the simulation using dataset at volcanic areas in Japan.

Original languageEnglish
Title of host publicationAdvances in Intelligent Systems and Computing
Pages405-414
Number of pages10
Volume208 AISC
DOIs
Publication statusPublished - 2013
Externally publishedYes
Event1st International Conference on Robot Intelligence Technology and Applications, RiTA 2012 - Gwangju, Korea, Republic of
Duration: 2012 Dec 162012 Dec 18

Publication series

NameAdvances in Intelligent Systems and Computing
Volume208 AISC
ISSN (Print)21945357

Other

Other1st International Conference on Robot Intelligence Technology and Applications, RiTA 2012
CountryKorea, Republic of
CityGwangju
Period12/12/1612/12/18

Fingerprint

Detectors
Computational efficiency
Mobile robots
Feature extraction
Wheels
Textures
Switches
Robots

Keywords

  • Feature detection
  • Outdoor environment
  • Visual odometry

ASJC Scopus subject areas

  • Computer Science(all)
  • Control and Systems Engineering

Cite this

Otsu, K., Otsuki, M., Ishigami, G., & Kubota, T. (2013). An examination of feature detection for real-time visual odometry in untextured natural terrain. In Advances in Intelligent Systems and Computing (Vol. 208 AISC, pp. 405-414). (Advances in Intelligent Systems and Computing; Vol. 208 AISC). https://doi.org/10.1007/978-3-642-37374-9_39

An examination of feature detection for real-time visual odometry in untextured natural terrain. / Otsu, Kyohei; Otsuki, Masatsugu; Ishigami, Genya; Kubota, Takashi.

Advances in Intelligent Systems and Computing. Vol. 208 AISC 2013. p. 405-414 (Advances in Intelligent Systems and Computing; Vol. 208 AISC).

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

Otsu, K, Otsuki, M, Ishigami, G & Kubota, T 2013, An examination of feature detection for real-time visual odometry in untextured natural terrain. in Advances in Intelligent Systems and Computing. vol. 208 AISC, Advances in Intelligent Systems and Computing, vol. 208 AISC, pp. 405-414, 1st International Conference on Robot Intelligence Technology and Applications, RiTA 2012, Gwangju, Korea, Republic of, 12/12/16. https://doi.org/10.1007/978-3-642-37374-9_39
Otsu K, Otsuki M, Ishigami G, Kubota T. An examination of feature detection for real-time visual odometry in untextured natural terrain. In Advances in Intelligent Systems and Computing. Vol. 208 AISC. 2013. p. 405-414. (Advances in Intelligent Systems and Computing). https://doi.org/10.1007/978-3-642-37374-9_39
Otsu, Kyohei ; Otsuki, Masatsugu ; Ishigami, Genya ; Kubota, Takashi. / An examination of feature detection for real-time visual odometry in untextured natural terrain. Advances in Intelligent Systems and Computing. Vol. 208 AISC 2013. pp. 405-414 (Advances in Intelligent Systems and Computing).
@inproceedings{55f935c44713496baf9db617b74540f3,
title = "An examination of feature detection for real-time visual odometry in untextured natural terrain",
abstract = "Estimating the position of a robot is an essential requirement for autonomous mobile robots. Visual Odometry is a promising localization method in slippery natural terrain, which drastically degrades the accuracy of Wheel Odometry, while relying neither on other infrastructure nor any prior knowledge. Visual Odometry, however, suffers from the instability of feature extraction from the untextured natural terrain. To date, a number of feature detectors have been proposed for stable feature detection. This paper compares commonly used detectors in terms of robustness, localization accuracy and computational efficiency, and points out their trade-off problems among those criteria. To solve the problem, a hybrid algorithm is proposed which dynamically switches between multiple detectors according to the texture of terrain. Validity of the algorithm is proved by the simulation using dataset at volcanic areas in Japan.",
keywords = "Feature detection, Outdoor environment, Visual odometry",
author = "Kyohei Otsu and Masatsugu Otsuki and Genya Ishigami and Takashi Kubota",
year = "2013",
doi = "10.1007/978-3-642-37374-9_39",
language = "English",
isbn = "9783642373732",
volume = "208 AISC",
series = "Advances in Intelligent Systems and Computing",
pages = "405--414",
booktitle = "Advances in Intelligent Systems and Computing",

}

TY - GEN

T1 - An examination of feature detection for real-time visual odometry in untextured natural terrain

AU - Otsu, Kyohei

AU - Otsuki, Masatsugu

AU - Ishigami, Genya

AU - Kubota, Takashi

PY - 2013

Y1 - 2013

N2 - Estimating the position of a robot is an essential requirement for autonomous mobile robots. Visual Odometry is a promising localization method in slippery natural terrain, which drastically degrades the accuracy of Wheel Odometry, while relying neither on other infrastructure nor any prior knowledge. Visual Odometry, however, suffers from the instability of feature extraction from the untextured natural terrain. To date, a number of feature detectors have been proposed for stable feature detection. This paper compares commonly used detectors in terms of robustness, localization accuracy and computational efficiency, and points out their trade-off problems among those criteria. To solve the problem, a hybrid algorithm is proposed which dynamically switches between multiple detectors according to the texture of terrain. Validity of the algorithm is proved by the simulation using dataset at volcanic areas in Japan.

AB - Estimating the position of a robot is an essential requirement for autonomous mobile robots. Visual Odometry is a promising localization method in slippery natural terrain, which drastically degrades the accuracy of Wheel Odometry, while relying neither on other infrastructure nor any prior knowledge. Visual Odometry, however, suffers from the instability of feature extraction from the untextured natural terrain. To date, a number of feature detectors have been proposed for stable feature detection. This paper compares commonly used detectors in terms of robustness, localization accuracy and computational efficiency, and points out their trade-off problems among those criteria. To solve the problem, a hybrid algorithm is proposed which dynamically switches between multiple detectors according to the texture of terrain. Validity of the algorithm is proved by the simulation using dataset at volcanic areas in Japan.

KW - Feature detection

KW - Outdoor environment

KW - Visual odometry

UR - http://www.scopus.com/inward/record.url?scp=84876206768&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84876206768&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-37374-9_39

DO - 10.1007/978-3-642-37374-9_39

M3 - Conference contribution

SN - 9783642373732

VL - 208 AISC

T3 - Advances in Intelligent Systems and Computing

SP - 405

EP - 414

BT - Advances in Intelligent Systems and Computing

ER -