Localization of a tracked robot based on fuzzy fusion of wheel odometry and visual odometry in indoor and outdoor environments

Hyeong Ha Ham, Sung Ho Hong, Jae Bok Song, Joo Hyun Baek, Jae Kwan Ryu

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Tracked robots usually have poor localization performance because of slippage of their tracks. This study proposes a new localization method for tracked robots that uses fuzzy fusion of stereo-camera-based visual odometry and encoder-based wheel odometry. Visual odometry can be inaccurate when an insufficient number of visual features are available, while the encoder is prone to accumulating errors when large slips occur. To combine these two methods, the weight of each method was controlled by a fuzzy decision depending on the surrounding environment. The experimental results show that the proposed scheme improved the localization performance of a tracked robot.

Original languageEnglish
Pages (from-to)629-635
Number of pages7
JournalTransactions of the Korean Society of Mechanical Engineers, A
Volume36
Issue number6
DOIs
Publication statusPublished - 2012 Jun

Keywords

  • Fuzzy fusion
  • Harris corner detector
  • Slip velocity
  • Stereo camera
  • Tracked robot

ASJC Scopus subject areas

  • Mechanical Engineering

Fingerprint Dive into the research topics of 'Localization of a tracked robot based on fuzzy fusion of wheel odometry and visual odometry in indoor and outdoor environments'. Together they form a unique fingerprint.

  • Cite this