Visual-based simultaneous localization and mapping and global positioning system correction for geo-localization of a mobile robot

Sid Ahmed Berrabah, Hichem Sahli, Yvan Baudoin

Research output: Contribution to journalArticlepeer-review

Abstract

This paper introduces an approach combining visual-based simultaneous localization and mapping (V-SLAM) and global positioning system (GPS) correction for accurate multi-sensor localization of an outdoor mobile robot in geo-referenced maps. The proposed framework combines two extended Kalman filters (EKF); the first one, referred to as the integration filter, is dedicated to the improvement of the GPS localization based on data from an inertial navigation system and wheels' encoders. The second EKF implements the V-SLAM process. The linear and angular velocities in the dynamic model of the V-SLAM EKF filter are given by the GPS/INS/Encoders integration filter. On the other hand, the output of the V-SLAM EKF filter is used to update the dynamics estimation in the integration filter and therefore the geo-referenced localization. This solution increases the accuracy and the robustness of the positioning during GPS outage and allows SLAM in less featured environments.

Original languageEnglish
Article number124003
JournalMeasurement Science and Technology
Volume22
Issue number12
DOIs
Publication statusPublished - Dec 2011

Keywords

  • geo-localization
  • simultaneous localization and mapping

Fingerprint

Dive into the research topics of 'Visual-based simultaneous localization and mapping and global positioning system correction for geo-localization of a mobile robot'. Together they form a unique fingerprint.

Cite this