Skip to Main content Skip to Navigation
Theses

Indexation d'une base de données images : Application à la localisation et la cartographie fondées sur des radio-étiquettes et des amers visuels pour la navigation d'un robot en milieu intérieur

Younès Raoui 1
1 LAAS-RAP - Équipe Robotique, Action et Perception
LAAS - Laboratoire d'analyse et d'architecture des systèmes
Abstract : This document is related both to indexing methods in image data bases and to localization methods used in mobile robotics. It exploits the relationships between research works on Perception made in the Robotics department of LAAS-CNRS in Toulouse, and on Data Mining performed by the LIMIARF lab at the Rabat university. Computer Vision has become a major source of sensory data on mobile robots for about ten years; it allows to build especially representations of the environment in which a robot has to execute motions, either by geometrical maps or by appearance-based models. Concerning computer vision, only appearance-based representations have been studied; they consist on a data base of images acquired while the robot is moved in an interactive way during a learning step. Robot self-localization is provided by searching in this data base, the image which looks like the one acquired from the current position: this function is performed from indexing or data mining methods, similar to the ones used on Internet. It is proposed an indexing method based on interest points extracted from color and textured images. Moreover, geometrical representations are also considered for an RFID-based navigation method, based on particle filtering, used either in a classical formal way or with a more intuitive approach. Finally, very preliminar results are described on a multi-sensory approach using both Vision and RFID tags in order to improve the accuracy on the robot localization
Document type :
Theses
Complete list of metadata

Cited literature [79 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00633980
Contributor : Younès Raoui <>
Submitted on : Thursday, October 20, 2011 - 12:40:55 AM
Last modification on : Thursday, June 10, 2021 - 3:04:22 AM
Long-term archiving on: : Saturday, January 21, 2012 - 2:21:57 AM

Identifiers

  • HAL Id : tel-00633980, version 1

Citation

Younès Raoui. Indexation d'une base de données images : Application à la localisation et la cartographie fondées sur des radio-étiquettes et des amers visuels pour la navigation d'un robot en milieu intérieur. Automatique / Robotique. Institut National Polytechnique de Toulouse - INPT, 2011. Français. ⟨tel-00633980⟩

Share

Metrics

Record views

631

Files downloads

8052