Skip to Main content Skip to Navigation
Theses

Simultaneous Localisation And Mapping with Monocular Vision

Thomas Lemaire 1
1 LAAS-RIS - Équipe Robotique et InteractionS
LAAS - Laboratoire d'analyse et d'architecture des systèmes
Abstract : This thesis tackles the Simultaneous Localisation And Mapping problem (SLAM for short). When the robot moves in an unknown environment, it has to incrementally build a map of the environment while using this map to localise itself. A SLAM algorithm is a fundamental part of the architecture of a fully autonomous robot. Several elements are required to solve SLAM, among which perception is of main importance since it produces observations of the objects of the environment (referred as landmarks) which compose the map. In this work, the focus is put on artificial vision as the main perception mean for the robot: the map and the robot location can be estimated in the full 3D space. Digital cameras are well suited for embedded systems and produce a valuable information on the environment. But a camera does not give a measure of the distance to the objects, only partial observations are available. The addition of a new landmark in the map is then a difficult problem. An initialisation method for 3D point landmarks is proposed, it is based on a multi-hypotheses generation and selection scheme. A full SLAM solution for a rover is described, in particular the use of a panoramic camera delivers a 360 degrees field of view. This architecture has been implemented on an ATRV rover. A map of 3D points is relevant for robot localisation, but gives a limited information on the structure of the environment. An algorithm to introduce 3D line segments in the map is proposed and tested on real data
Document type :
Theses
Complete list of metadatas

Cited literature [98 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00452478
Contributor : Anne Bergez <>
Submitted on : Tuesday, February 2, 2010 - 2:57:21 PM
Last modification on : Friday, January 10, 2020 - 9:10:15 PM
Long-term archiving on: : Friday, June 18, 2010 - 6:21:24 PM

Identifiers

  • HAL Id : tel-00452478, version 1

Citation

Thomas Lemaire. Simultaneous Localisation And Mapping with Monocular Vision. Automatic. Ecole nationale superieure de l'aeronautique et de l'espace, 2006. English. ⟨tel-00452478⟩

Share

Metrics

Record views

425

Files downloads

1237