Skip to Main content Skip to Navigation
Theses

Localisation ensembliste de drones à l’aide de méthodes par intervalles

Ide Flore Kenmogne Fokam 1, 2
2 RAINBOW - Sensor-based and interactive robotics
Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : This thesis aims at characterizing a pose domain for the localization of drones, using interval set-membership methods. This characterization is enables to quantify the positioning uncertainty. In the absence or failure of GPS, in constrained and indoor environments, an alternative is to use the camera. The image measurements and those coming from the drone sensors as well as the models parameters are very often tainted by errors. Classical estimation methods provide a point estimate of the drone pose (position and orientation), assuming a probabilistic model of errors. However, it is sometimes difficult or impossible to describe precisely the probability distributions of these disturbances. In the set-membership framework, these errors can be represented by intervals. Interval analysis can then be used to propagate uncertainties, even in the presence of outliers. This work proposes an interval set-membership localization method based on image measurements; a quantifier elimination method for taking into account the uncertainties on the landmarks positions; and a method for solving the set-membership cooperative localization problem. Each of these methods has been implemented and tested in simulation and on real data acquired in an indoor environment. making their comparisons with existing classical methods possible.
Document type :
Theses
Complete list of metadata

Cited literature [178 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02921454
Contributor : Abes Star :  Contact
Submitted on : Tuesday, August 25, 2020 - 11:18:17 AM
Last modification on : Tuesday, February 23, 2021 - 3:27:49 AM
Long-term archiving on: : Tuesday, December 1, 2020 - 7:07:56 AM

File

KENMOGNE_FOKAM_Ide_Flore.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02921454, version 1

Citation

Ide Flore Kenmogne Fokam. Localisation ensembliste de drones à l’aide de méthodes par intervalles. Robotique [cs.RO]. Université Rennes 1, 2019. Français. ⟨NNT : 2019REN1S084⟩. ⟨tel-02921454⟩

Share

Metrics

Record views

65

Files downloads

94