OATAO - Open Archive Toulouse Archive Ouverte Open Access Week

PAVO: a Parallax based Bi-Monocular VO Approach For Autonomous Navigation In Various Environments

Vivet, Damien and Debord, Adrien and Pages, Gaël PAVO: a Parallax based Bi-Monocular VO Approach For Autonomous Navigation In Various Environments. (2019) In: The International Conference on Digital Image & Signal Processing (DISP’19), 29 April 2019 - 30 April 2019 (Oxford, United Kingdom).

[img] (Document in English)

PDF (Author's version) - Depositor and staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
772kB

Abstract

This paper presents PAVO (Parallax based Visual Odometry), a new feature-based bi-monocular-cameras visual odometry. It can operate in both short or long range without any switch between mono or stereo algorithms and can be degraded into monocular case. The methodology relies on the use of the parallax information in order to estimate the confidence of each landmark. We have made PAVO to be used on any kind of vehicle or platform and for indoor as well as outdoor environments without tuning. The approach has been validated on popular datasets: KITTI for ground vehicle, EuRoC for UAV but also experimented on bicycle and underwater datasets from ISAE-SUPAERO. The obtained results for the camera only localization show that PAVO gives good trajectory results on KITTI dataset while it is not using any loop-closing, graph optimization and uses the same parameters values for all the experiments.

Item Type:Conference or Workshop Item (Paper)
Audience (conference):International conference proceedings
Uncontrolled Keywords:
Institution:Université de Toulouse > Institut Supérieur de l'Aéronautique et de l'Espace - ISAE-SUPAERO (FRANCE)
Laboratory name:
Funders:
DGA/MRIS under grant n°2015.60.0090.00.470.75.01 AV1
Statistics:download
Deposited By: Damien Vivet
Deposited On:20 Feb 2019 13:12

Repository Staff Only: item control page