Vis enkel innførsel

dc.contributor.advisorBrekke, Edmund Førland
dc.contributor.authorLegernæs, Eirik Worren
dc.date.accessioned2019-09-11T11:42:22Z
dc.date.created2017-06-05
dc.date.issued2017
dc.identifierntnudaim:16471
dc.identifier.urihttp://hdl.handle.net/11250/2616096
dc.description.abstractThe Multi-State Constraint Kalman Filter for visual-inertial odometry with online estimation of extrinsic and intrinsic camera parameters using the fisheye camera model is presented, and its performance in terms of accuracy and consistency is tested on synthetic and real world datasets The effect of some key parameters in the estimator is discussed and quantified with Monte Carlo simulations, and the sensitivity to initial conditions is investigated. A non-linear optimization based initialization procedure is proposed, and shown to outperform a linear closed-form method from the literature.en
dc.languageeng
dc.publisherNTNU
dc.subjectKybernetikk og robotikk, Navigasjon og fartøystyringen
dc.titleFusion of Inertial Navigation and Visual Odometry by Multi-State Constraint Kalman Filteren
dc.typeMaster thesisen
dc.source.pagenumber102
dc.contributor.departmentNorges teknisk-naturvitenskapelige universitet, Fakultet for informasjonsteknologi og elektroteknikk,Institutt for teknisk kybernetikknb_NO
dc.date.embargoenddate10000-01-01


Tilhørende fil(er)

Thumbnail
Thumbnail

Denne innførselen finnes i følgende samling(er)

Vis enkel innførsel