Vizualna odometrija i kartiranje vanjskog prostora

master's thesis

• Title (english): Visual odometry and mapping of outdoor environments
• Author: Piero Anić
• Mentor: Marija Seder (mentor)
• Committee member: Ivan Marković (član povjerenstva)
• Committee member: Juraj Peršić (član povjerenstva)
• Committee member: Luka Petrović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2023-07-11, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering

Abstract

Istodobna lokalizacija i izgradnja karte (SLAM) je jedan od osnovnih problem robotike i računalnog vida čiji je cilj omogućiti autonomnim sustavima istovremenu procjenu vlastite pozicije i izgradnju karte okolišnog prostora. Ovaj rad istražuje napretke, izazove i primjene SLAM sustava, s fokusom na teoretsku kao i na primijenjenu implementaciju. Prva dva poglavlja pružaju sveobuhvatan pregled ishodišnih koncepata i tehnika koje SLAM obuhvaća. Pokrivaju matematički opis temeljnog problema kao i definiciju algoritma korak po korak. Treće poglavlje pokriva razvoj otkrivanja, opisa i podudaranja značajki nakon kojeg slijedi opis pretvorbe video zapisa u kartu u četvrtom poglavlju. Tu se raspravlja i o optimizaciji projekcije značajki i kako to utječe na izgradnju karte. Dodatno su spomenuti i problemi kao scale-drift kao i potencijalna rješenja problema. Naposljetku je pokrivena i tema rekonstrukcije površine u petom poglavlju, rekonstrukcija zatvorenih površina kao i nove metode rekonstrukcija površina koje nisu nužno zatvorene.

Abstract (english)

Simultaneous Localization and Mapping (SLAM) is a fundamental problem in the field of robotics and computer vision that aims to enable autonomous systems to simultaneously estimate their own position and create a map of the surrounding environment. This thesis explores the advancements, challenges, and applications of SLAM techniques, with a focus on both the theoretical foundations and practical implementations. The first and second chapter of the thesis provides a comprehensive review of the underlying concepts and techniques involved in SLAM. They cover mathematical definition of the underlying issue as well as a step by step definition of the algorithm. Third chapter covers the development of feature detection, description and matching which is followed by the process of turning the video to a map in chapter four. There it is talked about optimization of feature projection and how it fits into the map building. Additionally issues like scale drift are mentioned and potential solutions. In the end surface reconstruction is covered in fifth chapter, both closed surface reconstruction and new methods for reconstructing open space.

• Keywords: lokalizacija; izgradnja karte; otkrivanje; opis; podudaranje značajki; rekonstrukcija površine
• Keywords (english): localization; mapping; feature detection; description; matching; surface reconstruction
• Language: croatian
• URN:NBN: urn:nbn:hr:168:774848
• Study programme: Title: Electrical Engineering and Information Technology; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije (magistar/magistra inženjer/inženjerka elektrotehnike i informacijske tehnologije)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-10-11 08:17:10