Pose estimation through two non-overlapping orthogonally mounted cameras for fiducial markers.

Date
2024-02
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH ABSTRACT: This thesis explores the use of ArUco markers for precise pose estimation in Concentrated Solar Power (CSP) plants, aiming to enhance the accuracy and functionality of Unmanned Aerial Vehicles (UAVs) within these environments. The study is structured around three main objectives: assessing and implementing the Gazebo simulation engine’s applicability to real-world scenarios to create a tested for the next two objectives, creating a framework for selecting optimal markers, and evaluating different camera setups for improved system performance. Key findings include the successful validation of the Gazebo engine for simulating UAV operations. The research further delves into marker selection criteria, addressing aspects such as accuracy, focal length, resolution, and the impact of motion blur as well as giving insights into the operating distance and angle for different camera-marker configurations. Additionally, the comparative analysis of monocular, stereo, and orthogonal camera configurations reveals no significant performance disparity, suggesting that resolution adjustments for a monocular camera could mitigate the benefits of more complex setups. Although ArUco markers show promise for CSP applications, the study acknowledges potential limitations related to UAV operational distances and marker scalability. The conclusions drawn show the importance of tailored technological solutions, suggesting the incorporation of advanced sensor fusion and filtering strategies for enhanced system precision. Recommendations for future work include improving Gazebo’s simulation accuracy by adding motion blur effects and expanding the analysis to cover a wider range of focal lengths and marker sizes, aiming for a closer approximation to real-world conditions.
AFRIKAANSE OPSOMMING: Hierdie tesis ondersoek die gebruik van ArUco-merkers vir presiese posisiebepaling in Gekonsentreerde Sonkrag (CSP) aanlegte, met die doel om die akkuraatheid en funksionaliteit van Onbemande Lugvoertuie (UAV’s) binne hierdie omgewings te verbeter. Die studie is gestruktureer rondom drie hoofdoelwitte: die assessering en implementering van die Gazebo-simulasie-enjin se toepaslikheid op werklike scenario’s om ’n getoetste basis vir die volgende twee doelwitte te skep, die skepping van ’n raamwerk vir die kies van optimale merkers, en die evaluering van verskillende kamera-opstellings vir verbeterde stelselprestasie. Sleutelbevindinge sluit in die suksesvolle validering van die Gazebo-enjin vir die simulasie van UAV-operasies. Die navorsing verdiep verder in die kriteria vir merkerseleksie, en spreek aspekte soos akkuraatheid, fokale lengte, resolusie, en die impak van bewegingsvervaging aan, asook insigte in die werkafstand en hoek vir verskillende kamera-merker konfigurasies. Daarbenewens toon die vergelykende analise van monokulêre, stereo, en ortogonaal gemonteerde kamera-opstellings geen noemenswaardige prestasieverskil nie, wat suggereer dat resolusie-aanpassings vir ’n monokulêre kamera die voordele van meer komplekse opstellings kan teenwerk. Alhoewel ArUco-merkers belowend toon vir CSP-toepassings, erken die studie potensiële beperkings wat verband hou met UAV-werkafstande en merkerskalering. Die gevolgtrekkings beklemtoon die belangrikheid van pasgemaakte tegnologiese oplossings, en stel voor dat gevorderde sensorfusie en filtreringstrategieë ingesluit word vir verbeterde stelselpresisie. Aanbevelings vir toekomstige werk sluit in die verbetering van Gazebo se simulasie-akkuraatheid deur die toevoeging van bewegingsvervaging-effekte en die uitbreiding van die analise om ’n wyer reeks van fokale lengtes en merkergroottes te dek, met die doel om ’n nouer benadering tot werklike toestande te bewerkstellig.
Description
Thesis (MEng)--Stellenbosch University, 2024.
Keywords
Citation