Automated High-quality Image Acquisition for UAV-enabled Bridge Visual Inspection

Bridges need regular inspections to identify potential defects and provide maintenance recommendations. Recently, camera-mounted Unmanned Aerial Vehicles (UAVs) have emerged as a promising tool for bridge inspection, potentially replacing traditional inspection methods using rigging, scaffoldings, or heavy bucket trucks. However, UAV-enabled bridge inspection practices are still in their early stages, facing challenges in experience-based subjective errors by pilots, difficulties in accurate UAV localisation in GPS-denied bridge areas, and potential image quality issues. These challenges hinder the widespread adoption of UAVs in real bridge inspection. This study aims to establish a systematic framework for enabling UAVs to automatically capture high-quality images of bridges, thereby generating a photorealistic 3D bridge model and supporting subsequent damage detection and bridge condition assessment. Firstly, to reduce subjective errors by UAV pilots, a set of objective rules was designed for planning effective camera viewpoints considering specific inspection requirements, photogrammetry constraints, and UAV flight safety considerations. Secondly, to enable UAV autonomous navigation in GPS-denied bridge areas, a new low-cost UAV localisation method was proposed for estimating the UAV’s global location in areas underneath bridge girders. This method combines Stereo Visual Inertia Odometry (SVIO) and fiducial marker-based measurements. By periodically registering the local pose estimation into a global frame and correcting accumulated error by the SVIO, this method can provide robust global location estimation during long-distance flights under multiple bridge girders. Thirdly, to address potential image quality issues that may affect inspection effectiveness, an In-flight Image Quality Check (IIQC) framework was proposed to evaluate critical quality aspects of UAV-captured images for inspection purposes and to enable UAV pilots to promptly address any identified image quality shortcomings. The performance and effectiveness of these three approaches have been rigorously verified through extensive testing in both simulated environments and real-world bridge scenarios. Finally, a comprehensive framework for automated UAV-enabled high-quality image acquisition was introduced by integrating these three components into a unified system. The findings demonstrate that this integrated framework can not only substantially enhance the level of automation in UAV-enabled image collection across a wide range of bridge structures and under various weather conditions but also guarantee the UAV’s ability to capture high-quality imagery data.

Name thesis:degree_name

PhD

Level thesis:degree_level

Doctoral

Discipline thesis:degree_discipline

Civil Engineering

Grantor dc:publisher

ResearchSpace@Auckland

Year dc:date.issued

2023

Author dc:creator

• Wang, Feng

Advisors dc:contributor.advisor

• Zou, Yang
• del Rey Castillo, Enrique
• Lim, James B.P.

Statement dc:rights

• Items in ResearchSpace are protected by copyright, with all rights reserved, unless otherwise indicated.

Licence dc:rights.uri

• https://researchspace.auckland.ac.nz/docs/uoa-docs/rights.htm

Handle dc:identifier.uri

https://hdl.handle.net/2292/69795

OAI identifier oai:identifier

oai:researchspace.auckland.ac.nz:2292/69795