Building facade inspection remains a safety-critical yet operationally challenging task, particularly for high-rise and glass-dominant structures where manual methods are hazardous, subjective, and increasingly insufficient under tightening regulatory demands. This thesis presents a systematic review of the state of the art in UAV enabled facade inspection, covering UAV data acquisition, AI-driven defect detection and segmentation, and 3D/BIM integration, with specific attention to the optical and geometric challenges posed by glass facades. Five representative peer-reviewed studies are critically analysed through a structured case-study methodology and synthesised thematically across five dimensions: data acquisition and UAV operations, image preprocessing and glass-specific enhancement, AI-based detection and segmentation architecture selection, 3D modelling and BIM integration strategies, and glass facade-specific failure modes. Drawing on this synthesis, the thesis proposes an integrated five-layer sequential framework that structures how inspection requirements translate into sensing configurations, glass-aware preprocessing, segmentation-oriented AI pipelines, spatially anchored 3D/BIM representations, and decision-support deliverables. Complementary decision matrices enable context sensitive customisation of the framework according to risk level, facade composition, and organisational BIM maturity. Evidence-informed best-practice guidelines are derived for each layer, emphasising traceability, uncertainty communication, and human-in-the-loop governance. The findings indicate that while individual pipeline components have reached promising levels of performance, fully autonomous end-to end systems for glass-rich facades remain an aspiration; reliable deployment requires co-design across all layers with explicit quality assurance and professional oversight. The thesis concludes by identifying priority research directions, including larger glass specific datasets, end-to-end validation protocols, and regulatory frameworks for UAV-AI-BIM adoption in facade maintenance practice.
L'ispezione delle facciate degli edifici rappresenta un'attività critica per la sicurezza ma operativamente complessa, specialmente per strutture di grande altezza con facciate in vetro, dove i metodi manuali risultano pericolosi, soggettivi e inadeguati rispetto alle normative vigenti. Questa tesi presenta una revisione sistematica dello stato dell'arte nell'ispezione delle facciate tramite UAV, coprendo l'acquisizione dati, il rilevamento dei difetti basato sull'intelligenza artificiale e l'integrazione 3D/BIM, con particolare attenzione alle sfide ottiche e geometriche del vetro. Cinque studi sottoposti a revisione paritaria vengono analizzati criticamente e sintetizzati tematicamente su cinque dimensioni: operazioni UAV, pre-elaborazione specifica per il vetro, architetture AI di segmentazione, strategie di integrazione 3D/BIM e modalità di guasto delle facciate vetrate. La tesi propone un framework integrato a cinque livelli sequenziali che traduce i requisiti ispettivi in configurazioni sensoriali, pipeline AI orientate alla segmentazione, rappresentazioni BIM ancorate spazialmente e prodotti decisionali. Matrici decisionali complementari consentono la personalizzazione in base al rischio, alla composizione della facciata e alla maturità BIM dell'organizzazione. Vengono derivate linee guida basate sull'evidenza per ciascun livello, enfatizzando tracciabilità, comunicazione dell'incertezza e supervisione umana. I risultati indicano che, sebbene i componenti individuali della pipeline raggiungano prestazioni promettenti, sistemi end-to-end completamente autonomi per facciate vetrate restano un'aspirazione; l'implementazione affidabile richiede co progettazione trasversale con protocolli di garanzia della qualità. La tesi identifica infine direzioni di ricerca prioritarie, tra cui dataset specifici per il vetro, validazione end-to-end e quadri normativi per l'adozione UAV-AI-BIM nella manutenzione delle facciate.
UAV-enabled building facade inspection: a systematic review and proposal of an integrated layered framework with decision matrices and BIM-based deliverables
Shabbir, Mohammad
2025/2026
Abstract
Building facade inspection remains a safety-critical yet operationally challenging task, particularly for high-rise and glass-dominant structures where manual methods are hazardous, subjective, and increasingly insufficient under tightening regulatory demands. This thesis presents a systematic review of the state of the art in UAV enabled facade inspection, covering UAV data acquisition, AI-driven defect detection and segmentation, and 3D/BIM integration, with specific attention to the optical and geometric challenges posed by glass facades. Five representative peer-reviewed studies are critically analysed through a structured case-study methodology and synthesised thematically across five dimensions: data acquisition and UAV operations, image preprocessing and glass-specific enhancement, AI-based detection and segmentation architecture selection, 3D modelling and BIM integration strategies, and glass facade-specific failure modes. Drawing on this synthesis, the thesis proposes an integrated five-layer sequential framework that structures how inspection requirements translate into sensing configurations, glass-aware preprocessing, segmentation-oriented AI pipelines, spatially anchored 3D/BIM representations, and decision-support deliverables. Complementary decision matrices enable context sensitive customisation of the framework according to risk level, facade composition, and organisational BIM maturity. Evidence-informed best-practice guidelines are derived for each layer, emphasising traceability, uncertainty communication, and human-in-the-loop governance. The findings indicate that while individual pipeline components have reached promising levels of performance, fully autonomous end-to end systems for glass-rich facades remain an aspiration; reliable deployment requires co-design across all layers with explicit quality assurance and professional oversight. The thesis concludes by identifying priority research directions, including larger glass specific datasets, end-to-end validation protocols, and regulatory frameworks for UAV-AI-BIM adoption in facade maintenance practice.| File | Dimensione | Formato | |
|---|---|---|---|
|
2026_03_Shabbir.pdf
non accessibile
Descrizione: Thesis text
Dimensione
1.84 MB
Formato
Adobe PDF
|
1.84 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/252984