AI and Remote Sensing for Resilient and Sustainable Built Environments: A Review of Current Methods, Open Data and Future Directions

Abstract Critical infrastructures such as roads, bridges and buildings are increasingly exposed to damage from climate change, natural hazards and human-induced events, making rapid and reliable damage assessment essential for resilient infrastructure management. This review critically examines how artificial intelligence, combined with modern data acquisition technologies, is applied to damage assessment and monitoring across these three infrastructure classes. Following PRISMA 2020 guidelines, 244 peer-reviewed studies published between 2018 and 2026 are analysed for AI models, datasets, data sources and applications. Rather than cataloguing methods, the review interrogates why the field has converged on particular configurations. Three methodologically distinct ones emerge: detector-heavy pipelines on UAV road imagery, a split bridge literature between close-range visual inspection and InSAR-based long-term monitoring, and a satellite-based building literature focused on post-event change detection. The analysis shows that model choice is structurally constrained by sensing modality and data representation rather than architectural superiority, and that data availability is the binding constraint, not model capability: the literature remains concentrated on a small number of benchmark datasets, with limited cross-site validation and scarce satellite and SAR usage, despite their operational value. These same constraints help explain why operational adoption by the engineering sector remains limited. The review consolidates these findings into a conceptual taxonomy and a three-tier monitoring framework, and sets out a research agenda spanning foundation models, multimodal fusion, physics-informed AI, digital twins, edge AI, and explainable AI.