Leading engineers, researchers, and infrastructure specialists gathered in London for the Geotechnics Forum 2025 hosted by the Mott MacDonald Geotechnics Practice, where the transformative potential of artificial intelligence in underground infrastructure resilience took centre stage.
The Forum, held on 14 November, featured a keynote address by Dr Brian Sheil, Director of the Centre for Smart Infrastructure & Construction and Laing O’Rourke Associate Professor in Construction Engineering. His presentation, titled “AI Meets Mechanics: Integrating Deep Learning and Physics for Underground Infrastructure Resilience”, explored how the field of geotechnics is rapidly evolving through the fusion of data-driven methods and physics-based modelling.
Dr Sheil’s keynote charted the progression of geotechnical engineering from empirical methods to advanced hybrid approaches that leverage the vast amounts of data now produced by modern sensing technologies. He highlighted the growing challenge of manually processing LiDAR, fibre optic and ground-penetrating radar (GPR) datasets, emphasising that automated, physics-informed AI is becoming essential for detecting early-stage deterioration, interpreting complex signals, and supporting robust engineering decisions.
A key focus of the presentation was the application of deep learning and simulation tools to tunnel asset management. Dr Sheil demonstrated:
- AI-driven defect identification powered by high-fidelity synthetic LiDAR datasets and real-world scans.
- Advanced GPR reconstruction techniques capable of revealing hidden subsurface features with nearly 94% accuracy.
- A multi-modal “Digital Inspector” platform integrating LiDAR, GPR, inspection reports and mechanics-based models to prioritise defects and recommend maintenance actions.
His work included examples from ageing UK infrastructure, such as the 200-year-old Islington Tunnel, illustrating how innovative AI tools can modernise maintenance and extend asset life. The Forum also examined the industry-wide shift toward physics-informed machine learning, with discussions highlighting its importance for producing trustworthy, explainable and generalisable AI systems-qualities essential in safety-critical engineering contexts.
With strong attendance and engagement, the Geotechnics Forum 2025 provided a timely platform for exchanging ideas on how digital technologies, sensing and AI can strengthen infrastructure resilience and improve decision-making across the UK’s geotechnical sector.