Research Areas

This research area focuses on advancing BIM through model conditioning and enrichment, transforming incomplete, inconsistent, or unstructured BIM data into structured, standardised, and decision-ready digital assets. Model conditioning involves cleaning, restructuring, and aligning BIM models with project-specific requirements, industry standards, and classification systems, ensuring interoperability and usability across different platforms and stakeholders.

Building on this, data enrichment and validation processes are applied to augment BIM models with missing attributes—such as material properties, cost, carbon, and asset information—while ensuring data accuracy, consistency, and compliance with relevant standards. These processes enable BIM models to evolve from geometric representations into rich, semantically meaningful information environments that support downstream applications.

Furthermore, the integration of Large Language Models (LLMs) enables advanced semantic querying and intelligent data interaction, allowing users to access, interpret, and reason over complex BIM datasets using natural language. This significantly enhances data accessibility, automation, and decision-making efficiency across design, construction, and asset management phases.

This research area focuses on developing unified methods to represent and understand the physical construction environment by leveraging multi-modal data sources, including 3D LiDAR point clouds, 360° imagery, and video streams. The core objective is to encode each data modality into high-dimensional, semantically rich representations that capture geometry, appearance, temporal dynamics, and contextual relationships within construction sites.

Advanced machine learning and computer vision techniques—such as point cloud learning, vision-language models, and self-supervised representation learning—are employed to extract meaningful features from heterogeneous data. These representations are designed to be robust to noise, incomplete observations, and dynamic site conditions, enabling reliable perception in complex construction environments.

A key focus is the fusion of multi-modal data into a unified representation space, where information from LiDAR, images, and videos can be aligned, correlated, and jointly interpreted. This unified representation enables cross-modal reasoning and supports seamless integration with BIM and digital twin systems. The resulting framework underpins a wide range of downstream applications, including semantic data querying, automated asset identification and management, progress monitoring, and embodied intelligence for construction robotics. By bridging the gap between raw sensory data and structured digital models, this research area establishes a foundational layer for intelligent, data-driven construction systems.

This research theme focuses on advancing next-generation construction robotics by integrating robotics, artificial intelligence, computer vision, BIM, digital twins, and multimodal sensing. The aim is to develop intelligent robotic systems that can perceive complex and dynamic construction environments, understand tasks and risks, and support or autonomously perform activities such as site inspection, material handling, component positioning, quality checking, and human-robot collaboration.

A key emphasis is placed on embodied intelligence, where robots learn from human activities, site observations, and engineering knowledge to make context-aware decisions and adapt to unstructured and changing site conditions. By combining real-world sensing data with BIM models, task planning, and safety constraints, this research seeks to create robotic solutions that are not only technically capable, but also practical, safe, and deployable for real construction and infrastructure projects.

The overall goal is to improve construction productivity, safety, quality, and sustainability through intelligent robotic systems tailored for the needs of the built environment.

This research area focuses on leveraging advanced digital technologies to estimate, monitor, and control carbon emissions across the construction, infrastructure, energy and resources industries. It begins with BIM-based carbon estimation, enabling early-stage decision-making by embedding carbon data into design models to support low-carbon design alternatives, material selection, and scenario analysis.

Building on this, the lab develops AI-enabled carbon tracking methods that utilise validated construction documents—such as invoices, contracts, delivery dockets, and receipts—to capture actual, evidence-based emissions during project delivery. By combining document intelligence with data analytics, the approach ensures high-fidelity, auditable carbon accounting that bridges the gap between estimated and actual emissions.

In addition, the lab advances AI-driven carbon auditing and reporting, automatically generating transparent, traceable carbon reports aligned with industry standards and regulatory requirements. These reports link emissions directly to primary evidence, improving trust, compliance, and decision-making. Through the integration of BIM, AI, blockchain, IoT, and data analytics, this area aims to deliver end-to-end carbon intelligence—from early design to construction and audit—supporting real-time monitoring, optimisation of resource use, and implementation of effective emission reduction strategies. Ultimately, this work enables organisations to meet sustainability targets, comply with evolving regulations, and transition towards a low-carbon future.