Automatic As-built Digital Twin

The goal of this project is to create a novel approach to automatically generate as-built Building Information Models – also called Digital Twin, of built environment assets.
Description of the Project: 

Key to achieve visions like Smart Cities or Digital Britain is the availability of reliable Digital Twins of built environment assets (buildings, infrastructure industrial complexes). Such semantically-rich 3D models are important to support Operations and Maintenance activities, including for the effective and safe operations of robots (e.g. to operate a valve or simply navigate some environments during dangerous or emergency operations).

However, while these assets may be designed with care and detail, their actual construction often results in changes that must be recorded in the Digital Twin. Furthermore, the facilities age and deteriorate, so that monitoring the appearance and evolution of defects is critical to ensure safe and sustainable operation. Detecting construction discrepancies as well as defects of time is challenge and these activities currently require significant work and capital. Some methods (and commercial software) have been proposed to process point cloud data of as-built assets to generate digital twins, but these are very limited in capacity and level of automation.

Method: the project will aim to interpret as-is data (including coloured point clouds, images, thermal maps, etc.) of a facility given a design model to identify discrepancies and defects. This will be achieved by integrating various techniques, such as Scan-to-BIM [1], Scan-vs-BIM [2], and AI techniques.

Evaluation: We have access to a number of case study point clouds that can be provided by various companies in Europe and beyond.

Resources required: 
Possibly access to a terrestrial laser scanner, but the team already has access to datasets
Project number: 
300004
First Supervisor: 
University: 
University of Edinburgh
Second Supervisor(s): 
First supervisor university: 
Heriot-Watt University
Essential skills and knowledge: 
2D-3D data processing
Desirable skills and knowledge: 
Machine learning / AI
Funding Available: 
References: 

[1] Murali S., Speciale P., Oswald M. R., Pollefeys M. (2017), “Indoor Scan2BIM: Building Information Models of House Interiors”, IEEE/RSJ Proceedings of IROS, 2017. [pdf]

[2] Bosché F., Ahmed, M., Turkan, Y., Haas, C., Haas, R. (2015), "The value of integrating Scan-to-BIM and Scan-vs-BIM techniques for construction monitoring using laser scanning and BIM: The case of cylindrical MEP components", Automation in Construction, Vol. 49, pp. 201-213. [pdf]