Most of the buildings that exist today were built based on 2D drawings. Building information models that represent design-stage product information have become prevalent in the second decade of the 21st century. Still, it will take many decades before such models become the norm for all existing buildings. In the meantime, the building industry lacks the tools to leverage the benefits of digital information management for construction, operation, and renovation. To this end, this paper reviews the state-of-the-art practice and research for constructing (generating) and maintaining (updating) geometric digital twins. This paper also highlights the key limitations preventing current research from being adopted in practice and derives a new geometry-based object class hierarchy that mainly focuses on the geometric properties of building objects, in contrast to widely used existing object categorisations that are mainly function-oriented. We argue that this new class hierarchy can serve as the main building block for prioritising the automation of the most frequently used object classes for geometric digital twin construction and maintenance. We also draw novel insights into the limitations of current methods and uncover further research directions to tackle these problems. Specifically, we believe that adapting deep learning methods can increase the robustness of object detection and segmentation of various types; involving design intents can achieve a high resolution of model construction and maintenance; using images as a complementary input can help to detect transparent and specular objects; and combining synthetic data for algorithm training can overcome the lack of real labelled datasets.

Digital twins have started to diffuse within architecture, engineering, construction, and operations (AECO), based on their emerging and anticipated benefits to the various stakeholders involved in the building life cycle. However, their applications are still at an early stage, and much effort is still needed to exploit their full potential. This chapter explains some key notions to help understand digital twins in AECO. It exposes the various definitions of digital twins and illustrates the basic steps and relevant methods for creating a digital twin. The chapter also provides an overview of the state-of-the-art deep learning methods for digital twins and discusses some real-life use cases. Finally, the chapter discusses the benefits and challenges associated with the adoption of digital twins.

Geometric fidelity is a fundamental prerequisite for reliable cognitive digital twins (CDTs). An incomplete as-is geometric foundation leads to model uncertainty, thereby undermining CDT reasoning and maintenance decisions. However, point cloud data acquired through laser scanning is inevitably subject to occlusions caused by structural obstructions, sensor limitations, and complex MEP configurations. Existing methods fail to address the high-precision full spectrum of occlusion scenarios in real-world Scan-to/vs-BIM workflows, particularly for border truncations and irregular non-Manhattan geometries. This paper proposes a method that systematically reduces geometric uncertainty in CDT construction through three stages. First, a gap classification scheme categorises incomplete point clusters by occlusion severity and object class, enabling targeted treatment of each scenario for both Manhattan and non-Manhattan geometries. The completed clusters are then tessellated into high-resolution meshes and converted into IFC models through a new purpose-designed data structure that embeds laser-scan timestamps and as-designed global identifier links. Together, these stages form a closed-loop geometric DT updating system that continuously reflects the as-is physical state of the building, directly reducing the geometric uncertainty that would otherwise propagate into CDT-based maintenance reasoning and risk assessment. The method is validated on various real-world datasets. The performance is evaluated with ground truth and Chamfer distance, with results benchmarked against LOA requirements to confirm practical ability. The resulting timestamped IFC models provide a geometrically reliable foundation for CDT-based predictive maintenance and facility risk assessment.