In this edition of our HumanTech technologies series, our colleague Mahdi Chamseddine, an M.Sc. researcher at the German Research Center for Artificial Intelligence (DFKI), shares the keys to the scan-to-BIM technology we are developing.
What is scan-to-BIM and why is it important?
Building Information Modeling (BIM) is a process in architecture, engineering, and construction (AEC) that uses 3D models to centralize building project information. BIM integrates design, visualization, and collaboration, thus enhancing efficiency, accuracy, and communication throughout a building’s lifecycle, from planning to construction and management.
Scan-to-BIM is the process of generating BIM models from 3D scans of existing structures, buildings, or construction sites.
“Scan-to-BIM is the process of generating BIM models from 3D scans of existing structures, buildings, or construction sites.”
Scanning technologies provide accurate 3D point clouds of the target site. The scans reflect up-to-date information and real-world conditions, thus allowing for the generation of “as-built” models. The “as-built” building model can defer from the architectural plans, and thus, an automated approach to BIM generation is desired. This invaluable tool streamlines projects, facilitates efficient facility management, and empowers informed decision-making throughout a building’s lifecycle.
How does scan-to-BIM work?
Scan-to-BIM starts with data acquisition. There are different methods for 3D scanning of buildings, such as terrestrial scanners or photogrammetry. The scanning devices capture high-resolution and detailed point clouds of buildings or sites. Some scanning technologies can capture colour as well as 3D points, which allows for more information to be extracted.
The point clouds are then processed to remove noise and clutter. A typical scan will include data that is not relevant for the BIM generation, such as neighbouring trees or buildings and objects outside the designated site. The processed point clouds are then segmented using specialized machine-learning algorithms that are trained to recognize structural elements in scans. The AI model classifies points into relevant classes such as walls, floor, ceiling, doors, columns, and other classes. The segmented point cloud is used to create structured 3D models for the different objects depending on their type and dimensions. The collection of the 3D models, as well as their relationships and other information, constitutes the BIM model.
Scan-to-BIM applications and challenges
Generating a BIM model of existing buildings is labour-intensive, but automating the process can allow us to reap the benefits faster and more cost-effectively.
“Generating a BIM model of existing buildings is labour-intensive, but automating the process can allow us to reap the benefits faster and more cost-effectively.”
It allows for automated progress monitoring of construction sites by generating models at regular intervals. Changes to the project are tracked and stored, and deviations from the plan can be detected and highlighted at an early stage, eliminating the need for costly corrections down the line.
Scan-to-BIM can be used, among other things, in renovation, historic preservation, and industrial projects to create precise digital replicas of existing structures. However, some challenges, including maintaining data quality, managing large datasets, scanning hazardous or hard-to-access areas, and adapting to unique and different structures, still exist.
Advancements in scan-to-BIM technology promise increased efficiency and accuracy for the AEC industry.
Learn about another of the technologies we are developing at HumanTech: the 360° ToF camera. Its speed, portability, accuracy, and efficiency in 3D data collection are unparalleled, and it is set to transform the way construction sites are monitored and managed.
Stay tuned to our news, newsletter, and social media channels (LinkedIn, Twitter and YouTube) to follow our journey toward accelerating the digital transformation of construction!