Using artificial intelligence in building design
The possibilities offered by artificial intelligence (AI) have not yet been charted by the Finnish construction sector. Sweco is examining how machine learning methods could be utilised when designing buildings.
What kind of problem is this project trying to solve?
The government’s objective is to make Finland the leading applier of artificial intelligence in the world. Contrary to many other sectors, the construction industry has yet to explore the potential of applying AI to its operations. The project’s objective is to find out how machine learning methods could be utilised when designing buildings.
The construction process, with its many parties and numerous execution alternatives, is very difficult to manage as a whole. There simply are not enough resources for calculating different alternatives, with the selection often falling on the solution that has been used the longest.
With machine learning methods, alternative solutions could be generated with computer programs. Modelling a building’s details according to an image is a significant part of design, so even a partial boost to efficiency in this work could be immediately applied in practice.
Objective: to examine the applicability of machine learning to building design
Machine learning is a sub-area of AI. Its aim is to make computers generate viable solutions by combining a large amount of starting data that would be impossible to analyse with other technologies.
The main objective of the project is to find out how BIM data models of buildings could be applied as starting data for machine learning and what the requirements are for being able to apply the method to building design. A BIM (Building Information Model) is a digital and three-dimensional set of data on a building’s entire life cycle.
The problem is how to present the data contained in the design models in a format understood by AI.
What is being done in this project?
The study focuses on a long-term digital archive that is very rare by international standards and that is maintained by a large construction sector design agency. The project strives to test the use of the material for solving a limited problem. Various AI methods are applied for analysing the material, and there is testing of whether a proposal for a design solution can be generated on the basis of this material. The subject of a test can be, for instance, the creation of a detail for a joint in a single building,
Once the project is completed, we should have information about how the BIM archive should be compiled and utilised in the future. Moreover, we will aim to assess what is needed for the method to be utilised more widely and what financial impacts would result from widespread use of the system.
A study related to the application of AI in this project is being conducted at the Laboratory of Signal Processing at Tampere University of Technology.
Construction engineers, architects and everyone involved in the designing of buildings.
Duration: 10 October 2017–31 December 2018
Implemented by: Sweco
Partner: Tampere University of Technology
Mauri Laasonen, Development Manager, Sweco