Drawing Floor Plans with Machine Learning
Just as AI has enabled and advanced other industries, it is also growing in the field of architectural design. In this workshop, we will explore ways in which this technique can influence the future of architectural design. Traditionally, architects have always drawn buildings from scratch. With ML, we can reuse existing building designs to generate new ones.
In the workshop, participants will get a hands-on tutorial on how to incorporate ML into a small architectural design workflow. More specifically, we will look at ways in which we can automatically furnish a room within in a building, using Machine Learning. This will include preparing a dataset, training a ML model, and evaluating its performance on an unseen dataset. At the end we will explore the implications of using this technique, regarding speed, diversity, and human-computer interaction.
The workshop is targeted at MSc and PhD students, as well as professionals, who are specifically interested in incorporating Machine Learning in their work.
The outcome will be the evaluation of our trained ML model in an unseen dataset. Participants will be encouraged to create difficult test data points. We will compare the performance of the ML model to the manually drawn test drawings of the participants.
All participants should have intermediate experience with Rhino, Grasshopper3D and should have some experience with the Python programming language. They should bring their own laptop with Rhino and Grasshopper3D installed.
Jeroen is an architect and computer scientist. He holds a MSc in Architecture from Delft University of Technology (NL), and a MSc in Computer Science from University of Bristol (UK). After having worked as an architect and Artificial Intelligence (AI) researcher, he founded PlanFinder, a small software company which focuses on using Machine Learning (ML) to accelerate architectural design workflows. In the past, he has thought at Delft University of Technology as tutor in CAD and BIM software.