A famously modular building is a good place to ask what modularity actually does to circulation. Floor plans become graphs, and the graph gets interrogated at three scales: many floors, one floor, one apartment.
Context
IAAC · Barcelona
Year
2026
Stack
Graph analysis · Machine learning · Python
Brief
La Cité Radieuse investigates how the spatial organization of Le Corbusier’s Unité d’Habitation shapes circulation, accessibility, and connectivity between apartments and rooms. Floor plans are converted into spatial graphs and examined at three scales: multiple floors, a single floor, and an individual apartment. Graph-based measurements reveal recurring patterns within the building’s modular organization, while machine-learning models test whether room functions can be predicted from their position, connectivity, and relationships within the overall spatial network.
01 · Context
The Unité d'Habitation organises circulation, accessibility and connectivity through repetition. Converting its floor plans into spatial graphs makes that organisation measurable rather than described.
02 · Scales
The same graph measurements are applied at three scales, which is what separates a property of the building from a property of an apartment. Recurring patterns surface where the modular logic repeats, and break where it does not.
03 · Prediction
Machine-learning models are given a room's position, connectivity and relationships within the network, and asked for its function. Where they succeed, the plan's logic is legible in its topology alone.
