Tron

Since the AI focuses on energy supply and spatial efficiency, the design uses an L-system algorithm based on the shortest path to generate a dendrogram. The paths within this dendrogram form the building’s corridors, and the spaces at each corridor endpoint are generated using Voronoi shapes. These spaces and corridor structures are calculated through programming, which ultimately produces a steel frame and floor panels. Through this process, over 1,000 different building combinations were generated.

​

Among these designs, the most suitable building was selected based on criteria such as maximum space, structural stability, and the shortest distance between spaces. In essence, this building was nearly 100% designed by AI.

For the second building, the design method and process remain the same. However, since it’s intended for human residence as well, the Voronoi spaces were modified into rectangular shapes, allowing for horizontal and vertical floor panels suitable for human living, and AI could also inhabit this space.

​

As these buildings are generated by algorithm, they can be created automatically by setting the plot’s length and width, making them adaptable to any site. Additionally, I used Python to calculate the optimal site conditions for this building. Analyzing suitable geological conditions for the L-system-generated building indicated that urban and rock-based geologies are the most suitable.

​

In summary, this building was designed entirely using a parametric design approach to create an AI-friendly residence from start to finish. During the design process, I trained the AI with diagrams from books I collected, enabling it to incorporate these references.