How can we define generative strategies to help inhabitants or experts to organize their own environment?

Many computational approaches to spatial generation were developed and debated in the 1970s and 1980s but they were too specialized and deterministic and so… ended up not influencing real architecture practice. Today, we have access to many techniques of form-finding and digital fabrication. However, in this quest for new shapes and among the widespread complex geometric techniques available for architecture, spatial organization remains an secondary topic.

Facing this gap, I searched for recent experiments of generative design related to spatial organization. The first example is Space Syntax programme and external form generator by Frano Bazalo.

This project is based on a excel sheet set to register the characteristics of the rooms (size, height, level and name) and its connections. This set is an input to a grasshopper definition that generates a three-dimensional space syntax diagram (a graph of the functions). Based on the attributes of each node or room, the algorithm generates an overall volume that is divided by a 3d voronoi diagram.

On one hand it is interesting to note that the author adopts part of these design strategies to develop some specific design experiments, using the resulting 3d voronoi diagram to generate also the structure. On the other hand, this strategy seems to work without real building constraints such as circulatory systems or site conditions. Besides, it needed to be more interactive to stimulate design exploration.

Another interesting project that I found is a multi-agent systems called PROBOTICS, developed at the Architectural Association Design Research Lab by Jose Sanchez, Knut Brunier, Anica Taneja, Diego Rossel (Course Tutor: Alisa Andrasek).

It seems to be formed by primary agents that move around and interact in vector fields. The environment is also populated by secondary structures that can leap to the orbit of these agents generating different patterns of organization. In this interaction of two levels of agents, the team was able to organize different infrastructures.

It is even more experimental than the first project presented, so it does not deal with real constraints and does not consider a specific programme as a pre-requisite. Besides, it is based on agents and it is more adaptable to a specific environment, which increases the space of possible solutions.