C.Soddu, E.Colabella
Artificial intelligence and Architectural Design

THE LOGICAL STRUCTURE OF "BASILICA"

The logical and operative structure of the simulation system of Basilica is based on the use of a main cycle, with auto-organization capability, and a set of ever growing secondary cycles. All are bound together.
Each cycle represents a whole structure in simulating the decision choices. It operates the transformation of the answers into possible shapes. This device is designed by:
1. The use of a paradigm to control the auto-organization procedures. This tool represents and controls the gained complexity and, in the meantime, is adaptive to incoming developments. This is the device that allows us to reply to a question that puts one of the possible formal matrixes into the paradigm.
2. The identification and sharing of the random margins between questions and shaping replies. The system uses and represents these margins as "operable fields" for the design choices. This improves the project evolution.
3. The set of possible formal matrixes, which are abstract shapes but usable in giving body to a set of possible performances. These formal matrixes are not a data base. They are generated by the interconnected cycles, by a set of simultaneous devices operating in a series of different fields, like geometry, dimension, materials, technology, complexity, and so on.
Therefore every formal matrix is the extemporary transformation of the contaminations and resonances into a set of different subsystems performed as a subsequent paradigm/random margin/formal matrix. All are in a subsequent omothetic complexity that looks like a fractal shape. At the end of every cycle (and of the related and multiple progressive nidifications) the result is:
1. an increasing complexity, and the related passage into a more evolved representation of answers, together with the proliferation of the same answer.
2. The production of needs, for the reason that every event we design was born also using subjective and random approaches. It was not necessary before but it became necessary after the choice: it is a part of the project history. This happens also if we, later, remove it because we consider this event as an obsolete one.
At first Basilica was only a research software to simulate, using AI procedures, the decision approach to design. Now we can use it for the management of architectural and environmental design, as well as for teaching. To use Basilica as learning tool we have built a user interface to evaluate and control, in different but simultaneous fields, the morphogenesis of the paradigm structure, its evolution and the sharing of the random margins to operate subjective choices oriented towards the intersubjective performances of the project.
The designer can build his own paradigm, change the structure of the relationship between possible events, change the geometry and stratify multiple possible geometries within the paradigm, define the quantity of possible exceptional events and the relationship between these events and the normal structure of the architecture. And, also, the designer can choose the time of evolution that he wishes to simulate. Mainly, Basilica is a tool to design the morphogenetic code of the artificial environments that we can manipulate through a set of quantic parameters.
With Basilica it is possible to design an artificial DNA able to generate a set of ever different, unpredictable and individually characterised artificial events. Every scenario, that is a 3D computerised virtual model of architecture, is recognisable as an individual of the same species. So we can identify, for example, a species as a typical design of an architect, the universe of possible solutions that the same architect can generate when faced with the same design problem. We have used our software in professional experiences, designing the incoming scenarios of the town environments and of new architectures.
For example, we have designed the DNA of the Italian Medieval towns. The problem was how to control in progress the identity and the recognisability of these environments in front of the increasing complexity of the actual needs. The challenge was to save the differences identified as morphogenetic code of this type of towns. The images (Figure
generative design Medieval town variation 11, generative design Medieval town variation 2 2 , generative design Medieval town variation 3 3 , generative design Medieval town variation 4 4, generative design Medieval town variation 5 5)
are a sequence of views of different 3D models generated directly with Basilica, that are different but belonging to the same "species". There are a set of nowhere virtual medieval towns that lived a parallel virtual time.
We have also used this program in a more usual professional experience: the development, with new architectures, of a quarter of the centre of Rome (Figure
Roma, Flaminio, generative design variation 1 Roma, Flaminio, generative design variation 2
Roma, Flaminio, generative design variation 3 Roma, Flaminio, generative design variation 4
Roma, Flaminio, generative design variation 5 Roma, Flaminio, generative design variation 6
Roma, Flaminio, generative design variation 7 Roma, Flaminio, generative design variation 8
Roma, Flaminio, generative design variation 9 Roma, Flaminio, generative design variation 10
Roma, Flaminio, generative design variation 11 Roma, Flaminio, generative design variation 12
6-17).
The sequence of these images shows two different parallel scenarios generated automatically with our software, and belonging to the same morphogenetic code that we have designed. The first 3D model is one of the multiple prototype and the second is the scenario we have used in our project. This scenario has been generated the last. It belongs to the last, more complex and "interesting" generation of scenarios.
The last two images

(Figure 18-19) show an architectural project we have done for Taiwan. We have gained the complexity of these buildings, and I think also this quality, using a sequence of 3D models generated with our software.

Last review 10 December 1995
The Authors
Bibliography
Return to the index of the paper
Return to the first page
Mail to Prof. Celestino Soddu