A Heuristic Approach to Design. Transitions from Generative Algorithms to Parametric Models | Un approccio euristico alla progettazione. Transizioni da algoritmi generativi a modelli parametrici

The paper critically addresses the issue of the automatic generation of distributional layouts that respect topological/spatial coherence and are obtained exceptionally quickly. This activity is of absolute interest within several scientific communities that collaborate fruitfully with each other: from IT experts, to scholars in the areas of Architectural and Urban Representation and Composition to name the main ones. We use the heuristic potential of digital drawing to explore some possible approaches inherent in the generation of layouts based on images, graphs, specialized structures and related hybridizations. Particular attention is paid to generative processes by training Generative Adversarial Neural Network (GANN) networks. The paper outlines the first outcomes of a research project involving a collaboration between Politecnico di Torino, Links Foundation and the Dutch gymnasium company Basic-Fit. In the first phase, the project critically analyzes the state of the art of Artificial Intelligence algorithms used in architecture; then, to demonstrate the potential of these tools, a first prototype was designed that respects the primary design constraints and criteria derived from the Company’s experience and is also characterized by a high level of interoperability in the generation of the selected solutions in the BIM environment.