Artificial intelligence makes more agreeable cities
Artificial intelligence can be of great assistance in urban planning, urban design and architecture. AI drives predictive tools and computational modeling to improve the quality of city living via complex adaptive systems, which is part of a methodology that is still incipient in architecture and urbanism but that, in Brazil, is already in use for an initial urban volumetric modeling project.
The interactions between methodology and urban design were addressed by the research project “Complex adaptive systems and rule-based design: applications in architecture and urban design”, with Maria Gabriela Caffarena Celani as the principal investigator.
Celani is a professor in the University of Campinas’s School of Engineering, Architecture & Urbanism (FECAU-UNICAMP) in Brazil. The project is supported via a cooperative agreement between FAPESP and the University of Melbourne in Australia.
“This is an activity that’s just getting under way in architecture and urban design,” Celani told Agência FAPESP. “It isn’t commonplace yet because it requires programming skills, which most architects don’t have and also because questions have to be answered before actions are implemented, which requires a culture change on the part of city governments. Nevertheless, although we’re still developing the methodology, it’s gradually becoming more widespread in policymaking, land use, and urban planning and design.”
While the modeling of complex adaptive systems originated in physics and the natural sciences, it offers a multidisciplinary theoretical framework that is highly appropriate for the study, analysis and design of sustainable cities with enhanced qualities for their inhabitants.
Justyna Karakiewicz, Associate Professor of Urban Design at the University of Melbourne and a member of the Melbourne Sustainable Society Institute, is also leading the project alongside Celani. For Karakiewicz, the performative attributes of cities are best understood as using complex adaptive systems (CAS) rather than through prescriptive regulation.
“It’s a methodology that enables you to see cities as open systems,” she said. “Cities are non-linear, unpredictable. We need to recognise the city’s behaviour as an emerging phenomenon".
"This, in turn, will help us focus increasingly on issues relating to resilience and on sustaining the features of the urban ecosystem and human functions instead of concerning ourselves with specific fixed parameters such as the density, functional mixtures, populations or the number of stories of buildings.”
Urban volumes were studied in this project using a type of CAS tool called cellular automata. An article published during the PhD research of André Araújo (a mentee of Celani) at FECAU-UNICAMP discusses how generative and parametric designs can be integrated with structural performances to enhance the design flexibility and control of different stages of the design process.
Cellular automata originated with the Game of Life, devised in 1970 by the British mathematician John Horton Conway. It is not a game in the conventional sense, as there are no players and no winning or losing.
The initial configuration was a two-dimensional grid consisting of cells with two possible states, alive or dead. Each cell interacts with its eight neighbors to create or replicate patterns.
Cellular automata can be used to predict and analyse how structures, plots of land or entire cities might interact with their surroundings. In the system devised by Araújo, the algorithm scans cells across the grid to identify the local conditions in specific cells in relation to the environment.
The article discusses the example of the Water Cube at the 2008 Beijing Olympics, but Araújo has also analysed the Bird’s Nest stadium, which is also in Beijing. “Some efficient solutions from the standpoint of structural performances do not necessarily represent a rational solution in terms of feasibility,” he said.
“Is this important when you’re working on a master plan, for example? Yes. In the future, it will be possible to simulate the effects of regulation in terms of imposing conditionalities on the site,” Celani said. “That’ s not what happens nowadays. Generally speaking, our urban laws don’t take local situations into account.”
The rules currently apply to large areas and overlook the needs of specific sites or small plots. “The idea is to use CAS in predictive simulations and then to test the results,” she added. “If we introduce conditional legislation applied to the situations of each plot and take specific aspects into consideration, what would happen if this were propagated to an entire region of the city?”