Generative and AI methods in architecture and urbanism: urban networks, smart mobility, and movement path planning, design, and computation [keynote] [predavanje po pozivu štampano u izvodu]

Abstract

The talk will address one of the case studies performed as a test of possibilities of generative and AI methods application in architecture and urbanism, or more broadly in spatial engineering and sciences. The research has been situated within the areas of urban computing and informatics, urban design, and urban analysis and analytics, being part of the broader strategy of smart, intelligent, and programmable cities, particularly targeting the subcategory of urban mobility and transportation regarding all layers of complex urban systems.

In that respect, a two-stage generative design problem has been defined, firstly at the scale of the urban city-region and territory (the construction of the transportation system as a dynamic and transformable network and geometry for specific performances), and secondly focusing individual actors and their performance within the chosen and designed system (the construction of agents' movement paths, experiences, or activities in general, as being determined by the defined parameters and with respect to the constructed system). Greater Paris (Grand Paris Express) has been chosen as the network and field of operation, implying the boundedness of each particular movement scenario to its domain of control and range of "legal actions" (the movement policy with regards to the spatial and relational arrangement and constraints of the designed network). The computational set of problem-solving methods that compose the path starting from the generative problem identification and definition and ending with its solution and visualisation will be presented, including specific exercises created for each part or a phase of the decomposed methodological set. Each exercise can be considered an autonomous problem or a part of the stated problem-solving and decision-making chain, which makes them multivalent considering possible applications.