A generative design method for the functional layout of town planning based on multi-agent system

Abstract

In recent years, with the development of artificial intelligence and digital architecture, more architects begin to wonder how to generate urban planning and urban design through computational method. For the purpose of generating urban planning digitally using computational algorithms, we design a series of algorithms to develop a system that evaluates initial features of the site such as the strength of sunlight, water, landscape. These parameters related to the function zoning of the town were determined based on the data extracted from case studies. These data were integrated into a Markov chain mathematical model for the sake of analyzing the function of grid points. Finally, an algorithm of a multi-agent system was used to optimize the function that could evaluate the grade of each raster point of the town, which could be used to decide the function of a specific region.