ML for Urbanity

Workshop lead at the eCAADe '20 conference at the Technical University (TU) Berlin, Germany

Having the machine take on more decisions than the user has been ever so popular, it’s implemented through almost all segments of life nowadays. In the scope of urbanism and urban design, the optimal choice is often quite difficult. There are much more parameters and regulations to be aware of and we are limited as beings to make such complex decisions in a rather short amount of time. Even when writing these lines. That is the key point where we introduce machine learning, to help people make better decisions and thus reducing their potential errors. Starting from choosing the perfect plot to build upon and then building up through automated choices until we reach to the end where we would have an optimal urban scenario to put our design intent into. This workshop will take students through single\multi objective optimisation, regressions, clustering and then utilising some neural network autoencoders to accurately predict and define the perfect surrounding for any design intent.

urban morphology (re)generation from existing contextual data through optimisation and machine learning methods/models.

*piece out of Cottbus, Germany*

participant work - James Walker | Fredericia, Denmark

participant work - James Walker | Fredericia, Denmark

participant work - James Walker | Fredericia, Denmark

participant work - James Walker | Fredericia, Denmark

From extracting city data to regenerating the selected city block according to chosen urban parameters and regulations | Cottbus, Germany

From extracting city data to regenerating the selected city block according to chosen urban parameters and regulations | Cottbus, Germany