Learning Machines -
New Methodologies for hyper-dense urbanism
Urban Design Studio - Technion 2017-2018
“Learning Machines - New methodologies for hyper-dense urbanism” is a thematic ‘research by design’ studio exploring cutting edge Informed Planning and Machine Learning methodologies for urban design. The studio focuses on the design of high density neighbourhoods not through top down masterplanning but through bottom up generative processes where quantifiable data and its translation into (machine) recommendations play a definitive design role.
The rapid urbanisation process taking place in Israel and the current strategic plans for more than 1.5 million units in the coming 20 years (literally multiplying the existing local residential market) demand urgent novel methodologies in urban design which can tackle the creation of quality urban living within a hyper dense urban environment. The aim of this studio is to explore and develop Integrative design solutions focusing on density not as a restrictive concept but as a measure of quality. This studio repositions the concept of density as a connective term in both urban planning and design; an instrument defined in quantitative terms density must fuse both planning and design in a qualitative manner.
Structuring a systemic approach governed by natural algorithms ( evolutionary clustering methods -EA, generative processes, Modular neural network decision making ) and driven by new explorations of evidence based analysis, simulation and parametrics may be understood as an attempt for total solutions in the urban scale. However, this research will provide nuanced understanding of computational systems, where resilience and adaptation are key concepts guiding a discussion which avoids spatial determinism and gives room for a wide design search; scenario based strategic thinking demonstrating hands-on the balancing act of stakeholders in the urban environment. Furthermore Generative, rule-based systems suggest a complete mirror image to the traditional masterplanning process which normally develop from the top down - from the abstract whole to local specificities. Generative design methodology flips the process on its head - moving from local rules of construction to global order through a series of constructive dualities - Difference-Repetition ,Homogeneity-Diversity and Whole-Parts
Students work by : Doron Revach, Rotem Lewinson, Lev Zitink, Boris Levin, Meisam Matar, Hanan Tansara, Rasha Boirat, Amit Eilam, Shahar Ben Efraim, Avi Cohen, Itai Ohad