DigitalFUTURES: Artificial Intelligence for Resilient Urban Planning



Workshop Leaders: Angelos Chronis + Serjoscha Düring + Diellza Elshani + Nariddh Khean + Theodore Galanos / City Intelligence Lab, Austrian Institute of Technology

This workshop focuses on practical applications of AI within urban design. Through project-driven pedagogies, this workshop aims to take students with a basic understanding of machine learning, to applying almost state-of-the-art algorithms to urban-scale issues. In teams, participants will learn to identify problems amenable for AI-driven intervention, develop and train AI models to gain hands-on intuitions about hyperparameter tuning, and build interprocess pipelines for deploying AI models locally. Teams will be asked to develop project objectives which could range from assessing the quality and performance of urban spaces, to managing dynamic sub-systems such traffic through the use of reinforcement learning.

Artificial intelligence (AI) has permeated almost all facets of our digital and physical world. By now, it is a rarity to find an aspect of our day-to-day lives that has not been somehow influenced by an AI-driven technology. Even within the built environment, an industry that is known for its sluggish uptake of new technology, AI has been applied toward detecting anomalies in city-wide datasets, generating and optimising urban morphologies, and increasing the speed of onerous urban-scale microclimate simulations.

This AI in Urbanism charrette focuses on practical applications of AI within urban planning and design. Through project-driven pedagogies, this workshop is aimed at taking students with a basic understanding of machine learning, to applying state-of-the-art algorithms to urban-scale issues. Students will learn to identify problems amenable for AI-driven intervention, develop and train AI models to gain hands-on intuitions about hyperparameter tuning, and build interprocess pipelines for deploying AI models locally.

Leveraging several almost state-of-the-art ML models (such as generative adversarial neural networks and deep q-networks for reinforcement learning), linked to Grasshopper, participants will explore AI in urbanism within a sandbox model of an urban quarter.

In a hackathon-like fashion we will form teams and work on different problems related to urban planning, design, and simulation. This could cover issues such as generating streets and buildings using neural networks, assessing the quality and performance of space, managing dynamic sub-systems such traffic or policies through the use of reinforcement learning.

source

Save This Post
ClosePlease login