Environmental AI
DigitalFutures 2021
Course: DigitalFutures 2021 Workshop
Instructor: Jiawei Yao, Minggang Yin, Chenyu Huang
Date: June.- July., 2021
The emergence of environmental performance-based design aims to provide solutions and strategies for cities and buildings to cope with climate change and provide more comfortable living spaces. Traditionally, environmental performance is only used as a conceptual design guidance and evaluation factor. The adjustment of the space shape is mainly carried out by designers and consultants, and the passive design methods of "blind box" and "experience" are carried out. Under the circumstances, environmental performance-driven design, which is spawned by the rapid growth of artificial intelligence tools, is an active design idea. The method emphasizes on the environmental performance in the stage of schematic design, through simulation analysis, result feedback, and iterative optimization to guide generative design.
KEYWORDS:
Environment Performance, Digital Design, Artificial Intelligence, GAN (Generative Adversarial Network)
OBJECTIVES
- Advance research on environment performance optimization
- Basics of machine learning and applications of AI algorithm, such as CNN and GAN
- Multi-objective optimization, genetic algorithm, reinforcement learning
- Environment performance simulation, dataset preprocessing
- Using trained model for environment performance evaluation, sensitivity analysis
Research Framework
