Research
Food for AI
Automating 71,334 Unique and Realistic Architectural Plans
University of Kansas
Georgia Institute of Technology
University of Notre Dame
2025
In collaboration with:
Cameron Ernst, M.Arch
School of Architecture and Design
University of Kansas
Athanassios Economou, PhD
Professor
School of Interactive Computing
School of Architecture
Georgia Institute of Technology
Yichao Shi, MS.Arch
PhD Candidate
School of Architecture
Georgia Institute of Technology
Xu Han, PhD
Assistant Professor
School of Architecture
University of Notre Dame

This project focuses on generating at least 5,000 unique architectural plans for single housing units using DrawScript and the Shape Machine. Each plan is rigorously crafted to meet stringent quality standards. In our first attempt, we successfully generated 71,334 unique plans. This dataset is publicly available, and anyone is welcome to download the plans via the link below:
71,334 Single House Plans
The goal of this project is to synthesize a dataset of high-quality, diverse designs that can be used to train advanced AI models and AI agents focused on residential architecture. By exposing the AI to a wide range of plans, the model can learn to recognize design patterns, optimize space usage, and adapt to various environmental and aesthetic requirements.
This large and diverse collection allows the AI to develop a deeper understanding of key architectural principles such as spatial flow, natural lighting, and circulation, as well as practical considerations like safety and efficiency. Over time, the AI will be able to generate new, optimized housing designs tailored to specific criteria or offer suggestions for design improvements.
This project not only pushes the boundaries of AI in architecture but also contributes to the development of AI-aided design systems aimed at enhancing the efficiency, quality, and creativity of future residential design and construction.





























