Expertise, playfulness, analogical reasoning: three learning mechanisms to train Artificial Intelligence for design applications.

By Gabriele Mirra
PhD Student

How can AI be trained in architectural and structural design and integrated with CAD software to support the design process?

Following the success of AI in statistical regression, image generation, and decision-making tasks, since 2014 new computational tools based on AI have been proposed for design applications. Engineers have used AI to improve the efficiency and computational speed of existing software for structural analysis and optimisation, whereas architects have started exploring the potential of AI for conceptual design applications. However, current applications of AI in design do not consider such aspects as AI-CAD integration and what and how different AI models learn. By focussing on these aspects, this PhD project aims to assess the potential of AI for developing more autonomous and participative CAD systems and contributes to understanding how AI will affect design practice in the future.

This PhD project hypothesises that AI can be trained to acquire design knowledge by simulating three learning mechanisms: expertise, playfulness, and analogical reasoning. In design education, expertise is related to studying and analysing design precedents; playfulness is linked to model making; and analogical reasoning pertains to finding inspiration in domains other than architecture, such as nature, art, music, and literature.

AI model trained to generate shell structure depth maps from the recombination of visual features extracted from a dataset of design precedents.

Through several conceptual and performance-driven design applications, this PhD demonstrates the potential of each learning mechanism as a strategy to train AI in design. It also shows how the trained AI models can be integrated with CAD software and interact with the designer to support the design process.

Outputs

Mirra, G. and A. Pugnale (2022). “Exploring a Design Space of Shell and Tensile Structures Generated by AI From Historical Precedents.” Journal of the International Association for Shell and Spatial Structures (published in fast track), DOI: 10.20898/j.iass.2022.008

Mirra, G. and A. Pugnale (2022). "Expertise, playfulness and analogical reasoning: three strategies to train Artificial Intelligence for design applications." Architecture, Structures and Construction 2(1): 111-127, DOI: 10.1007/s44150-022-00035-y.

Mirra, G., A. Holland, S. Roudavski, J. S. Wijnands and A. Pugnale (2022). "An Artificial Intelligence Agent That Synthesises Visual Abstractions of Natural Forms to Support the Design of Human-Made Habitat Structures." Frontiers in Ecology and Evolution 10,DOI: 10.3389/fevo.2022.806453.

Mirra, G. and A. Pugnale (2021). "Comparison Between Human-Defined and AI-Generated Design Spaces for the Optimisation of Shell Structures." Structures 34: 2950-2961, DOI: 10.1016/j.istruc.2021.09.058.

Mirra, G. and A. Pugnale (2020). Sketches of Thought: Inside the Black Box of AI. Future Prototyping Exhibition Catalogue, Melbourne School of Design: 100-101.

Mirra G., Expertise, playfulness, analogical reasoning: three strategies to train AI in architectural and structural design. DigitalFUTURES Doctoral Consortium, 21st June 2021, Web Meeting. Presentation accessible at: https://youtu.be/ttvU-6YOszU?t=3618.

Mirra G., “Expertise, playfulness, analogical reasoning: three strategies to implement AI in architecture”. DigitalFUTURES Young talk “AI & Creativity”, 28th February 2021, Web Meeting. Presentation accessible at: https://www.youtube.com/watch?v=Gj-fFXz1kqc&t=467s

Mirra G. and A. Pugnale, “Testing Generative Models (VAEs) for the design of shell structures”. IWSS 2020: 1st Italian Workshop on Shell and Spatial Structures, 25th-26th June 2020, Web Meeting. Presentation accessible at: https://www.youtube.com/watch?v=_GTnitGgT_s

Biography

Gabriele Mirra, PhD Student

Gabriele Mirra is an architect and a computational design specialist. He is an expert in acoustic design, structural design, robotics, and machine learning. He is currently completing a PhD in Artificial Intelligence in Design at The University of Melbourne, where he teaches computational design and acoustics.

Graduate Research Profile
g.mirra@student.unimelb.edu.au
www.gabrielemirra.com

Supervisors


Dr Alberto Pugnale, Main Supervisor
A/Prof Wally Smith, Co-Supervisor
Dr Eduardo Velloso, Co-Supervisor