Back to search

The Open University

Minds, Brains and Autonomous Human-Like Embodied AI Physical Drawing Machines for Generating Architectural Design Concepts

Abstract

dc:description.abstract

This thesis formulates part of an ongoing project to exploring and developing <i>Human-Like Embodied AI</i> for <i>Architectural Design</i> purposes. It seeks to take “… one step closer to emulating human-like intelligence” for architectural design. <br></br><br></br> Presently, <i>connectionist deep neural network (DNN)-based AI approaches</i>, such as <i>generative diffusion models</i>, have captivated the world. Aside from being accessible both in terms of their text-based interface and open-access across the internet, the flexibility, wide-ranging, greater details and speed in their content generation capabilities have definitively contributed to ushering ‘AI’ into the mainstream consciousness and professions, including <i>architectural design</i>. However, such computer-based alone are limited. There is ample interest towards AIs that can interact with the physical world through a physical body to extend these machines capabilities. <br></br><br></br> Research and developments related to embodied AI (EAI) using human-like form and structure such as the robotic arm and connectionist deep neural network (DNN) are in abundance. However, there remains ample mystery on machines as <i>emulation of human architectural designers’</i> embodied cognition in the <i>generative design</i> process via <i>traditional physical hand sketching</i>. <br></br><br></br> For human architectural designers, the features and functionalities of their embodied cognition in the generative architectural design concepts are complex. One of which includes the <i>morphology<sup2</sup></i> of the arm, i.e., form and structure. With it, they influence the human arm’s range of motion and the ability to produce marks. Just as therein lies the decisions made to move their arm, producing different marks within this design mark-making space. Lastly, as there is a crucial role in the interplay between the visualisations in the human’s mental imagery in the sketching process<sup>3</sup>, control and behaviour in their generative sketching task, which can be seen to pertain to the brain-body interpretation of embodiment<sup>4</sup> and structural-coupling. <br></br><br></br> This thesis therefore asks, <i>“what are the implications for machines to have a human-like embodiment such as the arm, sensor and brain to generate architectural design concepts via physical sketching?”</i> <br></br><br></br> To answer this question, this thesis responds to them in two parts: <br></br><br></br> Part I begins the inquiry from a theoretical and philosophical standpoint. Reviews on literature and terminologies on three core topics, <i>“Human-Like AI for Generating Architectural Design Concepts”, “Human-Like Embodied AI in Architectural Design”</i> and <i>“Physical Drawing Machines for Architectural Design”</i> demonstrate the aforementioned research gap as much as craft a theoretical framework for <i>embodied AI machines for generating architectural design concepts</i>. A taxonomy is created based on the interpretations found in the literature to a selection of reviewed <i>physical drawing machines’</i> embodiment and their similarities to human architectural designers generating architectural design concepts via physical hand sketching. Their findings are cross-compared with the outputs to reveal the relationship between their embodiment, embodiments’ human-likeness, and capabilities to <i>generate architectural design concepts</i>. <br></br><br></br> Part II investigates the question from a practical and technical standpoint. It reports on the modelling of the human arm a 5-DoF BraccioV4 Robotic Arm. With it, further studies are explored through its morphospace visualisation, custom-developed inverse kinematic (IK) solver, and integration with DNN-based image generative approach, such as DALL-E 2, creating DALL-E 2-BraccioV4. The latter can process can be thought to bear parallels with human architectural designers’ externalising their generated design ideas in their <i>visual mental imagery via traditional physical hand sketching</i>. <br></br><br></br> Finally, the research outcome of the two parts is combined. The robotic arm augmented DALL-E 2 system is evaluated using the taxonomy and through a similar manner in Part I. Simultaneously, insights from Part II also contribute towards the embodied framework that helped construct the taxonomy in Part I.

Degree

thesis:*
Name dc:type.qualificationname
phd
Level dc:type.qualificationlevel
doctoral
Grantor dc:publisher.institution
The Open University
Year dc:date.issued
2026

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Nguyen, Duong

Rights

Language dc:language
en

Chain of custody

source
Harvested from
The Open University
Base URL
oro.open.ac.uk/cgi/oai2
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Nguyen, Duong. Minds, Brains and Autonomous Human-Like Embodied AI Physical Drawing Machines for Generating Architectural Design Concepts. doctoral thesis, The Open University, 2026.