开元体育app

Tutor:

Ximing Zhong (Aalto University)；

Jiadong Liang (Microfeel Company)；

Prof. Pia Fricker (Aalto University)

Studio Coordinator:

Xianchuan Meng

Teaching Assistant:

Chao Cao

Participants:

Mengchen Fan, Min Feng, Ye Feng, Heyu Gao, Hongmin Su, Junjie Su, Qiaomin Su, Yijing Wang, Yiping Xu, Jingwen Yang

The AIGC workshop, a collaborative venture between the School of Architecture and Urban Planning at Nanjing University and the School of Art, Design and Architecture at Aalto University, commenced on the 13th of June, 2024, and culminated on the 21st of the same month. The workshop's primary focus was to investigate the application of artificial intelligence (AI) technology in uncovering avant-garde challenges within the architectural discipline, with the aim of pre-emptively addressing prospective future challenges.

Introduction

This workshop is dedicated to exploring the potential applications of AI technologies, particularly Large Language Models (LLMs), Graph Neural Networks (GNNs), and Deep Generative Modeling (DGM), within the architectural domain. These technologies introduce a novel paradigm of collaboration, methodologies, and thought processes for architects, enabling precise understanding and simulation of complex design issues, performance optimization, and innovative expression. The central research encompasses:

1. Application of AI in the design process: Investigating the role of LLMs, GNNs, and GMs in architecture to uncover new workflows, such as rapid concept generation, structural optimization, and automated design iteration. Early-stage assessments aid architects in selecting superior schemes, revealing the prospects and potential of AI in contemporary architectural design.

2.Interdisciplinary collaboration: Emphasizing cross-disciplinary cooperation among computer science, architectural design, and urban planning, fostering knowledge exchange and technological integration, and promoting scientific research development, innovative thinking, and practical application. Exploring the training of future architects with AI tools to enhance adaptability to interdisciplinary collaboration and stimulate innovative thinking and technical capabilities.

3. Addressing challenges in the architectural field: Analyzing current and future challenges, such as automated design and intelligent planning, and researching AI's strategic responses. Through the analysis of outcomes and the summarization of practices, the workshop seeks to explore the theories and methods by which technological innovation can assist the industry in advancing towards intelligence and sustainability.

Presentation

The design outcomes were presented and defended on June 21, 2024, in the Jianliang Building. Ten students from the School of Architecture and Urban Planning at Nanjing University showcased promising design results, exploring the potential of Large Language Models from various perspectives in the fields of architecture and urban studies. The seminar was honored to invite Professors Ziyu Tong, Youpei Hu, and Yushu Liang from the School of Architecture and Urban Planning at Nanjing University. The professors provided insightful critiques and discussions on the workshop's outcomes, integrating their own reflections and practical experiences.

Group 1: Yiping Xu; Yijing Wang

Project Title: Chat with Dynamic Map

Based on the traditional physical space model, textual information such as crowd policies and norms is integrated. By clicking on any point and inputting user questions, design suggestions can be provided.

Group 2: Jingwen Yang; Min Feng

Project Title: Behaviour System and Space

Characters from the series "Westworld" were placed within the AI UGC framework we provided, cohabiting with the characters of "Westworld" and constructing the "Westworld" town themselves. From behavior to systems to spatial layout, the exploration delved into the different spatial outcomes resulting from the distinct behavioral patterns of various AI characters.

Group 3: Hongming Su; Junjie Su

Project Title: “Say Something” to Enhance Data-Driven Space Design

Convey your feelings and needs in a single sentence to influence the form and specific design of the space. Modify the data bias in the results generated by the graph neural network. Link the features of the agent with those of the graph network.

Group 4: Qiaomin Su; Ye Feng

Project Title: Interior Design Inspiration Generator: Hard Decoration Design Quotation and Furniture Layout

Provide prospective homebuyers with stylized hard decoration design estimates and furniture arrangement inspiration through floor plan uploads and language input at the initial stage.

Group 5: Mengchen Fan; Heyu Gao; Chao Cao

Project Title: Investment Decision Simulation in Virtual World Under a Certain Initial Environment

The AI agent makes spatial decisions based on cash flow, requirements, population, and the proportion of public space, offering decision-makers multiple cycles of AI construction simulation results and economic model assessments, thereby making spatial decisions more scientifically effective.