| id |
ascaad2023_035 |
| authors |
Cheng, Chi-Li ; Nagakura, Takehiko; Tsai, Daniel |
| year |
2023 |
| title |
A Synergy of AI Observation and Design Tool: Leveraging Multifaceted AI Techniques for Encoding Human Behaviors and Stories in Space |
| source |
C+++: Computation, Culture, and Context – Proceedings of the 11th International Conference of the Arab Society for Computation in Architecture, Art and Design (ASCAAD), University of Petra, Amman, Jordan [Hybrid Conference] 7-9 November 2023, pp. 502-516. |
| summary |
This paper presents an innovative AI-powered tool aimed at revolutionizing observational methods in architectural design. Its primary objective is to bridge the existing gap between designers and AI predictions, streamlining and enhancing the design process. The tool facilitates the creation of dynamic visualizations that predict human behaviours within 3D design models, adapting seamlessly to design alterations. This prototype showcases the potential for efficient AI-assisted design. The core of our system consists of an AI model that trains on data related to human behavior within environmental contexts. Our user-friendly interface empowers designers to interact dynamically with their 3D modelling tool, akin to playing an interactive chess game. Designers can populate their models with human characters, and the system, in turn, predicts the likely activities of these characters. Observational techniques are pivotal in architectural design, drawing inspiration from influential works such as those by Alexander and Whyte. They provide a comprehensive understanding of how spaces can foster human interaction and help architects, designers, and urban planners make informed decisions that enhance user-friendliness. Nevertheless, two key challenges hinder the effective utilization of this data. Firstly, there is a lack of an intuitive interface that seamlessly integrates with existing tools. Designers often struggle to translate the information into design parameters and interpret the data effectively. Secondly, architects must adapt to evolving living environments and cultural shifts, necessitating real-time observations. However, time constraints and biases impede this process. A solution allowing designers to easily update their data is imperative. Our system comprises three integral components: a pre-trained model adaptable to specific locations, depth prediction and segmentation models for spatial comprehension, and a recognition model for user-designed structures. These features, combined with a user-friendly interface, empower designers to interact intuitively with their models, facilitating more informed and responsive design decisions. |
| series |
ASCAAD |
| email |
|
| full text |
 file.pdf (614,279 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2024/02/13 14:34 |
|
