Buscar

BIM

AI Architecture Generative Design Housing

Por Rose Barfield 5 min 29 de septiembre de 2019
AI Architecture Generative Design Housing

Will computers replace architects one day? That's the idea being put forward by Stanislas Chaillou. But is AI architecture efficient? If so, how far are we away from achieving fully computer-designed buildings? Should architects start retraining?


The AI-designed architecture is created using Generative Adversarial Nets (GANs)...
The layouts are generated in 3 stages: the footprint, the layout, and the furniture. A human must manually input details such as the door and the main windows. They also have to ability to tweak the layout at different stages of the design process. The computer generates the designs by calculating the most probable layout based on existing layouts.


Generative Adversarial Nets, is there anything they can't do?

Chaillou's concept is to design apartment block layouts using Generative Adversarial Nets (GANs).

Currently, in the CAD design world, one can only go far by hearing about the seemingly limitless possibilities of GAN. In the past, we have explored GAN generated photographs and even, computer-designed 3-dimensional models. There have even been generatively designed automobile parts. In BricsCAD, you can use intelligent tools such as BIMify to detect and label areas automatically, so why shouldn't you be able to generate whole layouts automatically?

The design process is broken down into three stages:

  1. The footprint
  2. The layout
  3. The furniture

This design process still requires some human input, and the architect can tweak the design at each stage. You can try it yourself.

null

Human interaction can manipulate the result. Images courtesy of Stanislas Chaillou

But how is AI architecture design possible?

It might seem a little mind-boggling that a computer can generate such intricate designs; however, what it's doing is developing a series of colored pixels that are then converted into a design based on input data. Each colored square becomes an area of the house; orange is the bedroom, green is the living room, etc. The walls are marked in black.

The technique uses a pix2pix GAN-model, a clever open-source code he uses to convert simplistic blocks of color into a more sophisticated render. He then combines this with Google deep learning tools.

null

The three stages of an AI-generated architectural floor plan. Images courtesy of Stanislas Chaillou

The design processes

To begin, the computer was "trained" to design the shape of a building footprint. Images were fed into it from real layouts of the city of Boston, which taught the computer typical footprint shapes for a given plot of land, and it used these to generate new footprints.

null

Image demonstrating AI-generated layouts based on input data. Images courtesy of Stanislas Chaillou

For the second step, the architect must manually place the front door (green square) and the main windows (red lines). This is then sent back to the computer, and a layout is autogenerated based on input data from over 800 real-life apartment layouts.

null

Images show how the computer fills in possible layouts from a footprint and simplifies the shape of colors to form rooms. Images courtesy of Stanislas Chaillou

Once completed, the computer calculates the layout of the furniture. The final results are pretty darn impressive!

null

Image illustrating how the colored spots (left) are converted to a fully functional layout (right). Images courtesy of Stanislas Chaillou

Multiple stories

Chaillou intends for his layouts to be used as apartments. In an apartment block, not all layouts are the same across floors. An option was added to allow the architect to specify the individual footprint of any given floor. However, load-bearing walls are not currently specifiable. This could be fixed using the same door and window placement technique.

What's next?

In future research, Chaillou hopes to be able to push his technique to far more complex and a-typical layouts.

The generated designs are not vectors, meaning they can not be imported into CAD software. More work needs to be done to make this a possibility.

null null null
null null null
null null null
Images Courtesy of Stanislas Chaillou

Should Architects start retraining?

The designs have limitations, and the current process only works with human input. I certainly would only approve a building design with the human review first! There's every chance you could have no bathroom or a completely inaccessible living area.

The computer isn't designing anything new it's simply calculating similar results from existing designs. If computers were to design all future buildings, the world would be a boring place to live in. The current design is also limited to the input data from building layouts of one region and needs to reflect a global audience's design styles and requirements. For example, Washington DCDC Washington might not be suitable for an apartment block in South Africa or Paris.

However, this technique may have some merit in game design. In large-scale, free-roaming games, I can see a generative design solution like this being huge. It would save the game designer from repeating the same layout endlessly or manually inputting different layouts while giving players a constantly evolving game.


Further reading:

Explore BricsCAD

Download BricsCAD Free Trial | Sign Up For News & Updates | Shop Online For BricsCAD Desktop Software

Rose Barfield

por Rose Barfield - CAD User Experience & Interface Design Specialist

LinkedIn logo (b&w)

Rose es responsable de recibir los comentarios de los usuarios y mejorar el producto BricsCAD. Antes de llegar a BricsCAD, era usuaria de CAD y trabajó en las industrias automotriz, aeroespacial y de defensa como ilustradora técnica. Le encanta descubrir cómo funcionan las cosas, desmontarlas y (con suerte) volver a montarlas.

17 de septiembre de 2024 3 min

Customizing Line Types in BricsCAD

Creating custom line types in BricsCAD can significantly enhance the detail and clarity of your drawings. Whether you’re working on complex engineering plans or simple layout designs, understanding how to create your own line types can give your projects a professional edge. Follow this step-by-step guide to learn how to create both simple and complex line types in BricsCAD.

10 de septiembre de 2024 3 min

BricsCAD Partner Apps for Civil and Survey Workflows

At Bricsys, we provide a solid foundation for our specialist survey and civil app developers to create specialist civil and survey tools that deliver robust and efficient workflows for BricsCAD users. Let’s see what updates and improvements our various developer partners have made to complement BricsCAD V24's civil and survey workflows.

These civil and survey applications are among multiple third-party applications supported by BricsCAD, with our developers working to improve their functionality and remain on pace with our new features introduced with every release.

9 de septiembre de 2024 4 min

Embracing Digital Transformation in the Construction Industry with Bricsys® 24/7

The construction industry continually evolves, and digital transformation is reshaping how teams manage projects, communicate, and handle project documentation. When embracing digital solutions for construction projects, a common data environment (CDE) like Bricsys 24/7 is an excellent place to start. With a significant variance in the level of digital adoption across the industry, embracing and leveraging digital tools can help firms differentiate themselves by elevating operational efficiency and improving project outcomes.

Síganos en las redes sociales