Xiao Jin

Fully Automated Variable Multi-Unit Residential Design

Residential, Automated Generation, Ai, Customization, Modularization

The rise of the Metabolist movement in architecture during the 1960s was ignited by the Second Industrial Revolution. This sparked a new perspective that cities and buildings are not stagnant entities but dynamic organisms that undergo constant metamorphosis. Technological advancements have been instrumental in shaping this transformation of how we think about architecture. We now find ourselves at a pivotal juncture with the remarkable progress of AI technologies like ChatGPT and Midjourney, which have spurred the flourishing of innovative design paradigms.

Drawing from the current landscape, it is evident that densely populated residential environments in developing nations suffer from a glaring lack of diversity, rendering them incapable of catering to individual needs. Growing cities also face the challenge of disconnected supply chains and material transportation to the individual housing unit. Although infrastructure like water and electricity networks have reached residential units, essential resources such as deliveries, food services, and waste management remain unconnected. Therefore, the concept of object transportation becomes imperative, even extending to include the future prospect of transporting people, making it an integral aspect of material conveyance.

But why must architecture embrace adaptability and accommodate the evolving requirements of diverse individuals at different stages of their lives? The impetus lies in the increasing amount of time people spend within the confines of a particular space. The COVID-19 pandemic has stimulated patterns of working from home and extended periods of staying indoors. Furthermore, our aspirations for interstellar travel, space exploration, and habitation on distant planets all necessitate extended sojourns within a singular space. Consequently, it becomes imperative to establish a system that can meet the demands for flexibility, updates, and transformations, harnessing the power of advanced computing to record and optimize itself. This approach empowers individuals to customize and influence the overall appearance of their living environments, heralding the creation of visionary communities for the future.

Why prioritize high-density living environments? Our focus is on developing countries grappling with challenges posed by burgeoning populations, limited urban land resources, and a need for personalized, high-quality spaces. By implementing a new system, we have the opportunity to revolutionize low-quality, high-density living into high-quality residential realms.

Our approach entails training AI models using data from family single houses, enabling semantic control over housing programs. Promot like “100plan, 1 living room, 1 dining room, 1 kitchen, 2 bathrooms, 4 bedrooms, no balcony, no closet, no study room, program of a building, single-family house”. In this way to control the edge of the unit and program. Employing a modular stacking technique, we seamlessly combine multiple modules to generate collective clusters.

Component substitution plays a pivotal role, allowing for varying levels of components. At the micro level, we have windows, doors, and walls, followed by rooms at the intermediate level, and chunks at the macro level. Through meticulous categorization and the implementation of sophisticated algorithms, we ensure a profusion of viable options for component substitution.

In addition, I made a challenge in blurring the hierarchy of components. Compared with the traditional assembled building, which is divided into hierarchies at different scales, I used a holistic approach to blur such hierarchies. In addition, I use the analogy of the human body in terms of changeability, where some parts are less changeable like bones, and some parts are more changeable like skin and clothes, which is called the "anatomy of the building". It's not like modular construction where you would see the module, it's actually building up larger holes that read as a whole.

The program is classified into two categories: "Necessary" and "Special." The "Necessary" components encompass Entrance, Bathroom, Living room, Dining room, Bedroom, Storage & Check room, and Balcony, catering to essential functionalities. On the other hand, the "Special" components embrace captivating spaces such as Libraries, Climbing areas, Meditation corners, Swimming zones, Galleries, Fencing arenas, and Yoga studios, among others.

Embracing this pioneering methodology, our ultimate goal is to craft residential spaces that captivate not only visually but also functionally. Our research endeavors to transcend the confines of conventional architectural design, redefining the benchmarks of excellence in the realm of residential dwellings.

AI Typology Evolution

Component Details

Facade

Pixel to Vector to 3D

Prompt Control Program Test