EDEN
Terraforming with, for, and by Nature
Can buildings rewild ecosystems? Can we promote the biodiversity, resilience, and productivity of ecosystems for the mutual empowerment of humans and Nature? We present a data-driven design approach to uniting human-centric cultural typologies with Nature-centric needs to maximize ecological thriving.
Research team: Christoph Bader, Nic Lee, Neri Oxman, Khoa Vu, Tim Tai, Nitzan Zilberman
While the five kingdoms of life thrive in the full spectrum of environmental conditions, buildings are designed for the comfort of human inhabitants: controlled environments that reject bacteria, fungi, and all but a few plant species. The result is a narrative of displacing non-humans to create human spaces. Recognizing that humans could not survive without non-humans—plants produce 100% of the oxygen we breathe—EDEN reverses the narrative, instead moving us toward a future of direct collaboration and cohabitation with Nature.
The EDEN research and design platform considers buildings as ecosystems rather than objects in space. In this view, any decision associated with the building’s design and its incorporation on site has a direct impact on the local ecosystem. In contrast to ‘net-zero’ construction, which aims to minimize negative impact on an architectural site, we seek to maximize ecological well-being through new forms of ecosystem construction with value to the natural world, across kingdoms.
In this approach, the combination of building and site should elevate and even augment the ecological impact of the building on the site, compared with the ecological contributions of the site on its own. EDEN aims to create a new paradigm in architectural design, entitled “Ecological Programming”. An ecological program is like an architectural program in that it determines space and adjacency needs, yet it focuses on these requirements through the common denominators that enable all life on Earth to thrive.
Working across scales and harnessing recent developments across disciplines—from generative artificial intelligence (AI) and reinforcement learning to architectural design, landscape urbanism, and applied ecology—we can address the complex challenges of designing architectural structures that not only meet the needs of human occupants but also promote biodiversity, ecosystem resilience, and the performance of critical ecosystem services.
Position
Architecture and ecology have since biblical times experienced an inverse relationship: the art of space and place-making comes at the price of environmental well-being. As the fields of architecture and urban design prioritize human users and inhabitants, we seek instead to maximize ecological resilience and propose that human-made spatial environments can and should augment—perhaps even reincarnate—natural ones as ecosystems thriving across eras, meridians, and species. By neglecting non-human agents and ecosystems, modern-day building and planning practices contribute to habitat loss, environmental pollution, and biodiversity loss.
As urban populations expand and climate shifts, preserving and promoting thriving ecosystems holds the secret to flourishing civilizations. When integrated, landform and biota elevate the net positive ecological impact of the site, tackling both the biodiversity and climate crises while increasing ecosystem resilience, and thereby promoting socio-ecological sustainability.
EST. NUMBER OF HECTARES OF DEGRADED LAND WORLDWIDE
In contrast to 'net-zero' construction, which aims to minimize negative impact, we seek to maximize ecological well-being through new forms of ecological construction with value across kingdoms.
When integrated, landform and biota elevate the net positive ecological impact of the site more than the landform on its own, tackling both the biodiversity and climate crises while increasing ecosystem resilience thereby promoting socio-ecological sustainability.
Platform
The EDEN platform is a computational framework for data-driven cross-scale generative design guided by rapid environmental simulation and generative optimization. State-of-the-art AI tools employ site-specific data to analyze and produce a vast array of architectural designs.
AVERAGE NUMBER OF SITE CONFIGURATIONS EXPLORED
In contrast to the typical top-down process of architectural and urban design, our platform offers a bottom-up computational design approach whereby policies are established to guide the form generation process to achieve specific, measurable design objectives along three axes of environmental impact: biodiversity, resilience, and ecosystem services.
The platform enables the generation of complete 3D structures as a direct output to the process of simultaneously computing a vast array of parameters, thereby promoting a single solution that addresses a wide range of requirements and challenges. In this way, products, or buildings—including their shapes and behaviors—can be “learned” from their environmental context and corresponding interactions.
Platform Details
Distribution-based generative optimization can be applied to discrete data sets like mesh objects, point clouds, and vector fields, creating configurations that are iteratively optimized toward specific goals. In the animation, a vector field directs the flow of objects toward target points along an arbitrary surface.
Neural-field based generative optimization uses implicit representations to translate arbitrary inputs into outputs, enabling optimization across entire contour fields rather than just specific points. In the animation, a neural-field defines three-dimensional objects, evaluates their curvature, and optimizes them for maximal curvature.
The third application of generative optimization focuses on exploring a complex design space rather than producing a single optimal design. In the animation, a generative optimization system maps the relationships between low- and high-frequency features of objects while optimizing them for maximal area and curvature.
Rapid environmental simulations are a novel class of tools that enable designers to assess the impact of design decisions on environmental processes in real time, adapting to various contexts from residential homes to urban centers. By integrating with optimization frameworks, these tools support data-driven approaches that enhance airflow management, runoff drainage, and sunlight exposure for local plant species.
The Capsule provides a controlled environment that simulates designed ecologies, allowing researchers to determine the effects of biotic and abiotic factors on plant growth across different ecosystems, regardless of location or climate. Integrated AI and bio-sensing platforms enable real-time measurement and autonomous adjustment of growing conditions, informing rapid environmental simulations and generative optimization workflows to enhance the design process.
Project
Pushing the boundaries of conventional architectural and urban design, we present studies for ecologically programmed environments, from small-scale pavilions to large-scale towers. In these studies, we explore a vast array of spatial and species distributions that prioritize a site’s biodiversity, resilience, and ecosystem services, aiming for positive environmental occupancy impacts.
The goals of each distribution strategy are based on the needs of the site, such as the remediation of polluted landscapes or the creation of protective habitats for keystone species. Once deployed in the wild, on-site data related to species count and abundance, soil composition, climate, and atmosphere are continuously gathered in order to inform additional design decisions that will further improve the site’s health and overall performance.
The EDEN studies unfold in two phases: data collection and design generation. In the first phase, we gather large quantities of data associated with the biology, behavior, and ecological requirements of various plant and animal species, as well as data associated with environmental conditions and ecosystem interactions of a chosen site. This information provides the foundation for the computational design framework—a “computational rulebook”—revealing insights and increasing our understanding of the unique requirements and characteristics of the targeted ecosystems.
In the second phase, we apply generative optimization to the gathered data to explore the vast solution space of architectural configurations. Generative algorithms then iteratively refine the potential solutions to maximize ecological well-being while ensuring that human needs are met and minimizing resource use. Factors considered as part of the optimization process include environmental conditions, habitat connectivity, resource availability, ecosystem stability, and the provision of specific ecosystem services such as carbon sequestration or air purification.
Tower Scale
Can a tower act as the infrastructure for the delivery of regulating and provisioning of cultural ecosystem services to the city it resides in? With more than half of the entire human population residing in tall and dense urban environments, we ask how space and place-making in high-rise typologies can augment rather than degrade ecological services.
The EDEN Tower proposes a novel vertical typology whose urban footprint is minimized while ecological surface area is maximized. Structural, programmatic, and ecological elements are integrated as part of a single organizational system. The system centers on a primary truss level that functions as both a stabilizing structure and a suspension mechanism, supporting the four lighter lower levels while anchoring the heavier ecosystems positioned above.
Grassland and forest ecosystems growing on the tower’s exterior manage regulating services for natural processes such as thermal buffering and carbon sequestration. Natural water reservoirs emerge in the sloped and curved surfaces, enhancing the biodiversity of the ecosystem.
Transparent interior spaces facilitate human-centered cultural services such as recreation and education. Interstitial zones host services for provisioning material resources such as timber from a young forest, fibrous materials from open fields, and foraging and pollination processes in flower meadows.
The deliberate design of functional space creates a co-dependency between human and non-human systems. While the tower provides an infrastructure for ecosystems to flourish, the ecologies, in turn, generate essential ecosystem services that support human life, fostering a symbiotic relationship between the natural and the constructed, the human and the non-human.
The EDEN Tower showcases the ability of ecological programming to maximize a site’s biodiversity, resilience, and ecosystem services, while also acting as a public gathering space. Insights generated from this initial structure will inform future design interventions at the architectural scale.
Generative Optimization for the EDEN Tower
Next, these ecological niches are populated with local species according to specific ecosystem services to be provided by each floor. Neural-field based generative optimization specifies the location of every species on the tower according to its needs for sunlight and water, all while optimizing towards increased biodiversity and the provision of the specified ecosystem services.
Pavilion Scale
Beneath a pastoral English landscape of manicured gardens, we discover an origin story of spatial control and species manipulation, where the sumptuous art of monoculture has overtaken the dignified practice of biodiversity. The earliest records of the native ecosystem comprising this site are documented in the Celtic Ogham alphabet, where each ancient letter corresponds to a sacred plant.
In this design proposal for a small-scale architectural pavilion, we seek to resurrect a landscape that has been lost to a millennium of human-centric design. The EDEN Pavilion begins its life in a state of carbon negativity. It continues to sequester more and more carbon dioxide as it grows and stabilizes, over its lifetime and into its afterlife. In doing so, it nourishes an ecosystem where landscape augments architecture, architecture renews landscape, and human and ecological programs intertwine through nutrient cycles and energy flows.
Rooted in permaculture principles and the art of Hügelkultur, the material system—deployed across material, structural, spatial, and environmental modalities—involves the construction of elevated mounds using decomposing wood. This technique emulates the forest nutrient cycling processes, whereby fallen trees become ‘nurse logs’ that support other plants by providing a platform for root attachment and nutrient release as they gradually degrade.
One tree in the Ogham, Ash, now faces catastrophic decline due to the spread of invasive fungi. The tree’s death threatens to release stored carbon, accelerating the destabilization of ecosystems. By burying a sub-structure of infected ash trees, we not only provide a physical structure for the pavilion’s architecture but also trap stored carbon, providing a carbon-negative foundation for the structure that is augmented by the sequestration of atmospheric carbon by the plants it supports.
Generative Optimization for the EDEN Pavilion
Sunlight and rainfall are simulated for a typical year at the site of the EDEN Pavilion. Ecological niches or “zones” are then constructed based on simulated environmental factors and local soil conditions.
To determine the species layout, density, and distribution of plants throughout the site, we created a computational 'rulebook' that integrates both extrinsic and intrinsic parameters influenced by the spatial environment. This design system optimizes for three objectives: enhancing biodiversity, maximizing carbon storage, and promoting long-term growth and resilience by considering energy distribution and soil chemistry.
Praxis
Terrestrial and marine environments sequester over half of our greenhouse gas emissions each year—the single most effective mitigation of climate change on Earth (Friedlingstein 2020). As climate change intensifies, applying ecological programming to design and restoring biodiverse ecosystems with multiple layers of resilience can promote their reliable provision of ecosystem services over time, ensuring the long-term health of both humans and the natural world.
The ideal ecosystem cultivates relationships between organisms and their environment sustaining the system’s long-term health, as marked by its biodiversity, resilience, and productivity. Fungal networks decompose fallen leaves in deciduous forests; ground cover shrubs reach for sunlight that filters through a canopy. The effective provision of services is closely linked to an ecosystem’s overall biodiversity and resilience. Diverse ecosystems are more effective at leveraging the energy available to them, which in turn, reinforces their productivity.
Shown above is an example of generative optimization on a field of 2,000 solar panels. Over several hundred iterations, the position and orientation of every solar panel is optimized to maximize annual solar exposure and energy production.
Neural-field based generative optimization is implemented to train a continuous function that outputs the optimal orientation at every position on the terrain.
An application of neural-field based generative optimization in conjunction with rapid environmental simulation to maximize biodiversity on a one-hectare site, while ensuring that plants receive adequate sunlight and water.
Structural elements and large trees are designated as fixed elements which are optimized around. Over time, the system increases biodiversity while decreasing the number of plants receiving inadequate sunlight and water.
An example of diversity-search-based generative optimization exploring the relationship between greenspace, structural layout, and operational carbon emissions for a simplified master plan.
Structural elements and large trees are designated as fixed elements which are optimized around. Over time, the system increases biodiversity while decreasing the number of plants receiving inadequate sunlight and water.
The fourth method of generative optimization operates on the behavior of agents and policies. Reinforcement learning based generative optimization tunes the behaviors of semi-autonomous systems over time, directing them towards designed objectives.
In the animation above, a system optimizes a simulated phytoremediation strategy to maximize ecosystem stability during the remediation process.
While much of architectural innovation today focuses on improving operational energy costs to achieve sustainability, we see the solution in the forest.
By shifting the focus from purely human-centric approaches to ones that recognize the interconnectedness of all living beings, we seek to promote a sustainable and flourishing future for our planet across all kingdoms of life.
Credits:
Consultants: Voxeljet, VA-Arts, Salehi / Kushi Studio
Acknowledgments: Photography: Nicholas Calcott, Phillip Le; Technical Advisor: Jun Sato; Tobias Wallisser
