top of page

06 Oct 2020

The Robot Garden

SPAN, Matias del Campo y Sandra Manninger


SPAN, Matias del Campo y Sandra Manninger. Australian Pavilion for the Dubai Expo 2020, finalist, 2018.

The Robot Garden, Architecture & AI 

Architecture has rarely found points of intersection with the research conducted on Artificial Intelligence in a global scale1. Even today, the discussion of AI and Architecture has barely started. Considering the enormous potentialities of this area of research regarding its application in architecture, it is more than strange that this has not been discussed in wider circles within the discipline. In recent years we have seen a rapid development in the progressive methods emerging from AI research, resulting in applications that surround us continuously. 

Almost undetected, AI applications have seeped successfully into our daily life: voice recognition, ride sharing apps, banking apps, face recognition, AI Airline Pilots, smart home devices and more, are already naturally ingrained into our environment. More are in the pipeline that reach from AI driven Cars to farming with intelligent machines. The possibilities of these methods will transform all areas of our daily life and will mutate the planet. Consider the example of farming with intelligent machines. No need any more for giant monocultures of crops. If machines recognize the different species of crops in a quasi “natural” environment consisting of multiple plant species, they can become mechanical gatherers, roaming the landscapes, harvesting to feed the planet. 


SPAN, Matias del Campo y Sandra Manninger. Australian Pavilion for the Dubai Expo 2020, finalist, 2018.

The end of the monocultural farm as we know it, and the rise of synthetic natures to optimize the yield. There are several advantages to this method, one being that pathogens can not spread anymore like wildfire through a monoculture2 , another that the need for pesticides is reduced enormously – there are many more. For this to work, machines must learn to perceive and understand its environment and be able to respond to external stimuli in an autonomous fashion. Enter the Robot Garden.

How to Test a Robot

SPAN (Matias del Campo & Sandra Manninger) have been in touch with AI experts since the late 1990ies, when they first came in touch with the faculty of the OFAI, the Austrian Institute for Artificial Intelligence. One of the oldest of its kind, it was founded in 1969. These early meetings provided a basis for understanding the potentialities of this area of research, it was however the input by the Robotics Institute of the University of Michigan that turned out to be a game changer. 

SPAN, Matias del Campo y Sandra Manninger. Cassie, the main Biipedal Robot of Michigan Robotics.

After almost a year of conversations and experiments, the Robotics Institute offered SPAN the chance to design the Robot Garden, based on 2D to 3D Style transfer techniques that were specifically geared towards architecture design. What is the Robot Garden, you ask? First and foremost, it is a testing facility for robots. Michigan Robotics has specialized in exploring the possibilities of Bipedal Robots: robots on two legs. Combining its expertise in Machine Vision and Machine Learning, the team around Director Jessy Grizzle has made essential steps forward in the development of bipedal robots. 

These robots are designed to operate in areas normally designed for humans, such as factories, and in uneven terrain – think of the farming example mentioned above. In order o test these abilities, a testing ground was conceived, right next to Robotics new facility, the University of Michigan Ford Robotics Building. The outline given to SPAN called for a Robot Garden that contains a set of different terrains, from sand, to grass to gravel, to rockface. Inclinations and steps were part of the catalogue of features which were desired in order to interrogate the “last 50 feet”3 problem.

Posthuman Design is here

One of the first meetings we had about the Robot Garden was entirely dedicated to the conversation of agency in the age of AI’s. In recent years SPAN has discussed in lengths aspects of agency in a Posthuman environment. Just to clarify, when we talk about Posthuman, we do not mean “the age after humans”, but rather consider it an epoch which abandons the idea of human supremacy in the arts and design, and rather understands how other agents (or actors) can start to mingle with our own ideas about creativity and sensibility. 

SPAN, Matias del Campo y Sandra Manninger. Robot Garden, 2019.

The more we understood about AI, the more we saw how collaborative models of design can emerge from this area. The design technique used to design the Robot Garden is entirely embedded in this novel environment of design. Together with Alexa Carlson, a PhD candidate of Michigan Robotics, we developed a design method based on Deep Neural Networks. Deep neural networks have prevailed within a multitude of fields over the last few decades, including machine vision and natural language processing, due to their incredible accuracy at extracting salient features from input data and using these representations of our environment to perform tasks. This accuracy is thanks in part to the rapid development of powerful graphics computing technology, but also as access to big data has evolved, such that datasets can now start to capture the huge amount of visual variance that exists in the world.


Generative Adversarial Networks (GAN’s) came into existence in 2014 as a machine learning methodology devised by Ian Goodfellow4. Leon Gaty’s paper A Neural Algorithm of Artistic Style5 was published in 2015. SPAN’s experiments with the use of Neural Networks in architecture started around 20186 with the design of the Austrian Pavilion for the Dubai Expo and the first building project utilizing Neural Style Transfer (NST) as design method was the Robot Garden in 2019

SPAN, Matias del Campo y Sandra Manninger. First stage of the design, 2019.

Inhabiting the sphere of the imaginary can be a way to reflect on the aftermath of our current socio-spatial actions and what will be the future if we do not act and design the transition soon. What characterizes strongly the postmodern era is nostalgia that drives us towards an illusory flight to the imaginary, to the dream of community and the lost common soul. The related loss of any collectivity, the sense of the uncanny, the rise of individualism, consumerism and mass culture of virtual communication systems are some of the characteristics inherited from the 20th century to the 21st. In the introduction of ''The poetics of space'', Bachelard (1969) examines a mental capacity almost identical to the one that runs through the works of Heidegger: the poetic imagination. 

Big Data, AI and Architecture Design
Discussing Architecture and AI does mandate to touch upon the discussion about Big Data. Big Data allows Neural Networks to learn which goes beyond just collecting Big Data, it is not about Don’t Sort: Search7, but rather about how to crunch through this big data to extract the relevant information that allows to inform a project. It is literally about processing data to reveal information. Or to put it this way: Data is the new Oil8. Why? Because -in an analogy to crude oil- it is almost useless in its unrefined state, but needs to be refined into gas, plastics, chemicals etc. in order to create a valuable commodity.

In a similar fashion raw data is pretty much inert, as it is illegible to the human mind – it needs to be broken down and analyzed in order to reveal the valuable information. Yes, Data and Information are two distinctly different things. This is also what makes the use of Neural Networks so incredibly powerful. It would go far beyond the boundaries of this article to describe in detail the possible facets in the application of Neural Networks in architecture – reaching from site analysis, to plan analysis to improved methods of Building Information Modeling, to aspects of ecologic, economic and social impact of a project – the opportunities to reveal the profound nature of a project are gigantic. Using this notion, the project Robot Garden made use of a massive amount of satellite images in order to create databases that informed the distribution of different terrains on the given site. After various attempts, some successful - others less so- we found the right balance between the weights in the algorithm to come up with something that was useable as a testing ground for the robots. 

SPAN, Matias del Campo y Sandra Manninger. Closeup of Boulders showing the NN generated surface features, 2019.

In a second round, we designed the so-called Boulders in a similar fashion. The Boulders are designed to provide the test ground with a series of obstacles for the robots. We intentionally used the simple platonic solid of a dodecahedron as a basis for a deep neural network process in order to create the features on the boulders. These features include fissures, folds, pleats and erosion marks. All of these were entirely computer generated, through a machine learning process. All in all, this process presented itself as an opportunity to interrogate a posthuman design ecology, where human ingenuity and the abilities of Neural Networks to process large amounts of data created something novel that was neither fully controlled by humans, nor designed by an artificial agent, but was positioned somewhere between both these universes of thinking.


1. Spiller, N., (AI)CON, in Architectural Design Vol 65 11-12/1995, London 1995, pp XI - XII.


2. Newton, A.C., Exploitation of Diversity within Crops—the Key to Disease Tolerance? Frontiers in Plant Science, 2016; 7:665.


3. Xiachuon, L., Heng, L., Recognizing Diverse Construction Activities in Site Images via relevance Networks of Construction Related Objects detected by Convolutional Neural Networks. Smart Construction Lab, Hong Kong Polytechnic University, 2017.


4. I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Generative Adversarial Networks, Proceedings of the International Conference

on Neural Information Processing Systems (NIPS 2014). pp. 2672–2680.


5. L.A. Gatys, A.S. Ecker, M. Bethge, A Neural Algorithm of Artistic Style, aloiseprint 1508.06576, arXiv 2015


6. For example, the Austrian Pavilion for the Expo 2020 in Dubai, which utilized Style Transfer techniques to generate the ceiling. SPAN 2018 – Matias del Campo & Sandra Manninger. This design was a primitive effort in that it used the online solution by Google Deepdream to create Style Transfers. It used the Deep Style option to experiment with combinations between Baroque and Modern ceilings, transforming the resulting image into a 3D model using zBrush.


7. Carpo, M, The Second Digital Turn – Design Beyond Intelligence, Writing Architecture Series, the MIT Press, 2017


8. The phrase ‘data is the new oil’ was apparently coined in 2006 by Clive Humby, the British mathematician and architect of the Tesco Clubcard, a supermarket reward programme. (J. Bridle, New Dark Age – Technology and the End of the Future, Verso, 2019, P.245)


Originally published in Antagonismos Architecture Magazine, N6 Power. Buenos Aires, 2020.

bottom of page