A cyborganic sculpture ARBOR is a high-resolution 3D printed system designed by both human and non-human forms of intelligence and receptive for the interaction between human and non-human. Developed within the framework of the doctoral research under the supervision of Prof.Claudia Pasquero, the installation demonstrates a machine-learning-based design technic that is contextualised in a new form of material and formal articulation with an aim to impart biological intelligence into inorganic objects and synthetic environments. 




If we look at the microscopical pattern of wood we would see a great level of complexity embedded within its structure. It contains information about intelligent mechanisms of photosynthesis, growth, water and food distribution. ‘Biological structures adapt to external stimuli by growth-induced material property varia­tion resulting in hierarchically structured forms. Shape results from bottom-up material organization, sophisticated property gradation and functional hierarchies developed over time within single mate­rial systems’ (J.Vincent, O.A. Bogatyreva, Biomimetics: Its Practice and Theory. Journal of the Royal Society, 2006. pp471). The possibility to embed the organisationprinciples of an organic material into a digital systemallows to design a hybrid materiality which hosts biological intelligence within the digital structure.The project ‘Arbor’ outlines an approach for reading the intelligence of an organic timber structure by the means of machine learning algorithms, as well as rethinks the life cycle of wood, proposing a bio-artificial system which is alive in a cybernetic sense.


Developed within the framework of doctorate research by Maria Kuptsova and supervised by Prof.Claudia Pasquero at the Synthetic Landscape Lab, theproject proposes the use of Generative Adversarial Networks (GANs) as a method for extracting a material organisation principle from an existing database of timber structure for its further implementation in volumetric models.


GAN Style. CTL Tangential anatomical cut .  x200 

GAN Style. CT Transversal anatomical cut .  x200 


Wood is widely used in architectural construction because of its mechanical, environmental, economic, acoustic, elastic, thermal, hygroscopic and aesthetic properties. However, the heterogeneous properties possessed by wood in its living state are usually considered to be rather problematic. ‘Unlike other construction materials developed to meet specific manmade functional requirements, wood has evolved as a highly efficient biological system — a vascular tissue — to meet the support, conduction and storage requirements of trees’ (J.M.Dinwoodie; Building Research Establishment. Timber, its nature and behaviour. London, New York , BRE, with the support of the Centre for Timber Technology and Construction at BRE, 2000). To study the material organization of wood the project uses the large database on the botanical characteristics of different wood species developed by the ArchiWood project (https// Micro images of 995 species were collected and analysed with different contrasting techniques. ‘For each wood species, there are 3 anatomical cuts associated with the 3 planes of symmetry of the material taken in 3 different magnifications of the microscope: x40 (overview of the cut), x200 (main features anatomical from IAWA), x400 (highlighting fine lines such as punctuations)’ (F.Hallé, P.Détienne, P.Corbière, ArchiWood: dataset of legacy documents about wood anatomical, morphological, and architectural traits for plant species, 2017).The anisotropic properties are crucial characteristics of timber structure which represents its constant and reciprocal interaction with the surrounding environment. These qualities, which are usually considered as negative side effects, can be rethought and reused in a design of artificially alive systems which are capable to interact with the environment and develop functional capacities through this interaction. 


Volumetric model


The project proposes the use of GANs as a method for extracting a material organisation principle from an existing database of timber structure for its further implementation in volumetric models. For this research the StyleGAN2 variant of GAN was chosen in order to develop a technic which would allow an organic data to inform in-organic computational system with its own evolutionary neural network algorithm. A set of experiments was held to train it on all types of cuts and magnifications and the transverse cut in magnification x200 model was selected as the most consistent. This model was used to study the internal and external morphology of timber material organisation with the latent-vector interpolation translation into the Z axis of volumetric structure. In this process basic geometrical data describing the anatomical properties of timber structure of a thousand of timber species were extracted and memorized within a spatial topology. The discontinuous volumetric data set contains such information on the material organisation principles as: allocation of stiff and soft materials within a structure, gradients of fiber densities, as well as variation in hydrophilic properties.This network of curves and surfaces describes the distribution of material as behavioral pattern. Anisotropy of the wooden structure directly affects the distribution of material along the structure. constant and reciprocal interaction with the surrounding environment.



Fabrication technologies such as additive manufacturing allows for the development of adaptive fabrication methods informed by research in material behaviour. Wooden-based material system has been investigated through the large scale 3d printing experiments, suggesting a new regenerative life cycle of matter from the wood in its living state to a recycled woodenmaterial. 


Two different material systems have been tested for the development of synthetic prototype. First experiment has been conducted using wooden based filament containing 40% of wood and 60% of recycled PLA plastic. Second experiment has been focused on the development of a material system based on 99% content of wood with 1% of agar-agar as a binding material. For this test a custom-made end-effector has been designed to control the deposition of wooden paste. Material studies provide a new insight on how additive manufacturing can be used to develop novel approaches for timber construction based on bio-printing. Both methods have shown successful results and have been tested at a large scale. For the production of the “Arbor” sculpture wooden based filament with 40% content of wood powder has been chosen as the most suitable for a large-scale rapid prototyping.




Robotic manufacturing, regenerative life cycles of matter, as well as introduction of machine-learning-based design technic suggest the scenarios where cyborganic wooden structure could be artificially grown by the means of intelligent technologies. It challenges the processes of growth, decay and ontogenesis, introducing a form of cyborganic living object, bio-artificially grown in the city.



 Arbor sculpture, Potenziale 3 exhibition, Innsbruck, Austria, 2021


Author: Maria Kuptsova
Supervisor: Prof.Claudia Pasquero
Synthetic Landscape Lab, Institute of Urban Design, University of Innsbruck


Technical expert: Artem Konevskikh

3d printing material: Extrudr

Assembly support: Ernest Hager