Keywords: neurorobotics, embodiment, artificial intelligence, body art, sculpture

The AI Prostheses

World Premiere: 17 September 2020, Kontejner, Zagreb

The installation consists of the robotic, AI prostheses created by Marco Donnarumma, with a team of collaborators, for his 7 Configurations (2015-2019) cycle: performances and installations that reflect on the conflicts surrounding the human body in the era of artificial intelligence (AI). Through combined research on movement, dramaturgy, sound and engineering, each of the works in the cycle combines human bodies, robotic hardware, machine learning software and microorganisms into a particular ‘configuration’, that is, a specific assemblage of human and machine parts capable of performing through each other.

The prostheses of the 7 Configurations embody uncanny combinations of the machinic with the organic. They are useless prostheses, paradoxical objects designed for the body, but not to enhance it, rather to subtract or upend some of its functions: Amygdala is a skin-cutting robot with a steel metal knife; Rei is a facial prosthesis which blocks the wearer’s gaze with a mechanical arm (as seen in Eingeweide); C and B are two robotic spines that function as additional limbs without a body (performing in Alia: Zǔ tài).

All the robots are custom-made from the ground up, purposely avoiding off-the-shelf technologies and their limitations. Amygdala, the first artwork in the cycle, served as a blueprint for Rei, C and B, which can be understood as conceptual and physical offspring. Each of them was designed according to the anatomy and performative skills of the human performer that would engage with it on stage. The prostheses’ morphologies, their materials and functions are, thus, the embodiment of individualised somatic relations between a robot and a performer.

The prostheses have been created to act as performers with their own agency and interact with their human partners without being controlled externally. The machines are therefore not instructed to perform predetermined movements, nor do they follow a score. Instead, they possess an artificial sensorimotor system that senses its environments and guides how the machine “perceive” the world and how it can react to a range of possible interactions.

This is possible thanks to an architecture of biomimetic neural networks and machine learning algorithms that perform in tandem with the mechanical body and the haptic sensors of each robot. Inspired by biological nervous systems, biomimetic networks are information processing algorithms that can self-generate movement and alter a machine’s behaviour in response to real-time, sensory data. This technique endows the machines with basic, artificial cognitive and sensorimotor skills.

With these skills, the prostheses perceive their own bodies in space, as well as the bodies of other human performers and improvise movements in response to external stimuli such as touch, pressure, pull and torsion. As they move, the prostheses learn about their partners and environments, constantly adapting and reacting to them.

Awards

2021, Magic Machine Award, RosyDX, C. Rockefeller Center, Netzwerk Medien Kunst and Technische Sammlungen Dresden (DE)
First Prize in Art Category
2018, German Federal Ministry for Education and Research and WiD (DE)
Artist of the Science Year

 

Shows

  • esc medien kunst labor | Wüste der Wirklichkeiten
    Graz, AT, 2023
  • Kapelica Gallery | European ARTificial Intelligence Lab
    Ljubljana, SL, 2021
  • Kontejner | Touch Me Festival
    Zagreb, HR, 2020

Technique

Amygdala, 2016-2018, robotic sculpture from stand-alone installation.
Media: Artificial skin, artist’s hair, epoxy, beeswax, FPGA computer board, custom AI software (adaptive neural networks, reinforcement learning algorithms), servo motors, 3D-printed body, aluminium chassis, steel metal knife.

Rei, 2018, facial robotic prosthesis used in the performance Eingeweide.
Media: Bacterial cellulose, FPGA computer board, custom AI software (adaptive neural networks, reinforcement learning algorithms), servo motors, 3D-printed body, aluminium chassis.

C and B, 2018, robotic spinal prostheses used in the performance Alia: Zǔ tài.
Media: FPGA computer board, custom AI software (adaptive neural networks, reinforcement learning algorithms), servo motors, 3D-printed body, aluminium chassis.

Credits

An artwork by Marco Donnarumma in collaboration with Neurorobotics Research Laboratory and Ana Rajcevic Studio.

Marco Donnarumma – Artistic direction, AI programming, engineering
Neurorobotics Research Laboratory, Beuth Hochschule – Scientific partner
Ana Rajcevic – Robotics visual design
Christian Schmidts – Robotics 3D modelling and engineering
Margherita Pevere – Robot skin
Hana Josic | Kapelica Gallery – Photography

Realised in the context of the Graduiertenschule, Berlin University of the Arts.