Useful Uselessness? Teaching Robots to Knit with Humans

I am super excited that my first Pictorial was published at ACM DIS 2020. Fantastic colleagues at TU Berlin around Pat Treusch taught a commercially available robot to knit with them and with this pictorial Pat Treusch, Daniela Rosner and myself collectively reflect on this process through the visual format of an ACM Pictorial.

This pictorial uses imagery of human-robot collaboration, or cobots, as a site to examine the potential of queer use within design research. Through close documentation of our process, we reflect on acts of teaching a commercially available robot to knit with us—a messy and seemingly unproductive process. However, this uselessness of the chosen task allows us to re-consider the idealization of robotic collaboration. We question the optimization of a largely human labor force and the associated drive to increase efficiency within a range of sectors, from the service industry to industrial production. Building on non-use literatures examining technological limits, and drawing on performative explorations and critique, we show how knitting enlarges our capacity to visualize what might be a suitable use case for cobots.

Pat Treusch, Daniela Rosner and Arne Berger. 2020. Useful Uselessness? Teaching Robots to Knit with Humans. In Proceedings of the 2020 DIS Conference on Designing Interactive Systems (DIS) 2020. ACM. New York. USA ISBN 978-1-4503-6974-9/20/07. https://doi.org/10.1145/3357236.3395582

Robot arm on the left holds a needle in order for the human on the knit with them

It may seem odd to explore a technology that Time magazine declared to be one of the central innovations of 2018 in relation to the apparently “most boring cultural technology of the world”, namely knitting. Cobots are often culturally coded as requiring particular and rarified expertise while knitting appears to many as non-technical, simple or unthinking — a feminized hobby that has no relevance at a place like a robotics research lab. But it is precisely this sociotechnical paradox that we aim to examine through our design process.

The motion capture analysis (left image) of the practice of knitting reveals rather intricate and complex movements of a skilled human knitter.The right image shows a piece of knitwear produced in collaboration of human and robot. The robot appears to perform poorly at the practice of knitting.

The juxtaposition of human knitting (left) and robotic knitting (right) exposes some of the tensions emerging from a potential collaborative knitting practice. More specifically, the left hand of the human knitter had to adapt to the trained movements of the right needle which the robot is executing. Sometimes the needle turned out to be too slippery or not slippery enough — in both cases, a stitch might get lost and holes are produced.

four examples of neglected labor, done by humans to support the robot: plastics, 3d printing, masking tape, drinking straw

Some of the invisible human labor of adaptations and ergonomics that was needed for the robot arm to actually perform.