top of page
portfolio anne arzberger
THE ROBO-DOLL TOY
Ready to try out something ambiguous ?
Where does a doll become a robot? When does a robot become more doll then robot?
Aiming to search the space between doll and robot, we tried to measure the ambiguity of our doll-robot creations. Creating a highly interactive new practice we were able to transform the traditional Classification algorithm into a De-classification algorithm. Here is how it went:
Sketch your ideas
Curate your data
Never was the data curation part as important as when trying to use a classification algorithm to break categories. In an iterative and exploratory series of experiments, we created a set of images, ranging from toys over more simple sketches, textures and color schemes. Aiming to encapsulate our personal stereotype perception, we then went on with labeling those data. We intuitivley assigned each image to the binary categories of either 'feminine' or 'masculine'.
The DE-Classifier
.jpg)
With our curated data, we trained a classification algorithm. With our personalized tool to measure ambiguity, we started sketching ideas for our robo-doll. Each iteration was tested in front of the classification lens. Design ideas that tested to close to either the male or female category where not further explored. However the most ambiguous designs were followed and detailed.
The latent walk from unicorn to dinosaur
.jpg)
Learning about the aesthetic connections to gender by iterating through our design, lead to the definition of a few variable components. In order to capture texture and dimensionality, we then started putting our understanding of relevant variations of component into a clickable prototype. Adjusting arms, legs, bodies and capes, we were then able to receive more detailed feedback on our final design ideas. Although not all variations proved compatibility, we were able to identify and measure the weaknesses of our ideas. In further iteration we were then able to eliminate them.
Design final toy
bottom of page