Uncanny Valley, proposed by Japanese robotics expert Masahiro Mori (1970), is a controversial effect related to the design and likeability of robots. According to Mori, as similarity to humans increases, so should the perceived likeability and thus acceptance of robots. However, this is not consistently the case. In fact, too much human similarity seems to have an uncanny effect. To elaborate on the Uncanny Valley curve, two points in a coordinate system are necessary, the X-axis and the Y-axis.
The Y-axis is clearly given by the respondents' likeability scores, e.g., on a 5-point Likert scale. However, what about the X-axis? Which robot is more machine-like and which is more human-like? This question can perhaps be answered with 7 to 10 robots, but with more than 1000 it becomes much more difficult.
This problem confronted me in my master's thesis "Acceptance of Care Robots - Investigation of a Concept for the Care and Support of Elderly People in Germany". Originally intended as a tool for sorting robots, this problem developed into the main topic of my PhD thesis and led to the emergence of a novel categorization for robot designs. In this talk, I will explain the approach, the current state as well as the practical use for dealing with robot design, with the possibility to get indications for future robot acceptance.