Communications of the ACM
Crossing the ncanny Valley
Japanese roboticist Masahiro Mori posited that a not-quite-human robot will strike people as creepy.
Credit: cubo.cc
Developers of robots and computer-generated films face the difficult task of making their creations as human as possible without looking too human, a perceptual gap that experts call the uncanny valley, a concept first proposed by Japanese roboticist Masahiro Mori. Mapping the uncanny valley would be very useful to filmmakers and engineers who hope to design social robots.
Indiana University School of Informatics researchers are trying to cross the uncanny valley by studying the quality of eeriness. The researchers tried to isolate the factors that affect how people feel about simulated humans by showing volunteers five video clips of animations and five clips of robots. The test subjects were asked to apply ratings from several scales to each video.
The researchers were able to identify four key characteristics—attractiveness, eeriness, humanness, and warmth. A robot that shows warmth and attractiveness will be easier to interact with than one that looks cold and ugly, according to Indiana researcher Karl MacDorman, while humanness and eeriness are needed to explain how to bypass the uncanny valley. MacDorman says he can generate a chart of the uncanny valley by plotting perceived humanness along one axis and eeriness along the other, using real data about people's feeling toward a specific robot or animation.
From The Economist
View Full Article
Abstracts Copyright © 2010 Information Inc., Bethesda, Maryland, USA 
No entries found