Uncanny Valley Watch: Making Android Faces
The FACE Android can do semi-realistic facial expressions.
Credit: Face Team | University of Pisa

Can an android that looks convincingly happy or sad get beyond the creepiness of the "uncanny valley" trap? The FACE android's best efforts to imitate human expressions still leaves it floundering somewhere deep within the valley.

The idea of the uncanny valley coined by Japanese roboticist Masahiro Mori attempted to describe the uneasy or creepy sensation humans feel when seeing objects with incomplete human appearances or behaviors. An Italian team at the University of Pisa made the FACE humanoid robot to try mimicking realistic human facial expressions, according to New Scientist.

FACE uses 32 motors in its head and upper torso to make different expressions: anger, disgust, fear, happiness, sadness and surprise.

But even an android that looks human-like and flawlessly imitates human expressions can still fail the realism test in our brains. Researchers have suggested that the uncanny valley problem may arise from the mental disconnect between a robot's realism versus unrealistic appearance or behavior — a juxtaposition that partially tricks the brain into seeing a life-like human trapped within a creepy head or body.

The problem for the FACE android comes from its inability to move and behave realistically like a human. Instead, it holds each expression in a frozen manner before changing over to a different expression with an unnaturally slow smoothness.

On a scale where "1" means adorable and "5" means creepy, the FACE android earns a "4". Its appearance is not unpleasant, but seeing its slow, smooth transition between rigid expressions definitely brings out the creepy factor along with its fairly realistic appearance.

This story was provided by InnovationNewsDaily, a sister site to Live Science. You can follow InnovationNewsDaily Senior Writer Jeremy Hsu on Twitter @ScienceHsu. Follow InnovationNewsDaily on Twitter @News_Innovation, or on Facebook.