Ackerman, Evan. “Disney Research Makes Robotic Gaze Interaction Eerily Lifelike.” IEEE Spectrum: Technology, Engineering, and Science News, 2 Nov. 2020, 22:00 GMT, spectrum.ieee.org/automaton/robotics/humanoids/disney-research-robotic-gaze.
Article title:
Disney Research Makes Robotic Gaze Interaction Eerily Lifelike
Image: Disney Research
My summary of the article:
Disney Research collaborates with a team of researchers from California Institute of Technology, University of Illinois at Urbana-Champaign, and Walt Disney Imagineering and is at an ongoing collaborative process in creating a more humanlike 'illusion of life'. Their greatest challenge is to avoid the 'uncanny valley', a term that refers to a prevalent emotion of fear that is provoked amongst people when they are presented with an 'object' that is on the borderline between perfectly humanlike and not even nearly humanlike – i.e. that neither perfectly resembles a human nor completely fails to resemble a human. The team's focus is eye gaze. They believe that the way a robot (or an animated model) 'gazes' at a person is one of the biggest factors that define how humanlike it is – the more indistinguishable the artificial gaze is, the more humanlike the entity is. The robotic model in which the team is working on is a humanoid robot that reads a book without the presence of a person. To accurately resemble a real human's gaze behavior, the team has proposed four forms of gaze behavior: read, glance, engage, and acknowledgement. Which behavior the robot chooses to take is determined by the behavior (facial expression, eye contact, etc) of the person who the robot is attempting to interact with.
My response to the article:
When I noticed that one of the team's biggest challenge would be to overcome the 'uncanny valley', I was assured that the team is on the right track for creating a truly humanlike robotic or animatronic model. This is because I am fully aware of how visually creepy an attempt at creating a model that mimics human traits could be when it is neither perfectly humanlike nor completely non-humanlike. In fact, I still remember two memories of when I was very young, both of which directly refers to my experience with the uncanny valley. The first is actually from a non-horror film that I assume the director of it did not intend to provoke. The film is a 2004 film called "The Polar Express", and I still remember how oddly creepy the characters in that film looked like. Although the characters in that film were more "humanlike" than any other characters from animation films before, the fact that they were indeed more "humanlike" somehow made them more terrifying. The second is from a 2009 horror film called "Coraline". I still experience disturbance when I view images of faces with button eyes, and after realizing that this is due to the uncanny valley, I came to admire the director who cleverly utilized this phenomenon to deliver an indeed terrifying memory. But back to the research by the team, since it would not be desirable for the people who interact with the model to undergo the same experience due to the uncanny valley, I hope that the model they create do perfectly resemble an actual human.
Comments