menu
robots A-Zrobot timelinegamestv & moviesfeaturescontactfacebooktwitteryoutubetumblrrss feed
bg logo logobump



• Nico

Nico without his usual Yale sweatshirt

Nico is an upper-body humanoid robot developed at Yale University’s Social Robotics Lab, under the direction of Brian Scassellati (who cut his teeth on MIT’s Cog and Kismet).  Originally built circa 2004 to mimic the proportions of a one year old child, it has continued to serve as a research platform ever since.

Nico uses a similar mechanical set-up to the aforementioned bots, with a simple head (featuring two cameras per eye for wide and narrow fields of vision), a pair of arms, and a torso bolted to a table.  Microphones for speech recognition are placed apart from the robot’s body, along with a bank of sixteen computers used for image processing and other tasks. Its body (such that it is) has a total of 22 degrees of freedom (head x6, 2 arms x6, right hand x2, back and waist x2).

In 2007, Nico was touted as the first robot to recognize itself in a mirror.  Scassellati referred to the trick as “this dumb simple algorithm”.  Basically, Nico could classify what it was seeing as “self”, “other”, or “neither”. While it was not exactly the eureka moment it was made out to be in the press, it’s a start.

Nico’s vision system colors itself (reflection and limb) in green, & “other” in purple

He and Justin Hart, Ganghua Sun, and Marek Doniec programmed Nico to check itself out when confronted with the mirror test.  By having the robot move its arms, sensor data detecting the arm’s movements could then be compared with its camera images.  The vision side of the equation could estimate the arm’s position to within two centimeters in any direction, and if the sensor and image data matched, Nico could assume it was looking at a reflection.  But if it did not see any arm movement (or if it was off), then Nico would know with relative certainty that it was looking at another robot.

Nico has also been programmed to play a game of rock-paper-scissors (using simplified gestures).  An entertaining paper from 2010 entitled “No Fair!! An Interaction with a Cheating Robot” describes how a bunch of regular people were invited to the lab to play with the robot. Little did they know that Nico was programmed to verbally claim victory when it had actually lost, or (in an alternative scenario) would change its play after seeing its opponent’s hand.

As you can imagine, this twist had interesting effects on the participants. One of the problems with the verbal experiment was that people couldn’t quite make out if the robot was cheating or simply malfunctioning (“Did it misinterpret my hand gesture?”).  But with the action experiment, the cheating was more obvious (one person wrote “He cheats! Would change his sign after he saw mine!”).  In general people seemed to describe the robot as if it had a mind of its own, which is the impression humanoid robots ought to give.

Nico has played a role in a variety of research topics, including hand-eye coordination, language acquisition, shared attention, and evaluating autism.  The lab is also working on software that can interpret the emotional pitch, stress, and intonation of a person’s voice.  Some of the experiments are pretty amusing, like one where the robot instructed adults to throw text books into the trash, which they often followed – but only when the robot was physically present.  The experiment showed that people responded differently to robots in person than the same robot on a screen.

Further proof that the lab is a fun place: Nico can play a game of connect four, albeit using an electromagnetic appendage to lift colored ball bearings (although he might not be the most challenging opponent).  A video demonstration follows.  Yale is also home to the Grab Lab.

[source: Social Robotics Lab]

Further reading: Nico @ Yale Research (2), @ Yale Alumni Mag,
@ Yale Daily News (2), @ Yale News, @ Yale

Media

Video (Nico playing Connect Four):
YouTube Preview Image

Video:
YouTube Preview Image

Images:


Credit:
Seed Magazine