An MIT Alumni Association Publication

Meet Tega—one of the newest creations from the Personal Robots Group (PRG), led at the MIT Media Lab by associate professor of media arts and sciences Cynthia Breazeal SM ’93, ScD ’00.

PRG is working toward a future in which, simply put, we “live better with robots.” Its award-winning creations Nexi and Leonardo, for example, are designed to fit engagingly into peer-to-peer teamwork and family life. “Over the past few years,” Breazeal says, “our research has focused on advancing the artificial intelligence, user experience design, and application of social robots in the real world where they help people achieve long-term goals and can build personalized and positive relationships.” Educational goals are of particular interest: “There is huge need to help children enter school ready to learn, and social robots can offer something truly unique as an intervention both in schools and homes.”

Enter Tega, the product of extensive research on child-robot interaction and educational best practices. The development of Tega was led by former graduate student Jin Joo Lee SM ’11, PhD ’17, along with numerous contributors who designed and assembled early prototypes.

 

Research scientist Hae Won Park, pictured above, has spearheaded the interaction intelligence and deployment of Tega out in the field—most recently on a three-month literacy study in Massachusetts kindergartens, meeting weekly with children from 12 different classrooms.

Tega is equipped to tell stories to kids, then to conduct autonomous conversations about those stories, testing comprehension and vocabulary and making emotional or inferential prompts (“how did the frog feel?” or “what will happen next?”)—all while tailoring its hints and reactions to the child’s verbal and physical responses. Eventually, Tega invites the child to retell the story.

“By analyzing the story and speech sample, Tega can gauge that child’s language ability and which parts of a story the child is particularly interested in,” says Park. Relationship-building moments—such as conversations in which both child and robot share what they like about school—are key to nurturing richer, more personalized repeat interactions.

Tega’s bubbly, childlike demeanor makes it a unique research tool as well as educational platform. “As human beings, we are wired to learn from others,” Breazeal observes. And because it is designed to interact with kids as a peer rather than a tutor and to model productive mindsets, Tega offers a powerful, flexible social learning dynamic that PRG is doing rigorous experiments to better understand.

Research scientist Hae Won Park and a colleague with child-friendly robots.
Child-friendly robots line the walls of the Personal Robotics Lab.

Findings so far have reinforced the idea that “we learn not just knowledge and skills from others, but also important attitudes about learning—such as to be open and curious, to persevere through challenge, and to see mistakes as an opportunity to learn and grow.”

First published in Spectrum.

Next Up

  • Previous Article

    Making Machines that Speak Without an Accent

    Peter Dunn

    slice.mit.edu
    Today’s tech landscape features an increasing number of machines that talk, most of them in a neutral-sounding American accent. Pioneering research in the 1960s by Thomas P. ­Barnwell III is widely acknowledged to have helped launch this speech synthesis technology. Less well known is that his MIT friend William Jouris was the source of that quintessential accent. 
  • Next Article