Sam
MIT Media LabGNL, Projects, Sam
Publications   •    Media   •    TOT Talk '99   •    Resources
Avatars
Animal Blocks
BEAT
DollTalk and Zia
MACK
REA
Relational Agents
Sam
Research Team:  
Kimiko Ryokai
Austin Wang
James Dai
Jae Jang
Sam: a collaborative storylistening system for children
  Imagine a 3-D animated virtual child who can pass physical toys back and forth to real children, and construct stories with them collaboratively.

Bringing together our work in story-listening systems for children and in embodied conversational agents, we are developing a conversational character, Sam, who can act as a peer playmate to children and can create stories with them by sharing physical objects across the real and virtual worlds, and by listening and understanding the child's input.

The character and child share a play space and a set of story-evoking toys that can magically exist in both participants' worlds. Sam listens to what the child says and co-authors coherent stories with them. The background of Sam is the real-time video of the child's environment, so that Sam actually exists in the child's play space.

Recent developments include improving Sam's interactivity using speech recognition and natural language processing, improved models of turn-taking during fantasy play among children, an updated graphics engine, and optimized to run on one computer. Recent testing of the system shows the extent to which this toy engages children in creative and collaborative full-body play.

Publications   •    Media   •    TOT Talk '99   •    Resources