Related Projects:
SimRip
Grounding Verbs Via Motor Representations

Ripley: A Conversational Helping Hand
Cognitive Machines group

Researchers:		Prof. Deb Roy Nick Mavridis and Kai-Yuh Hsiao
Project Web site:		http://www.media.mit.edu/cogmac/robot.html
Group Web site:		http://www.media.mit.edu/cogmac

Ripley is a gripper robot that can learn to understand spoken language. He learns the meaning of words just as a child might—by looking at objects, manipulating them, and linking these experiences with spoken words.

Ripley is sensor-rich, with senses of touch, stereo color vision, stereo audition, gravity, and proprioception. Each of his joints includes both absolute position and force sensors.

Equipped with these sensors and an online visual simulator, Ripley can "imagine" what his environment looks like from different points of view and retain memories of objects he has seen in the past. He is able to learn about gestures using a Hidden Markov Model representation.

Possible future applications for Ripley include question-answering systems, language translation, assisting in deep-sea or space exploration, and assistive health care.

	TTT sponsors have access to more project details in the Sponsor section of the site. Learn more here.