Waseda University’s Perceptual Computing Laboratory has posted a video showing their third generation conversational robot, SCHEMA, engaging in conversation with multiple groups. The robot listens intently as two humans talk about movies. When another person arrives, SCHEMA asks the person to wait while it finishes what it was going to say. It then directs its attention to the new comer, whom it has already recognized using image processing. What is important here is how quickly SCHEMA can respond to human questions and interjections. It doesn’t display any of the usual awkward pauses normally associated with talking robots. In fact, it almost responds too quickly! This type of work is helping to make robots more sociable and natural, and is the best work I’ve seen on the subject. Like the previous video featuring ROBISUKE, the video is also subtitled in English.
The lab won the Yamashita Memorial Research Award from the Conference on Spoken Language Processing Society Technical Committee two years in a row for their work on spoken language processing.
[source: Matsuyama720 @ YouTube]