One of Brian Scassellati's studies inferred that the robot appearance does not matter very much as long as it is not overly human.

Should we still try to find an optimal robot appearance (i.e. through testing with puppets or something like that), or should we save time by simply picking a design and building it? Also, in order to attempt greater transfer, is it possible that, for example, we would want to make the robot more lifelike than a study on the children's interest might show? Or in other words, would finding out the appearance that the children are most interested in necessarily be the ideal, or would it be better to pick the type of therapy we want to achieve and design the robot to meet that therapeutic functionality?

The design doesn't need to be limited to a human head or body (although these might be easiest for many types of transfer). Dr. Scassellati used dinosaur robots, and other researchers have used Keepon (the snowman looking robot) for autism research. Would we want to use something completely inhuman like these, or focus on a human-like head and body?

---

« Discussion Page

Back to top