Page 74 - Modern Robotics Building Versatile Macines

54 Modern Robotics

SOLVING PROBLEMS: DOING ENOUGH WITH LESS

With the need to make My Real Baby simpler and cheaper, Brooks
and Angle had to modify both its components and its behavior. They
had to find ways to fit the software on cheaper processor chips. The
five separate motors the prototype had used to control the facial
muscles were replaced with a single one that still allowed for the
necessary expressions.
As for behavior, the original BIT acted more like a real baby in
that it would keep crying until “fed.” But since My Real Baby had to
meet the more limited attention spans and patience of children, it
was modified so that even if not fed, it would eventually get over its
hunger and be ready to play again. And instead of having the doll
determine whether its diaper had been changed for a new one, they
reconfigured the doll to make it satisfied even if the same diaper is
removed and put back on—from “virtually wet” to “virtually dry.”
These changes highlight the difference between the research
robots found in Brooks’s MIT lab and the products sold by iRobot.
Research robots are designed for exploring interesting concepts and
problems in as much depth as possible. But while what is learned
in research can be applied to create innovative products, when an
inventor becomes an entrepreneur, a different goal must be kept in
mind. The customer can be intrigued but must ultimately be satis-
fied—and at a price he or she is willing to pay.
Another key to iRobot’s success may have been the way it has
been able to reuse and build upon its technological base. Thus the
logic and behavior for a doll could be applied to a robotic pet as well.
The behaviors for an automatic vacuum cleaner should be applicable
to, for example, an automatic mopper or duster or even a robotic
lawn mower. Additional behaviors can be added to account for par-
ticular requirements.

My Real Baby was the first fruit of this agreement, reaching stores
just in time for Christmas 2000.
While sales of My Real Baby were disappointing, the toy was a
milestone in robotics. As Rodney Brooks noted to Joseph Pereira
of the Wall Street Journal Classroom Edition Online, “for the first
time our robots [had] to interact with countless thousands of real