Page 86 - Introduction to AI Robotics

69
3.1 Overview
psychology to formalize aspects of behavior. Schema theory has been used
successfully by Arbib to represent both animal and robot behavior. It is im-
plicitly object-oriented and so will serve as the foundation of discussions
through out the remainder of this book.

3.1.1 Why explore the biological sciences?

Why should roboticists explore biology, ethology, cognitive psychology and
other biological sciences? There is a tendency for people to argue against
considering biological intelligence with the analogy that airplanes don’t flap
their wings. The counter-argument to that statement is that almost every-
thing else about a plane’s aerodynamics duplicates a bird’s wing: almost all
the movable surfaces on the wing of a plane perform the same functions as
parts of a bird’s wing. The advances in aeronautics came as the Wright Broth-
ers and others extracted aerodynamic principles. Once the principles of flight
were established, mechanical systems could be designed which adhered to
these principles and performed the same functions but not necessarily in the
same way as biological systems. The “planes don’t flap their wings” argu-
ment turns out to be even less convincing for computer systems: animals
make use of innate capabilities, robots rely on compiled programs.
Many AI roboticists often turn to the biological sciences for a variety of
reasons. Animals and man provide existence proofs of different aspects of
intelligence. It often helps a researcher to know that an animal can do a
particular task, even if it isn’t known how the animal does it, because the
researcher at least knows it is possible. For example, the issue of how to
combine information from multiple sensors (sensor fusion) has been an open
question for years. At one point, papers were being published that robots
shouldn’t even try to do sensor fusion, on the grounds that sensor fusion was
a phenomenon that sounded reasonable but had no basis in fact. Additional
research showed that animals (including man) do perform sensor fusion, al-
though with surprisingly different mechanisms than most researchers had
considered.
The principles of animal intelligence are extremely important to roboti-
OPEN WORLD cists. Animals live in an open world, and roboticists would like to overcome
ASSUMPTION the closed world assumption that presented so many problems with Shakey.
CLOSED WORLD
Many “simple” animals such as insects, fish, and frogs exhibit intelligent be-
ASSUMPTION
havior yet have virtually no brain. Therefore, they must be doing something
FRAME PROBLEM that avoids the frame problem.