Robotics Reference
In-Depth Information
ated a version of his tortoise that was capable of learning, which he called
Machina Docilis
(machine capable of being tamed). His idea for the
Machina Docilis
was to train it in much the same way as the Russian
physiologist Ivan Pavlov had trained his dogs.
4
Pavlov employed two re-
flexes with his dogs, the ringing of a bell and the arrival of food. The
dogs became accustomed to getting food whenever the bell was rung and
so they associated the sound of the bell with being fed. As a result, when-
ever they heard the bell, Pavlov's dogs would salivate even when there was
no food in sight or within smelling distance.
Wa l t e r 's
Machina Docilis
was created by grafting on to the back of
his
Machina Speculatrix
what he called a Conditioned Reflex Analogue
(CORA)—a learning box. This provided the robot with the ability to
form conditioned reflexes, associating the physical action of being kicked
with the hearing of a whistle. This conditioning could be reinforced by
giving the robot five or six kicks, each one straight after a whistle, but
if the conditioning was not continually reinforced it wore off, so that
subsequent whistles would not create the same reflex action in the robot.
The learning box created a connection in the robot's “brain” between
the robot's light reflex and a whistle, and another connection between the
robot's contact reflex and the whistle. The robot could be trained so that,
for example, whenever the whistle was blown the
Machina Docilis
turned
and backed away from an obstacle which it imagined to be in its path,
just as the conditioned reflexes of Pavlov's dogs caused them to respond
to the sound of a bell by expecting the arrival of food.
Walter's tortoises had a tendency to explore the environment actively
rather than to wait passively for something to happen. They were never
still, except when “feeding”, i.e., recharging their batteries which they
did in a specially built hutch. At the top of the hutch was a lamp. Wal-
ter noticed that the tortoises wandered around a lot, but eventually, as
their battery gradually ran down, they would turn towards the hutch in
order to recharge themselves. Once a tortoise had “fed” itself in this way
it would leave the hutch to search its surroundings for new sources of
light:
4
Pavlov is renowned for his landmark study on conditioning. When he launched the study
Pavlov was actually observing the physiological effects of eating in dogs. He began by studying
digestion, but began to observe that the salivation of the dogs was very curious. He would place
meat powder or some other food morsel on the dog's tongue, waiting for the salivation to occur, and
soon observed that the dogs were salivating as soon as he entered the room, which was before any
food was even in sight. Since salivation in any animal is a reflex, Pavlov decided to probe deeper into
the conditioning of the dogs. This later became known as classical conditioning.
