Page 50 - Concise Encyclopedia of Robotics

Bladder Gripper
For speech recognition, specialties include vowel sounds, consonant
sounds, grammar, syntax, context, and other variables. For example, a
context specialty program might determine whether a speaker means to
say “weigh” or “way,” or “two,” “too,” or “to.” Another program lets the
controller know when a sentence is finished and the next sentence is to
begin. Another program can tell the difference between a statement and
a question. Using the blackboard as their forum, the specialty circuits
“debate” the most likely and logical interpretations of what is heard or
seen. A “referee” called a focus specialist mediates.
For object recognition, specialties might be shape, color, size, texture,
height,width,depth,and other visual cues.How does a computer know if an
object is a cup on a table, or a water tower a mile away? Is that a bright lamp,
or is it the sun? Is that biped thing a robot, a mannequin, or a person? As
with speech recognition, the blackboard serves as a debating ground.
See also OBJECT RECOGNITION and SPEECH RECOGNITION.
BLADDER GRIPPER
A bladder gripper or bladder hand is a specialized robotic end effector that
can be used to grasp, pick up, and move rod-shaped or cylindrical objects.
The main element of the gripper is an inflatable, donut-shaped or cylin-
drical sleeve that resembles the cuff commonly used in blood pressure
measuring apparatus. The sleeve is positioned so it surrounds the object to
be gripped, and then the sleeve is inflated until it is tight enough to accom-
plish the desired task. The pressure exerted by the sleeve can be measured
and regulated using force sensors.
Bladder grippers are useful in handling fragile objects. However, they
do not operate fast, and they can function only with objects within a
rather narrow range of physical sizes.
See also ROBOT GRIPPER.
BONGARD PROBLEM
The Bongard problem, named after its inventor, is a method of evaluating
how well a robotic vision system can differentiate among patterns. Solving
such problems requires a certain level of artificial intelligence (AI).
An example of a Bongard problem is shown in the illustration. There
are two groups of six boxes. The contents of the boxes on the left all have
something in common; those on the right have the same characteristic in
common, but to a different degree, or in a different way. To solve the
problem, the vision system (or you) must answer three questions:
• What do the contents of the boxes to the left of the heavy, vertical
line have in common?
• What do the contents of the boxes to the right of the line have in
common?

