Page 30 - Robots Androids and Animatrons : 12 Incredible Projects You Can Build

Maintenance
Maintenance robots specially designed to travel through pipes,
sewers, air conditioning ducts, and other systems can assist in
assessment and repair. A video camera mounted on the robot can
transmit video pictures back to an inspecting technician. Where
there is damage, the technician can use the robot to facilitate
small repairs quickly and efficiently.
Fire-fighting robots

Better than a home fire extinguisher, how about a home fire-fighting
robot? This robot will detect a fire anywhere in the house, travel to
the location, and put out the fire.
Fire-fighting robots are so attractive that there is an annual national
fire-fighting robot competition open to all robotists. The Fire-Fighting
Home Robot Contest is sponsored by Trinity College, the Connecticut
Robotics Society, and a number of corporations. Typically a fire-
fighting robot becomes active in response to the tone from a
home fire alarm. During the competitions, its job is to navigate
through a mock house and locate and extinguish the fire.
Medical robots
9
Medical robots fall into three general categories. The first category
relates to diagnostic testing. In the spring of 1992, Neuromedical
Systems, Inc., of Suffern, N.Y., released a product called Papnet.
Papnet is a neural network tool that helps cytologists detect cervi-
cal cancer quickly and more accurately.

Laboratory analysis of pap smears is a manual task. A technician
examines each smear under a microscope looking for a few abnor-
mal cells among a larger population of normal cells. The abnormal
cells are an indicator of a cancerous or precancerous condition,
but many abnormal cells are missed due to human fatigue and
habituation.
Scientists have been trying to automate this checking process for 20
years using computers with standard rule-based programming. This
was not a successful approach. The difficulty is that the classic algo-
rithms could not differentiate between the complex visual patterns
of normal cells and those of abnormal cells.

Papnet uses an advanced image recognition system and neural
network. The network selects 128 of the most abnormal cells
found on a pap smear for later review by a cytologist.

Team LRN In the beginning