ROBOTICS SCIENTISTS TURNING TO BUGS:
Forget the Humanoid C3PO from `STAR WARS',
Cockroaches make better models.
Maine Sunday Telegram / Portland Press Herald, May
31, 1998 Page: 1B © 1998
Guy Gannett Communications By Eric Blom, Staff
Writer
(Note: Text from the article
above)
Carlo Spirito studies
creepy-crawly things like cockroaches, crabs, spiders
and crayfish - a fact that made him popular at
theUniversity of Genoa robotics lab during his six-month
sabbatical there. The Italian engineers, who were
building a six-legged robot, peppered Spirito with
questions about how bugs move and how they sense the
world. They knew Spirito's critters could help them
build a better machine.
Lowly forms of life have
become a hot topic in robotics. Many engineers now see
bugs and beasts as easier to copy than humans, and as
potentially superior models of behavior. Biological
scientists, meanwhile, are starting to pay attention to
robots as powerful models for studying insect and animal
life.``Certainly, the boundaries are getting very
blurred'' between robotics and the natural sciences,
said Spirito, an associate professor of physiology at
the University of New England in Biddeford. ``There's
this group of us who want to use robots to understand
animals, and there are those who want to use animals to
understand robotics,'' he said.
This academic partnership
- broadly called "biologically-inspired robotics'' - has
implications for everyone. Proponents say it may soon
lead to new consumer, industrial and military robots -
things like robot vacuum cleaners, lawn mowers, cooks,
dusting machines and soldiers. And it is helping science
better understand living creatures.
Biologically-inspired
robotics also is raising ethical and
philosophical issues. Some observers say it is blurring
the definition of life itself - as robots are taught to
feed themselves, fight for survival, learn from their
environment and, in some cases, reproduce and evolve.
``Really, we're on the cusp of the next technological
revolution,'' said Scott Jantz, a doctoral candidate at
the University of Florida Machine Intelligence
Laboratory. The lab has built 30 insect-inspired robots
and a prototype robot lawn mower. ``It'll make the
(personal computer) revolution look like small
potatoes,'' Jantz said. ``You only have one PC in your
home, but you could easily have four or five robots in
your home, involved in your daily life.'' The
collaboration between engineers and natural scientists
is a relatively new development.
Randall Beer, an associate
professor of biology and computer science at Case
Western Reserve in Cleveland, compares it to peeking at
an answer in the back of the textbook. Nature has
already solved many of the problems robots confront.
``It's silly to ignore these real-world examples,'' Beer
said. ``Many of the things we'd like to do with robots
are done by animals as well as by humans.'' Engineers
now regularly look at how people and animals walk,
shifting weight from limb to limb and altering their
gait with the terrain. They create artificial spinal
cords, muscles, nervous systems and bodies for robots
based on animal models. They even write computer
programs to mimic the decision-making process of animals
and people. Biological modeling is particularly
important as robotics tries to overcome its biggest
hurdle: developing machines that work outside the
laboratory. ``Animals really have the same problem as
robots,'' Beer said. ``They don't have people walking
around after them, picking things up and helping them
out of jams.''
Still, until very
recently, robotics mostly focused on humans for whatever
biological inspiration it gleaned from the natural
world. When engineers and America's popular culture
envisioned robots, they thought of humanoid devices in
the style of C3-PO from ``Star Wars'' or Will Robinson's
friend on ``Lost in Space.'' ``They thought that if they
could come up with a machine with reasoning ability,
everything else would be trivial,'' Spirito said. But
experience demonstrated that not only were human brains
difficult to copy, but the physical skills of people are
hard to duplicate as well.
Setting sights
lower: So in recent years, many robotics
engineers have started to set their sights lower - all
the way down to the world of insects and other
invertebrates. Now robots frequently look more like the
spider in ``Charlotte's Web'' than the android on ``Star
Trek.'' For example, Beer is part of a team that is
building a robot cockroach - 17 times as large as the
real insect - that ``has got enough strength to run and
whatnot.'' ``That's the level of computer intelligence
we have available to us now,'' Jantz said of insects and
other creatures at the bottom of the food chain.
Bug-bots may not be smart, but they can be superior in
some physical respects to humans. They get around with
more stability, for example, on eight legs than people
do on two. And their mechanical bodies can be more
durable, powerful and responsive than flesh.
Richard Peck of Newport is
building a six-legged robot as his senior project at the
University of Maine. He chose that model because of its
relative simplicity and functionality. ``You have got a
lot more stability if you have four legs on the ground
at any time,'' Peck said. Researchers in Massachusetts
are using lobsters' superior ability to detect odors as
a model for their robot, which will identify or track
down chemical spills. They call it Robo-Lobster. A Yale
professor is using the sound waves produced by bats and
dolphins as a model for creating robots that can ``see''
in three dimensions, rather than just the two dimensions
available to humans. Researchers in several locations
are showing how even high-level activities can be
performed by robot bugs taught to interact with each
other in a kind of artificial-insect society. ``What
evolves out of that society (of insects interacting
together) is a very high level of function,'' Jantz
said. ``Ants can farm. They can do all kinds of things
that we associate with humans.'' This insect model is a
more realistic picture of how robots might really be
used in our lives than the science fiction ideal of
helpful, humanoid machines.
Doing jobs people
shun: ``You don't want to put doctors or
lawyers out of business,'' Jantz said. ``You need robots
for the jobs that people don't want to do. Those are the
robots that will be really valuable.'' You want robots
to take out the trash, defuse bombs, handle toxic waste
and demolish nuclear power plants. Some insects have
shown that they can thrive in environments like these
without complaint. ``Between the 1960s and the 1980s,
the artificial intelligence field had these lofty goals
of creating human-type intelligence,'' Jantz said.
``That goal was more academic than practical because you
don't want robots thinking for themselves. ``You don't
want them thinking `Do I really want to go to the bottom
of a volcano?' '' he said. Also, insect-inspired robots
would be a lot cheaper to produce than people-like
machines, and thus more useful in the real world. ``You
turn loose 500 of these autonomous cockroaches, and if
two or three of them stop working in the first couple of
days or hit a wall or whatever, it's no big deal,''
Spirito said. ``People aren't thinking about the perfect
robot anymore.''
Indeed, researchers at
Vanderbilt University in Nashville are designing robot
insects that will crawl and fly for the military.
Soldiers on battlefields and pilots in the sky would
release ``swarms'' of these robo-insects - each about a
third the size of a credit card - to spy on the enemy
and detect the presence of chemical or biological
weapons. And this kind of mechanical creature is just
the beginning of what biologically inspired robots could
accomplish. Scientists can take lessons from bug-bots
and move up the line to more complicated animals such as
reptiles, mammals and people. In fact, the robots can
help. They now often ``evolve'' in the same way that
biological creatures do in the natural world, Jantz
said. It's called ``emergent functionality'' among
robotics engineers. For example, Jantz programmed
several robots to recharge themselves from a specific
electrical outlet and set them down in the same room.
Survival of the
fittest: ``Without even programming it in,
the robots would fight for this resource and kill each
other off,'' Jantz said. ``The strongest one survives.''
This kind of so-called ``artificial evolution'' will
help scientists develop higher level robots - maybe even
approaching the human level - by learning from the
experiences of these lower-level forms of ``artificial
life,'' Jantz said. However, there will always be a role
for robotic insects and animals in the future, he said.
``The view of Rosie in `The Jetsons' won't happen, not
for technological issues but because of civil rights
issues,'' Jantz said. ``You can't enslave a
higher-intelligence being in your home.''
Staff photo by John Ewing
PHOTO Caption: Carl Spirito, an associate professor of
physiology at the University of New England in
Biddeford, Maine, is interested in how robots can be
built to simulate bugs' motions.
Click HERE to return to the Resume
Page
|