March 2004
Sciencecraft
By Patrick L. Barry and Tony Phillips
Probes that can distinguish between "interesting" things and "boring" things are
vital for deep space exploration, say JPL scientists.
Along with his colleagues in NASA's Space Technology 6 Project (ST6), JPL's
Steven Chien is working to develop an artificial intelligence technology that
does just that. They call it the Autonomous Sciencecraft Experiment, and it's
one of many next-generation satellite technologies emerging from NASA's New
Millennium Program.
As humanity expands its exploration of the outer solar system-or even
neighboring solar systems!-the probes we send suffer from two unavoidable
handicaps. First, commands radioed by mission scientists on Earth take a long
time to reach the probe: six hours for the planned New Horizons mission to
Pluto, for example.
Second, the great distance also means that data beamed back by the probe
trickles to Earth at a lower bandwidth-often much less than an old 28.8 kbps
modem. Waiting for hundreds or thousands of multi-megabyte scientific images to
download could take weeks. And often many of those images will be "boring," that
is, they won't contain anything new or important for scientists to puzzle over.
That's certainly not the most efficient way of using a multi-million dollar
probe.
Even worse, what if one of those images showed something extremely
"interesting"-a rare event like a volcanic eruption or an unexpected feature
like glaciers of methane ice? By the time scientists see the images, hours or
days would have passed, and it may be too late to tell the probe to take a
closer look.
But how can a probe's computer brain possibly decide what's "interesting" to
scientists and what's not?
 |
|
The Autonomous Sciencecraft technology that
will be tested as part of NASA's Space Technology 6 mission will use artificial intelligence to select and transmit only the scientifically significant images. |
"What you really want is a probe that can
identify changes or unique features and focus on those things on its own, rather
than just taking images indiscriminately," says Arthur Chmielewski, one of
Chien's colleagues at JPL.
Indeed, that's what Chien's software does. It looks for things that change. A
mission to Jupiter's icy moon Europa, for instance, might zero in on
newly-formed cracks in the ice. Using artificial intelligence to set priorities,
the probe could capture a complete movie of growing fractures rather than a
single haphazard snapshot.
Until scientists can actually travel to deep space and explore distant worlds in
person, they'll need spacecraft "out there" that can do some of the thinking for
them. Sciencecraft is leading the way.
Learn more about Sciencecraft at
http://nmp.nasa.gov/st6. Kids can make a "Star Finder" for this month and
learn about another of the ST6 technologies at
http://spaceplace.nasa.gov/st6starfinder/st6starfinder.htm.
This article was provided by the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration.