By Pierre Home-Douglas
TACKLING THE MYSTERY OF
SAND DUNES, A 175-MILE ROBOTIC RACE, AND EXTRAPLANATORY
ROVER DESIGN, UNDERGRADUATES AT CALTECH ARE SURFING
THEIR WAY TO INNOVATIVE RESEARCH.
It sounds like every undergraduate engineering student's
dream: doing groundbreaking research on a robotic vehicle
capable of completing a snaky 175-mile course through
the desert. An added bonus: helping your school pick
up a check for $2 million as winner of the Grand Challenge
race sponsored by the Defense Advanced Research Projects
Agency (DARPA). For 20-year-old computer science student
Jeremy Gillula, this opportunity was made possible by
a unique program at the California Institute of Technology
in Pasadena that pairs science and engineering students
with mentors in Caltech's engineering department
and at NASA's nearby Jet Propulsion Laboratory.
As part of Caltech's Summer Undergraduate Research
Fund (SURF), students are expected to find a mentor
with whom they can work and then define and develop
a project. Their research projects are reviewed by a
faculty committee, which then recommends which students
will be awarded the fellowship research and a stipend
of $5,000. At the end of the 10-week-long program, students
submit a technical paper then deliver an oral presentation
at a symposium modeled on a professional technical meeting.
Gillula had already worked the previous summer and
throughout the school year on a computer code that would
speed up Caltech's entry in the DARPA Grand Challenge.
The Challenge is a major-league field test intended
to accelerate research and development in autonomous
ground vehicles that will help save American lives on
the battlefield. Vehicles participating in the
2005 race, which will run this October, can only use
publicly available navigation systems like GPS and cannot
be controlled remotely by an operator. Caltech's
entry, a souped-up Ford van affectionately known as
Alice, will have a pair of mounted cameras to triangulate
the position of objects in front of it. Gillula's
work involved making the stereo-vision code faster and
more accurate to enable the vehicle to complete the
obstacle-filled course in under the 10-hour maximum
limit. In the 2004 race, none of the 15 vehicles that
entered the race from Barstow, Calif., to Primm, Nev.,
made it farther than 8 miles from the starting line.
"I knew from the beginning of last summer what
needed to be done," Gillula explains, "I
knew what the flaws were in the code. I knew how to
fix them. I just needed a good 10 weeks. SURF gave me
the chance to do it."
Gillula
submitted his proposal to mechanical engineering professor
Richard Murray, a cosponsor of Caltech's entry
who has worked with more than 30 SURF students over
the past 12 years. Murray says that SURF is "all
about teaching undergraduate students what research
is like and learning along with them about new ideas.
It's much closer to the type of teaching you do
with graduate students as they're working on their
theses." He adds, "Of course, the level
is different with graduate students, but Caltech undergraduates
are incredibly bright. They are constantly surprising
me with new insights and approaches."
Murray met each week with Gillula and four other
SURF students involved in the Grand Challenge to discuss
overall activities for the team. He checked in with
Gillula several times over the 10 weeks to get updates
on his progress. The two then spent five days in the
desert with the rest of the Caltech team testing out
the vehicle.
For Murray, who also serves as chair of the Division
of Engineering and Applied Sciences, Gillula's
SURF project is a good example of a student tackling
a problem that needed solving but had eluded other researchers.
"He's done a great job of investigating
what the possibilities are, trying them out, and keeping
careful data to see which ones really worked. His efforts
over the summer are definitely going to make a big difference
in our algorithms for the next race." And for
Gillula, SURF is "an incredible hands-on experience.
In the classroom you learn the theory. You might get
a guest lecture who comes in and says, ‘Here's
what we do in industry,' but when you get to do
it yourself, you learn so much more—the ins and
outs of every detail of the system."
SURF grew out of a grant from Hollywood mogul Lew
Wasserman, head of Universal Pictures and a member of
Caltech's board of trustees. In the mid 1970s,
flush with the success of the film Jaws, Wasserman decided
to endow a grant that would make a significant contribution
to Caltech, a school that consistently ranks among the
top engineering universities in the United States. A
committee was formed to decide how the money should
best be used, and eventually an undergraduate summer
research program was designed that would allow students
to pursue hands-on work under the guidance of professors
and professionals.
Under the leadership of chemical engineering professor
Fred Sher, SURF started in 1979 with 17 faculty members
and 18 students. At the start, there were some professors
who felt that undergraduates were more suited to classroom
learning than serious research, but the quality of the
students drawn to SURF and the breadth of their work
turned cynics into believers. Any remaining doubts were
finally allayed in 1995 with the launch of SURFSAT,
a $3 million communications satellite created by a series
of SURF students over the previous seven years. The
probe was designed to test new ways to receive communications
from distant planetary probes by way of NASA's
Deep Space Network. But it was later adapted to help
debug communications between the VLBI, a series of radio
telescopes around the world that can be linked to operate
as a single telescope, probing celestial mysteries like
quasars and black holes.
JPL manager Joel Smith described this added use of
SURFSAT as alone "worth its weight it gold."
In the fall of 2004, nine years after launch, SURFSAT
was still in operation, "a hugely and wildly successful
run," says Carolyn Ash, director of student-faculty
programs at Caltech. She says that SURF has helped change
attitudes about undergraduate research at Caltech. "It's
now deeply embedded in the culture here." In the
summer of 2004, close to 400 students participated in
SURF. Most were from Caltech's engineering department,
but the program also attracts students from as far afield
as Singapore, France, and India.
Students hit the ground running with SURF since they
have already collaborated with their mentors and completed
a research proposal before they begin work. "With
a lot of other internships, the student shows up and
everyone says, ‘Gee, we knew you were coming,
but we're not ready yet'. With SURF, everyone
is poised to go before the students arrive, so they
can maximize the time spent in the program."
As is true of research in general, SURF projects
don't always produce nice, neat answers or end
up solving the problem that the student proposed. For
Richard Murray, that's actually one of the advantages
of SURF and why it fills a gap often left by graduate
students. "Graduate students, by their very nature,
don't want to go off on an avenue of research
that may lead them into a dead end." Plus, he
adds, some research work isn't "meaty"
enough to warrant a graduate student's attention.
PUBLISHING PAPERS
Still,
SURF projects are not simply "make-work"
assignments. Some of the work done is undeniably groundbreaking.
Senior electrical engineering student Lyle Chamberlain
spent his 2003 SURF summer developing the control software
that may help enable robotic vehicles traverse difficult
terrain on other planets 10 to 20 years down the line.
Chamberlain's mentor at the Jet Propulsion Laboratory,
Terry Hunstberger, part of JPL's Mechanical and
Robotics Technology Group, says he was impressed by
how quickly Chamberlain came up to speed and made top-quality
research contributions. "Once I got him started,
he took off," Huntsberger says with a chuckle.
His only regret was the length of the SURF project:
"I wish it were longer. I'd like to have
had him for six months." The two wrote a paper
together that was accepted by the International Conference
on Robotics and Automation.
"Paper publishing is definitely one of the
best things about a SURF," Chamberlain enthuses.
"I'm applying to grad school and I have
four papers on my CV. That's very rare for an
undergrad." In the summer of 2004, Chamberlain
completed another SURF, adapting his work from the previous
summer to help design a robotic vehicle with Caltech's
Murray.
The work that one SURF student may perform can sometimes
lead to fruition a year or more later. Mechanical engineering
professor Melany Hunt observed this with a SURF student
who built a probe during the summer of 2003 to take
samples below the surface of sand dunes. The work is
part of Hunt's efforts to unlock a centuries-old
mystery: why certain large sand dunes around the world
emit a deep, booming sound when the wind passes over
them, a phenomenon that has been noted by travelers
as far back as Marco Polo. She believes that the noise
is the result of a damp, reflective layer about 6 feet
down that causes the sound to reverberate back and forth
between the surface and the denser layer of sand below.
Getting samples from that wet level posed a challenge—and
a suitable goal for a SURF project.
Nora DeDontney, then a junior majoring in geophysics
and mechanical engineering, designed a 7-foot-long tube
with another tube inside that served as a retractable
collection basket—a "very clever design,"
according to Hunt. The trouble was, when they tried
out the rod in September, the end of it buckled when
it was hammered down the last few feet. "Nora
was crushed," Hunt recalls. But the experience
was far from a failure: "She learned a lot about
strength of materials and all the other engineering
aspects you have to think about when you build something
like that—lessons you wouldn't necessarily
get in a classroom."
Hunt meets regularly with her SURF students. "If
they're having problems, you have to make sure
you learn about them quickly and help them solve them.
If they don't make progress, there's something
they don't understand and they don't know
where to go next; you have to get them past that hurdle."
Last summer, a SURF student did a redesign of the
sampling rod and it worked flawlessly. "It hasn't
given us a conclusive answer yet," says Hunt,
"but it certainly supports our hypothesis about
what is happening." Hunt says the work on sand
dunes may have application for other geophysical flows
like avalanches and the kinds of materials and fluids
that petroleum companies are concerned with in pumping
or drilling operations. Chalk up another success—for
faculty and students—as part of Caltech's
SURF program.
Pierre Home-Douglas is a freelance writer based in
Montreal.
|