SPRING
2002 • VOLUME 6, NUMBER 3
Computer
Research Points Way to Virtual Environments
BY
DAVID LaCHANCE
| |
 |
| |
BEN
BARNHART
|
| |
(L
to R) Tasha Sakaguchi '04; Claude Fennema, chair of the College's
computer science department; Huma Jafry '03; Farial Anam '03;
and Larissa Winey '04 are figuring out how to connect a treadmill
to a computer that will project an artificial version of the
Clapp Lab hallway, seen here, onto a screen mounted in front
of the treadmill. |
Aboard
the starship Enterprise is a place called the holodeck, a fantastic
space where powerful computers can simulate almost any environment
with a degree of fidelity indistinguishable from reality. Here,
the ship’s crew can stroll the streets of Victorian London, for
instance, or ride the trails of the Old West. This is the stuff
of science fiction. But if this tantalizing bit of fantasy should
become reality decades from now, it may be in some small part
thanks to the work of Claude L. Fennema Jr., chair of Mount Holyoke’s
computer science department, and his team of student researchers.
Working
with student researchers at three other institutions—fellow computer
scientists at the University of Utah and perception scientists
at the University of Minnesota and Vanderbilt University— the
Mount Holyoke students are delving into the mysteries of the human/machine
interface, the ways in which a computer can create a full-scale
environment that human senses regard as real. The schools are
working together under a five-year, $4-million National Science
Foundation (NSF) grant for the study of so-called “locomotion
interfaces,” the technology that, as Fennema, says, “makes an
artificial environment seem real and makes it feel like you’re
really there.”
Virtual
reality has been with us for some time, in the form of computer-generated
special effects and flight simulators, to name a few of its uses.
Yet technology that would allow a person to walk through a full-size
physical environment remains beyond reach. The benefits of such
technology would not be limited to interstellar recreation—children
could visit Rome’s Colosseum, architects could walk through a
building still under design, and firefighters could be trained
to contain fires without being put at risk.
Designing
equipment to “create” a virtual environment, as on a computer
screen or a projection, has been the focus of much research, and
its fruits can be seen in the Sony PlayStation and the movie The
Lord of the Rings. The locomotion interface research takes that
work in a new direction, looking for the first time at the interactions
of visual and perceptual cues in the virtual world. In other words,
what sorts of things do our senses need to perceive for the world
to seem real? Which explains the existence of the treadmill in
a computer laboratory in Clapp Lab.
One
question that keenly interests the students in Fennema’s Information
Technology Research (ITR) group— Farial Anam ’03, Huma Jafry ’03,
Nausheen Malik ’03, Natasha Mohanty ’03, Tasha Sakaguchi ’04,
and Larissa Winey ’04—is the relationship between the motion of
walking and what we see—what Fennema calls “optical flow.” What
“feels” right at different paces, and different strides? What
happens if a person walking at a certain gait sees the scenery
passing by too quickly, or too slowly?
The
students are figuring out how to connect the treadmill to a computer
that will project an artificial environment—in this case, the
fourth-floor hallway of Clapp—onto a screen mounted in front of
the treadmill. Once the device is working, people will be able
to “walk” down the virtual hallway.
Even
this rudimentary interface raises many technical questions. During
a weekly meeting of the ITR group, Fennema presses students to
move past their urges to perfect the device in the design stage,
to “get it together and get it working and then see where the
shortcomings are.”
What
about textures? the students ask. Lighting? Shadows? Levels of
detail? How can the hallway be scaled to make it appear to be
the right size? Is there a way to get around the time the computer
needs to render its scenes, making motion appear jerky? These,
Fennema said, are next steps to be addressed once the system is
running. That is the goal for this spring.
Students
do more than tinker with software and meet weekly with Fennema.
Through the ITR project, they attend conferences, work with researchers
in other fields and at other institutions, and prepare an honors
project—all rare opportunities for undergraduates in research.
Participating students can work their entire junior year on the
project, spending ten hours a week in independent study, and getting
paid for an additional ten hours under the NSF grant.
Students
can also spend part of their summers at Utah, working on the Treadport,
a state-of-the-art locomotion interface with an eight-by-ten foot
treadmill developed at that school by William Thompson, the lead
investigator under the NSF grant. All of these opportunities are
“chances to get them involved in a really big way,” Fennema says.
“Working
on a project of this scale is very different from working in an
independent study,” says Mohanty. “It involves not only taking
care of my part of the project, but learning to work with other
people and learning from others. I want to go to graduate school
in the future, and this project is giving me a flavor of the kind
of work I will be doing.”
For
Anam, research held little interest until she learned of the ITR
project. “I realized that by not even giving research a chance
I would be denying myself a possibly great opportunity and career.”
That she may be helping to bring about the kinds of virtual reality
envisioned in Star Trek was an added inducement.
For
Fennema, ITR represents the latest chapter in a forty-year computing
career. In the 1960s, he was among the scientists who cobbled
together “Shakey,” a computer-controlled robot trumpeted by Life
magazine as “the first electronic person.” He helped develop the
technology that General Motors, Ford, and Caterpillar used to
create industrial robots, and later went to the 3M Company to
help turn research discoveries into marketable products, learning
how the marketplace helps focus research to have an impact on
society.
He
began to consider teaching as a way to have a greater impact and
came to the College in 1990, helping to develop the computer science
major. With three daughters—one a teacher of math and computer
science, a second who has a doctorate in cognitive science, and
a third with a degree in English and journalism—Fennema was particularly
enthused about having a role in encouraging women in the sciences.
Fennema
is proud of the track record of the department’s graduates: of
the sixty-four who graduated between 1992 and 2000, eleven entered
doctoral programs, and eight entered master’s programs. Two—Margarita
D. Bratkova ’01 and Sarah T. Erickson ’01—are now pursuing their
doctorates in computer science at Utah, says Thompson, the director
of the program there, “and we hope to have more.”
Fennema’s
experience tells him to teach his students to take a long view,
to ignore the trends, and to focus on the fundamental questions.
“I want them to think about a twenty-year cycle, to think far
enough out to create a career of interest,” he says.
“These
students will become technical leaders, able to persuade people
that this is the direction to go in,” he adds. “Remember—TV took
twenty-six years to make it to market.”
And
don’t be surprised if someday you find yourself waiting in line
for those holodeck tickets.
Can
a Robot Help You Get Into Graduate School?
If the Robot Is Susan B., Then The Answer is Yes
| |
 |
| |
BEN
BARNHART
|
| |
Jessica
Littman '02 (left) and Suchi Saria '04 with robot Susan
B. |
Susan
B. is a two-foot-tall, cylindrical robot created at Mount
Holyoke in 1990 as a platform for research into artificial
intelligence. Equipped with a television camera and connected
to a network of computers, Susan B. (named for Susan B.
Anthony) has helped Mount Holyoke students cut their research
teeth in such areas as computer vision, natural languages,
computer networking, electronics, and parallel processing.
For several, this research experience has been a springboard
to graduate schools, including the Massachusetts Institute
of Technology, Carnegie Mellon, and Stanford. Fifteen honors
projects and twenty-two independent projects have been carried
out under Susan B.’s umbrella. Lately, Susan B. has been
spending time with Jessica Littman ’02 and Suchi Saria ’04.
Littman is doing an honors project aimed at helping the
computer teach itself to figure out the implications of
commands given in common English. Through a kind of artificial
reasoning, Fennema says, the computer can learn how to do
what it’s told to do. Even a seemingly simple question like
“Where are you?” can require complex reasoning on the robot’s
part. “She may have to look at colors, decide what patterns
mean, and compare what she sees with her memories of the
environment. She may find that she has to move,” Fennema
says. Saria is developing a navigation system that will
allow the robot to navigate intelligently, using information
that it gathers along the way to plot its path. Built on
Susan B.’s software platform, the new navigation system
will be at the heart of “Susan C.,” the new robot Saria
is working to build with Fennema and Mount Holyoke computer
science professor Paul Dobosh.
|
|
