Sign of the Times: New Faculty
Robot Wins 3rd Place and Creativity Award
Use Robots in Egyptian Desert
Alums Return for Basketball Game
In the Dakhleh Oasis of Egypts Western Desert, Roelof Versteeg
pushed a jury-rigged cart made of PVC piping equipped with two cesium
gradiometers and a differential global positioning system receiver
to scan the remnants of an ancient town lying beneath the sand.
Versteeg, a geophysicist in the department of earth and environmental
engineering, is part of an interdisciplinary team of eight computer
scientists, archeologists, and art historians who are using 21st
century technology to discover the location of structures that once
were the Roman town of Trimithis. The incongruous juxtaposition
of robots and archeological sites is the result of an imaginative
proposal led by Peter Allen, professor of computer science, who
formed this interdisciplinary team that was awarded a $2 million
National Science Foundation grant. This grant brings together
a range of technology on historical reconstruction and applies it
to an area of research that is manual, slow and labor intensive,
said Allen. We have combined several areas of research that
will apply new digital technologies to the problem so we can create
tools that will model, visualize and analyze historic and ancient
sites more accurately, efficiently and with a minimum of invasiveness.
The five-year NSF project takes advantage of Columbia Universitys
excavations at Amheida, Egypt, which began in 2000. According to
archeologist Lynn Meskell, associate professor of anthropology,
the site comprises a substantial urban center with standing remains
dating to the Roman period plus mud brick tombs to the south that
date to Ptolemaic and Roman times.
Steve Feiner, Ken Ross and Peter Allen with Allen's robot.
Four members of the Engineering School faculty are involved in
the project: Allen; Steven Feiner (professor of computer science);
Kenneth Ross (associate professor of computer science); and Versteeg.
Allen and Versteeg will provide the surface and subsurface mapping
of the site to create a high quality 3D model of the site. Feiner
will create a wearable, augmented reality system for presenting
this information to mobile users, while Ross will take data accumulated
at the excavation site and make it accessible to everyone who might
be interested in archeology.
Allen will use his mobile site-modeling robot to create a photo-realistic,
geometrically accurate 3D model of the outdoor sites. The robot
will use range and imaging sensors, including a color camera, a
360-degree Omnicam, the omnidirectional camera developed by Prof.
Shree Nayar, and a Cyrax laser range scanner. The remote controlled
robot may also be equipped with Versteegs tools and used again
as the vehicle for scanning the subsurface.
This would be the best method to scan such a large area,
said Versteeg. Last month I walked over 100 kilometers to
collect information from several million data points. The area is
2 square kilometers and we need to have centimeter accuracy. With
robotic sensors it is simple to take data. Once initial surface
and subsurface data have been obtained, Feiners augmented
reality eyewear will overlay the 3D world with important information
for archeologists, including 3D models, and 2D images and text.
As we accumulate data about the artifacts that are discovered,
it will be possible to use our head-worn displays to view relevant
portions of this information as we look around the site, registered
with the things that we see in the real world. For example, when
looking at a piece of pottery, a researcher might be able to call
up for comparison images, models, and descriptions of related items
from different parts of the site or from other sites.
Feiner is also interested in how researchers can take advantage
of a site model that changes over time as work at the site progresses.
Imagine being able to turn the time dial, he said, and
see where you came from. You can go from viewing the area prior
to excavation, to seeing it in its current state, superimposing
subsurface data to visualize which places might be most promising
to explore next.
With large amounts of information coming out of the desert site,
organizing the information is a task that falls to Ken Ross. His
role in the project is to create a database system that allows archeologists
to perform interactive analyses. We will create a compositional
query language that will allow archeologists to build sophisticated
queries from simple parts so they can access the information they
need without having to be computer scientists.
Rosss longer-term goal is to apply the distilled wisdom obtained
from this project to develop a query system infrastructure. Such
an infrastructure would allow the rapid deployment of interactive
query systems in other domains (such as human anatomy) given just
the domain-dependent information.
Of the SEAS faculty, only Versteeg thus far has made the trek to
Amheida, which is a 13-hour drive from Cairo. Before taking their
respective research work on location, Allen and Feiner will refine
their tools on a site that is closer to home. Before we go
to Egypt, we will use the Cathedral of St. John the Divine as a
test bed to refine our tools, said Allen. We will be
using our scanning lasers and recovering interior and underground
data. We will start in April with a new scanner and begin in the
choir area, he said. This work is being done in close collaboration
with Prof. Stephen Murray of the Department of Art History and Archaeology,
who is an expert on St. John the Divine.
For Feiner, one advantage to using St. John the Divine is the treasure
trove of existing documents. There was so much political intrigue,
said Feiner, with one architect replacing another, and many
changes in plans. We have documentation for the design contest organized
for the cathedral, the revisions to the winning entry, the cathedrals
later redesign, and its convoluted construction history. For example,
we are interested in using augmented reality to make it possible
to look up at the Guastavino tile dome originally built as a temporary
roof over the crossing, and instead see the tower that was intended
to be there. The fascinating thing about this technology is that
it can allow you to experience different alternatives in the context
of the actual cathedral.
Allen has effectively synthesized these strains of independent
and divergent research within the University and is credited by
his colleagues in the School with putting it all together and making
it work. I didnt believe it would happen, but it did,
said Versteeg enthusiastically. Ross was equally vocal on the effectiveness
of the team. I am really excited about working on this team,
he said, We have been working together for about six months
and are very cohesive. Ive been on several teams before and
I am very pleased to serve on this team.