 |
|
September 29, 2006 Vol. 77,
no. 7F
 |
| UP
TO THE CHALLENGE:EECS postdoc Jonathan Sprinkle leads the Berkeley DARPA team.
JONATHAN SPRINKLE
PHOTO
|
Driverless
cars to race through the street
Berkeley sets its sights on the 2007 DARPA “Urban” Grand
Challenge
Berkeley is joining forces with the University of Sydney to build
an autonomous robotic vehicle to enter in next year’s DARPA Grand
Challenge. EECS postdoctoral researcher Jonathan Sprinkle has mobilized
25 faculty, postdocs, graduate and undergraduate students from three
different UC Berkeley departments to help design and develop a vehicle
for the November 3, 2007, competition.
The new Grand Challenge, sponsored by the Defense Advanced Research
Projects Agency (DARPA), challenges teams from around the world to
complete a 60-mile course in a mock urban area in less than six hours.
Autonomous vehicles must navigate their way through streets while negotiating
traffic, obeying traffic laws and avoiding obstacles. Dubbed the DARPA
Urban Challenge, the competition seeks to mimic military supply missions
in urban zones. The new competition builds on the last Grand Challenge,
where autonomous vehicles navigated rough desert terrain.
The enormity of the challenge isn’t lost on Sprinkle, but he
is taking full advantage of his role as executive director of CHESS,
the Center for Hybrid and Embedded Software Systems, enlisting the
help of faculty and researchers with specialties in vehicle automation,
control and robotics. “Berkeley will be concentrating on the”intelligent
control side, while the Australian Centre for Field Robotics will concentrate
on sensing systems and developing the autonomous machine,” Sprinkle
explains.
The harmonious marriage of hardware and software will be even more critical,
given the competition’s constraints. “You can’t have
the car suddenly go into reverse because of a software glitch, so you
need to have a great degree of confidence in your system,” he says.
This semester the Berkeley team is concentrating on building and testing
a software control framework that will optimize the correct path. “We’re
not even worrying about lanes and oncoming traffic at this point.”
To manage the project, Sprinkle is formalizing many of the team’s
processes, from asking members to meticulously document their work to
implementing checklists and setting up a wiki as a communication tool.
Sprinkle is also teaching a course this semester, EECS 290p, “Autonomous
Systems: Algorithms and Implementation,” that focuses on the DARPA
project.
In addition, he’s tapping the expertise of past Grand Challenge
entrants, namely IEOR Ph.D. student Anthony Levandowski (B.S.’02,
M.S.’03 IEOR), who led Berkeley’s quest for the Challenge
in 2005. Levandowski’s team entered an autonomous motorcycle, which
came close to competing in the big race, finishing two miles of a 2.3-mile
final qualifying round. Since next year’s Urban Challenge only
allows four-wheel vehicles, Berkeley will work with Sydney to create
a new vehicle design. But Sprinkle wants to reproduce the 2005 team’s
passion for their project. “They had a tight cadre, and that’s
essential,” he says. Sprinkle is recruiting students with keen
programming skills, a love of cars or a simple desire to take on this
Grand Challenge.
For more information, go to http://chess.eecs.berkeley.edu/dgc3.
|
|
