Pedal to the metal

Makers of robot cars push technology's edge in DARPA's Urban Challenge

What is an autonomous vehicle?

Basically, a modified car or sport utility vehicle
that carries a variety of sensor and computing
equipment including long-range radar; shortrange
lidar, mid-range lidar and long-range lidar
(lidar works like radar but uses lasers); a
Global Positioning System and behavioral
software to make tactical decisions;
and an Ethernet communications

LAS VEGAS - With a forward jolt, the Chevy
Tahoe begins navigating a course marked
with bright orange cones.

The starts and stops are jerky, as though a
student driver were using two feet on the
accelerator and brake simultaneously. The
turns are deliberate and sharp, with little concern
for the comfort of backseat passengers.
At a four-way stop, an encounter with a
hefty Hummer H2 is treated cautiously; the
Tahoe waits patiently to make sure the other
car has indeed stopped.

For a novice driver, the exercise is passable.
For a driverless car depending 100 percent on
technology, the trip is impressive.

The vehicle ? Boss ? was built by Tartan
Racing, part of Carnegie Mellon University's
Robotics Institute. The vehicle won the 2007
Urban Challenge, held by the Defense
Advanced Research Projects Agency, and its
$2 million top prize. November's race was the
third one DARPA sponsored.

Stanford Racing's Junior, from Stanford
University, took the second-place, $1 million
prize, and Victor Tango's Odin, from Virginia
Tech, received $500,000 for finishing third.
The DARPA competitions are designed to
foster development of autonomous robotic
ground vehicle technology for the battlefield.
As these technologies improve, their application
will provide business opportunities
beyond the battlefield. Boss was recently
demonstrated in Las Vegas as part of the
Consumer Electronics Show.

Vehicles competing in the Urban Challenge
were required to operate entirely autonomously,
without human intervention, as they
obeyed California traffic laws and performed
maneuvers such as merging into moving traffic,
navigating traffic circles and avoiding

The vehicles had to think like human drivers
and continually make split-second decisions
to merge into traffic, safely pass
through intersections and avoid colliding
with other vehicles.


The first Grand Challenge was held in 2004
on a 142-mile desert course between vehicles attempted the course, but none finished
and the $1 million cash prize went

The most recent race near Victorville,
Calif., was held in a simulated urban setting
and was the most difficult to date, said Chris
Urmson, director of technology at Tartan

"The idea was to have the vehicles interact
with one another, and with human-driven
traffic, much the same way drivers do when
they commute to work in the morning," he
said. "The vehicles had to be able to see
where they were in the world and to see other

DARPA officials gave teams maps of the
roads they could use ahead of time. The winning
vehicle had to complete the course in
the least amount of time in a safe manner,
while following the rules of the road.
Since work began in 2003 for the first
challenge, the technology used for perceiving
and tracking vehicles has greatly evolved,
Urmson said. Improved computing power is
also a critical factor in making the technology

All the teams relied on computers, laser
rangefinders, the Global Positioning System
and inertial measurement to navigate the

Team Victor Tango is composed of Virginia
Tech undergraduates, graduate students, faculty
and Torc Technologies, a Virginia Tech
spinoff company that works with autonomous
systems. The Odin vehicle is a converted
2005 Ford hybrid Escape.

The Stanford team, which won the 2005
Grand Challenge, used a converted 2006
Volkswagen Passat for the Urban Challenge.
The ability to avoid collisions and understand
human driving concepts, such as right
of way, is of great interest to Pentagon officials
who want autonomous vehicles for dangerous
missions, such as delivering supplies.

"This has a component of prediction," said
Mike Montemerlo, a senior research engineer
in the Stanford Artificial Intelligence Lab.
"There are other intelligent robot drivers out
in the world. Predicting what they are going
to do in the future is a hard problem that is
important to driving. Is it my turn at the
intersection? Do I have time to get across the
intersection before somebody hits me?"

All that prediction and evaluation gets
more complex in dense urban congestion.

"If you're in a lane and you're going to stay
in that lane, that's easy enough to do,"
Urmson said. "But when people are driving in
very dense traffic, there's a lot of social activity
that happens. You make eye contact with
somebody that's going to change lanes, you
might gesture in some way that you're either
letting them merge in, or that you want to go
change lanes. That kind of social interaction
is hard to do."

Understanding traffic lights is difficult
because the configuration of one intersection
to the next is usually different.

Predicting what another vehicle might do
is difficult enough, predicting what pedestrians
might do is even harder, Urmson said.

"With vehicles, you have that constraint
that they can only go the way the wheels are
pointed, they can't leap sideways," he said.
"If instead you had Emmitt Smith out in the
road, he can go anywhere. So modeling the
way pedestrians move is difficult."


Although it will likely be at least a decade
before fully autonomous vehicles are available,
General Motors' Larry Burns said the
company is committed to investing in the
research and development needed for the

"OnStar, for example, is 12 years old, and
we're on our eighth-generation hardware,"
said Burns, GM's vice president of research
and development. "The only way to position
yourself for leadership in technology in the
auto industry is to get out there and create
those opportunities and get them to a generation
one, level-of-proof concept and then see
what really plays out."

Business opportunities to develop new sensors
and computer systems to interpret them
are expected to increase. Improved laser and
radar are needed for the technology to
become mainstream.

"The car is going to become much more of
an information appliance," Urmson said.
"You're actually going to have sensors on the
vehicle that allow you to push information
back into the network, and that's a big deal.
So we can start to look at how you use that
information and what applications that
inspires in the future."

Doug Beizer ( is a staff
writer at Washington Technology.

About the Author

Doug Beizer is a staff writer for Washington Technology.

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

What is your e-mail address?

My e-mail address is:

Do you have a password?

Forgot your password? Click here


contracts DB

Washington Technology Daily

Sign up for our newsletter.

Terms and Privacy Policy consent

I agree to this site's Privacy Policy.