Every year, drivers on Los Angeles freeways spend a combined total of 9,000 years stuck in traffic. That mind-boggling amount is in addition to the time it takes to get from one place to another tooling along on the highway at the legal speed limit. Fortunately, research in UC Berkeley's Center for Information Technology Research in the Interest of Society (CITRIS) program could lead to a kinder, gentler commute.
The plan, according to principal investigator Pravin Varaiya,
is to tap California's pre-existing freeway sensor network for
data to intelligently deal with congestion. Already, using an
alpha version of a traveler information system, commuters in Los
Angeles are able to log on to the PeMS (Performance Measurement
System) Web site, click on two map points, and instantly be told
travel time at that precise moment or, based on historical precedence,
anytime in the future. Eventually, the same system could be accessible
via Internet-enabled cell phones. Next July, once the desired
features are finalized and the California Department of Transportation (Caltrans) staff are trained, the agency
will deploy PeMS statewide for use by the public as well as Caltrans.
And that's just the first stop on the PeMS path.
The real-time freeway figures stream into Caltrans from loop detectors, the hexagon-shaped wire sensors embedded in the highway every third of a mile. Numbering more than 5,000 in Los Angeles alone, the loop detectors count how many cars cross the loop and the average time a car is on top of the loop every 30 seconds. The result is gigabytes of impenetrable numbers in dire need of translation.
"Managers and engineers at Caltrans don't really know what's happening out there, right
now, last month or last year, for that matter," says Varaiya,
the Nortel Networks Distinguished Professor of Electrical Engineering
and Computer Sciences. "There was a lot of data being collected
but they just stored it. Nobody ever looked at it. It just sat
on tapes gathering dust."
Acting as an automated interpreter, the software algorithms
convert the raw loop data into "news you can use." For example,
the travel time algorithm employs a statistical model called "regression,"
a term borrowed from the phrase "regressing towards the average."
Essentially, the PeMS system is programmed to forecast travel
time partially based on the assumption that if current congestion
is especially bad compared to historical data, it's likely to
improve and vice versa.
Still, the PeMS Web site comes complete with a caveat: "No matter
how much data or computing power, we cannot look into the future
and know that in fifteen minutes from now Mr. Smith's tire will
blow on I-10 causing him to block the right lane."
For more detailed analyses, PeMS generates three-dimensional
graphical representations of the loop data. Then, historical data
can be mapped on top of real-time read-outs to determine how,
for instance, traffic patterns have changed. With the computerized
assessments available online, traffic engineers might alter operational
decisions regarding metering lights or highway closures due to
construction. On a larger scale, "planners can determine whether
congestion bottlenecks can be alleviated by improving operations
or by minor capital improvements," according to a paper Varaiya
and two colleagues prepared for the national Transportation Research
Board meeting next year.
Gerald
Stone/PATH photo |
"Now you can do an analysis in a day that would have taken two months to do by hand, which of course nobody would have done," he says.
The goal is that once engineers are better informed about the
true state of congestion at any given moment, equipment like ramp-metering
lights and scrolling message systems can be made more efficient.
For instance, current metering lights are preset to turn on at
certain times of the day rather than at the times they're needed
most. Automated metering, Varaiya says, could reduce freeway traffic
by two-thirds.
Further improvements will come from strategically-placed electronic message
boards displaying constantly-updated travel tips such as: "You're
going to spend six minutes on this ramp, you should proceed to
the next exit instead." The effectiveness of a similar signage
system has already been demonstrated in Paris, Varaiya says.
"It's a non-linear improvement curve," he explains. "If you
can divert between two to five percent of the traffic, the benefit
is dramatic."
The researchers expect the next research phase to be driven by the recognition that freeways are just one element in overall transportation systems that could entirely be integrated using PeMS. For example, public transportation could be increased on an as-needed basis so that if one mode is stressed, perhaps more buses or trains to a certain destination could be introduced to pick up the overflow. Meanwhile, stoplights on arterial streets could also be coordinated with freeway conditions. Even under-utilized high-occupancy vehicle lanes on freeways could be eliminated with real-time ramp metering. Carpools would still be encouraged with a bypass at the on-ramps, Varaiya says, enhancing the environmental benefits PeMS is already expected to provide by reducing travel times, hence lowering auto pollution.
Long-term, Varaiya envisions PeMS digesting real-time data from the electronic
toll-paying tags slowly catching on in California and interfacing
with devices deployed by the Partners for Advanced Transit and
Authority Program (PATH) administered by UC Berkeley's Institute
for Transportation Studies in collaboration with Caltrans. Current
PATH endeavors include the test of a freeway surveillance system
that, using computer vision technology in development at the College
of Engineering, automatically detects stalled cars and accidents.
On the PATH horizon are intelligent highways, where robotic vehicles
keep traffic flowing by driving themselves. An intelligent highway
test for heavy trucks and buses is scheduled for 2003 near San
Diego.
"Economists point out that all congestion is pure dead-weight
loss," Varaiya says. "We may have learned how to entertain ourselves
while we're stuck but transportation doesn't have to be so bad.
We have lots of ideas on how to improve it a lot and Caltrans
is a receptive audience."
PeMS Freeway Performance Measurement Project: transacct.eecs.berkeley.edu
Pravin Varaiya's home page: www.path.berkeley.edu/~varaiya
PATH: paleale.eecs.berkeley.edu
Caltrans: www.dot.ca.gov
