Photography by John Florence

Professor Pitu Mirchandani directs the ATLAS Lab

By Ed Stiles

Take a peek at what’s
in store for drivers
as we turn the corner
into the next millennium. Cars will drive themselves
and electronic traffic cops
will use blocks-long vision to
efficiently direct traffic.

That’s what engineers at the
ATLAS (Advanced Traffic and
Logistics Algorithms and Sys-
tems) Laboratory at the University
of Arizona in Tucson envision as
they create traffic systems for the
21st century.

Controlling traffic amounts to controlling space, explains Pitu Mirchandani, director of ATLAS and a professor of Systems and Industrial Engineering (SIE).

Take a freeway interchange, for instance, where an arterial street crosses the freeway. Cars going across the freeway and those entering and exiting all have to get through this interchange. At any given time only a few cars can occupy the space at critical crossing points. Traffic control amounts to deciding who gets to use that space and when – the goal being to move the most cars as quickly as possible.

During the next few months, Mirchandani, SIE Assistant Professor Larry Head and other ATLAS researchers will test a system that behaves something like an omniscient traffic cop to control such interchanges. Tests will be run at interchanges in Tempe, Ariz.; Tucson; Seattle, Wash., and possibly Santa Clara, Calif.

Called RHODES (Real-time Hierarchical Optimized Distributed Effective System), it employs video cameras at the interchange, radar detectors near the interchange and loop detectors in the pavement up to several blocks away. These all gather data on traffic volume and speed.

This data is fed to a computer (the omniscient traffic cop) that controls the signals at the interchange. Since the data is updated every couple of seconds or so, the computer "knows" what’s happening to traffic right now – in "real time," as engineers say. It then decides how to time the lights to optimize the traffic flow.

Currently, researchers are giving the computer its initial instructions based on traffic data they have gathered at an interchange. Doing this for every interchange on a freeway system would take thousands of hours. "Eventually, we want to automate this, so that we just install the equipment and within a matter of a few hours it has learned enough about how traffic flows through the intersection to set itself up," Mirchandani says. That way, traffic departments could simply install the generic traffic control system without first doing a detailed study of traffic flow.

"This system will decrease travel time anywhere from 20 to 50 percent at the interchange," Mirchandani says. "If the intersection can handle 3,000 cars an hour now, it might be able to handle 4,000 with this system. You have effectively increased the capacity of the interchange without the disruption and cost of a construction project to upgrade the intersection."

Mirchandani emphasized that he’s not building new technology. The electronics, signals, computers and other hardware needed to do this exist today. ATLAS is developing the algorithm and an integrated system that control the operation. Baking a cake is a good analogy for this, he adds. The ingredients for the cake are like the hardware. The recipe is the algorithm. And the properly mixed and baked cake is the integrated system.

In another project, Mirchandani, SIE Assistant Professor Frank Ciarallo and other ATLAS researchers are developing ways to control freeway access during times of high traffic congestion. Some cities already have traffic signals (referred to as ramp meters by traffic engineers) that control the rate at which vehicles enter the freeway from on-ramps. But MILOS (Multi-Objective Integrated Large-scale Optimized ramp-metering control System) will do this in real time.

"What we do is monitor large segments of a freeway and gather traffic data on the mainline as well as at all the on- and off-ramps," he says. "Then traffic moving onto the freeway is regulated by the ramp meters located on the on-ramps." MILOS gathers data at about 20-second intervals and updates the timing of the ramp meters about once a minute.

ATLAS researchers will test MILOS in about six months along a 10-mile-long section of a freeway in Phoenix.

The ATLAS center also has been working on an intelligent vehicle. Recently Mirchandani and SIE Associate Professor Feiyue Wang built a car that drives itself. It was tested on an unopened section of a freeway in Phoenix and successfully followed another car while completely under control of an on-board computer.

The car uses radar and a video camera to sense other cars, maintain its position in the lane and to control speed and braking. "We visualize a future where the driver of an intelligent vehicle would have a CD of a map charting his route, say
I-10 from Tucson to Phoenix," Mirchandani says. "This would be a very detailed GPS map. Your car will follow that map and get updated based on roadside reflectors." Such reflectors would be located about every few hundred meters along the freeway. As the car’s radar detected each of these, the car would automatically recalibrate itself with the map using the GPS data it was receiving from satellites.

Mirchandani believes the trucking industry will be the first to use intelligent vehicles. The driver in the lead truck would do the driving and the following intelligent trucks would drive themselves. "If the driver in the lead truck gets tired, another truck can take over at the front of the line and the first driver can just fall in line and sleep while his truck does the driving," Mirchandani says. 

Related link:
http://www.sie.arizona.edu/ATLAS