...related to February 6, 2005 Boston Globe article describing traffic foul-ups that brought traffic to a halt in sections of Boston on Tuesday, February 1, 2005.

The South Bay merge or an errant pothole crew were not to blame for Tuesday's debacle. It was the lack of information and advance warning.

According to traffic-management researcher Moshe E. Ben-Akiva, an MIT professor: "My general impression . . . is that I don't think there is necessarily one agency that should be blamed. What should be blamed is a lack of focus on coordinated and well-funded, intelligent traffic systems."

That is what Ben-Akiva researches -- how the vagaries of traffic can instantly change and how, when a road is nearing capacity, to warn drivers to take another route before it's too late.

At the center of his research is melding DynaMIT, a real-time traffic management system that combines speed and traffic data generated by video cameras and roadway sensors, with sophisticated simulators that help predict traffic congestion in an area up to an hour into the future.

And the core of DynaMIT's ability to predict future snarls is the traffic simulation, one of the first in the world to view traffic as a strange, unpredictable flow.

DynaMIT can forecast what is going to happen on a particular road. Researchers say they can even schedule a Patriots game into the simulator's equations to better predict backups. In a perfect world, variable message signs and in-car navigation systems would then issue warnings based on the predictions.

Ben-Akiva said much of this technology is already being used at the operation control centers of the Boston Transportation Department, the Turnpike Authority, and the Highway Department.

By coordinating these centers, which currently are not linked, and allowing all of them to communicate in real time, Ben-Akiva believes congestion like Tuesday's could be lessened, even if it means telling people to stay at work late.

"It's just a question of priority in the same way that after the Blizzard of 1978 more resources were allocated for snow removal and emergencies," he said. "Hopefully events like this will create more awareness and a need for dynamic traffic management."

Program ponders future of traffic congestion, roads

Inside a quiet Massachusetts Institute of Technology lab in Kendall Square, researchers are thinking hard about traffic jams, but not the ones they may face going home.

They're studying the congestion drivers could face years from now, and how to guide motorists -- some not yet born -- around the standstills of the future.

The researchers are studying intelligent transportation systems, or ITS. The ITS lab at MIT operates on a simple, if daunting, principle: Someday soon, there will be no more room for more roads.

Even in Boston, with the $14.6 billion Big Dig project, the transportation infrastructure will be at or near capacity. It's a concept called induced demand, which means: If you build it, cars will fill it.

For example, the Big Dig's new northbound tunnel was built to carry about 107,000 vehicles a day, 12,000 more than the old above-ground Central Artery. Project officials have said relatively uncongested driving should last until about 2010, when driving speeds are expected to slow. By 2020, officials predict, the new roadway could slow to a crawl similar to the tunnel's antiquated steel predecessor.

"The question is, what to do about that," said Moshe E. Ben-Akiva, director of the ITS program at MIT.

Researchers are using DynaMIT, a real-time traffic management system that combines speed and traffic data generated by video cameras and roadway sensors, and melding it with sophisticated simulators that help predict traffic congestion in a specific area up to an hour into the future.

The core of DynaMIT's ability to predict future snarls is the traffic simulaton, one of the first in the world to view traffic as a strange, unpredictable flow.

Traffic analysts had long viewed traffic flow as similar to water or sand trying to get through a channel. Nothing, however, was done to try to simulate traffic's vagaries -- the speeder, the slow driver, the confused, and the rude.

"People are not molecules," said Tomer Toledo, a research associate at the ITS lab.

By using algorithms and the second-by-second study of live traffic, Ben-Akiva and other ITS researchers developed a simulator that fit digital vehicles with individual personalities and destinations -- as many as 8,000 at a time. The program uses 15 factors to determine the personalities of those who drive each car, bus, or truck. There are digital drivers who ignore signs, change lanes behind slow drivers, or tailgate.

Called MITSIMLab, the simulator was used on the first major analysis of the Big Dig to assess the configuration of the toll plaza leading to the Ted Williams Tunnel.

At one point, Massachusetts Port Authority officials viewing a simulation of the tunnel entrance were in awe.

DynaMIT, financed in large part by the Federal Highway Administration, after federal officials saw what it did with the Big Dig, has been moving the process forward. Taking highway data and feeding it into the simulator, DynaMIT can forecast what is going to happen on a particular road. Researchers say they can add a scheduled Red Sox game into the simulator's equations to better predict backups.

"The better the quality of the data, the better the estimation," Ben-Akiva said.

Combined with variable message signs and in-car navigation systems, DynaMIT has the potential to not only predict traffic flow, but also reroute traffic to less congested roadways, helping drivers while cutting pollution caused by a snag of idling vehicles.

Quest to unclog traffic congestion

Inside a quiet Massachusetts Institute of Technology lab in Kendall Square, researchers are thinking hard about traffic jams, but not the ones they may face going home.

They're studying the congestion drivers could face years from now, and how to guide motorists -- some not yet born -- around the standstills of the future.

The researchers are studying intelligent transportation systems, or ITS. The ITS lab at MIT operates on a simple if daunting principle: Someday soon, there will be no more room for more roads.

Even in Boston, with the $14.6 billion Big Dig project, the transportation infrastructure will be at or near capacity. It's a concept called induced demand, which means: If you build it, cars will fill it.

For example, the Big Dig's new northbound tunnel was built to carry about 107,000 vehicles a day, 12,000 more than the old above-ground Central Artery. Project officials have said relatively uncongested driving should last until about 2010, when driving speeds are expected to slow. By 2020, officials predict, the new roadway could slow to a crawl similar to the tunnel's antiquated steel predecessor.

"The question is, what to do about that," said Moshe E. Ben-Akiva, director of the ITS program at MIT.

Researchers are using DynaMIT, a real-time traffic management system that combines speed and traffic data generated by video cameras and roadway sensors, and melding it with sophisticated simulators that help predict traffic congestion in a specific area up to an hour into the future.

The core of DynaMIT's ability to predict future snarls is the traffic simulator, one of the first in the world to view traffic as a strange, unpredictable flow.

Traffic analysts had long viewed traffic flow as similar to water or sand trying to get through a channel. Nothing, however, was done to try to simulate traffic's vagaries -- the speeder, the slow driver, the confused, and the rude.

"People are not molecules," said Tomer Toledo, a research associate at the ITS lab.

By using algorithms and the second-by-second study of live traffic, Ben-Akiva and other ITS researchers developed a simulator that fit digital vehicles with individual personalities and destinations -- as many as 8,000 at a time. The program uses 15 factors to determine the personalities of those who drive each car, bus, or truck. There are digital drivers who ignore signs, change lanes behind slow drivers, or tailgate.

The simulator will not, however, allow cars to have an accident unless a programmer calls for it.

Called MITSIMLab, the simulator was used on the first major analysis of the Big Dig, where engineers used it to assess the configuration of the toll plaza leading to the Ted Williams Tunnel. At one point, Massachusetts Port Authority officials viewing a simulation of the tunnel entrance were in awe.

"That's what we see every day outside our window," Ben-Akiva recalled them saying.

The simulator helped Big Dig designers determine that eliminating an off-ramp inside the new northbound Interstate 93 tunnel would cause some delays, but would also save the project about $40 million. The designers eliminated the off-ramp.

"We went from theory to actual practice in a virtual-reality setting," said Big Dig spokesman Sean O'Neill. "We were able to put the human behavior inside the design. Concrete and steel are inanimate objects. They don't drive like a maniac in the slow lane."

DynaMIT, financed in large part by the Federal Highway Administration after federal officials saw what it did with the Big Dig, has been moving the process forward. Taking highway and feeding it into the simulator, DynaMIT is able to forecast what is going to happen on a particular road. Researchers say they can add a scheduled Red Sox game into the simulator's equations to better predict backups.

"The better the quality of the data, the better the estimation," Ben-Akiva said.

Combined with variable message signs and in-car navigation systems, DynaMIT has the potential to not only predict traffic flow, but to also reroute traffic to less congested roadways -- helping drivers while cutting pollution caused by a snag of idling vehicles.

The system has been tested in Irvine, Calif., with some success. It is about to be tested in Los Angeles, Westchester County, N.Y., and Hampton Roads, Va., Ben-Akiva said.

In Boston, where the Big Dig has installed roadway sensors to help pinpoint accidents and route traffic around them, the potential to enhance the current system to include something like DynaMIT is farther ahead.

"It would be great to see this innovation," said O'Neill, who added that Turnpike Authority chairman Matthew Amorello supports the use of this new technology. Money and time, however, are major stumbling blocks.

"They took a very conservative approach," Ben-Akiva said. "They were only going to deploy proven technology."

Taming Traffic

New Projects attempt to predict congestion and help drivers steer clear

Even when they are on roads equipped with advanced traffic data collection and advisory systems, drivers know that a fender bender can turn the morning rush hour into an endless wait. Transportation researchers at the National University of Singapore say that although they can't prevent crashes or rubbernecking, they'll soon have a better way to disperse traffic and avoid jams: computers will identify the best response and change electronic highway signs to suggest alternate routes, for instance. The Singapore project is just one of several traffic prediction efforts that, by 2004, could be saving drivers' time in cities such as Tokyo, Los Angeles, Houston, and Stockholm where "intelligent highway" infrastructures are already in place.

In these and many other cities, cameras, magnetic loops in roadbeds, and even signal patterns from drivers' cell phones provide analysts with raw data on traffic speed and density. But today's systems for utilizing those data have two big shortcomings: When gridlock sets in, the responsibility for interpreting the data and recommending responses falls to human traffic managers, who sometimes err. Furthermore, their recommendations offer no look ahead. "Without good predictions, you cannot come up with good traffic management," says Henry Lieu, a transportation engineer at the Federal Highway Adminstration's research labs in McLean, VA.

The Singapore system aims to provide traffic predictions early enough for drivers to act on them. In the aftermath of an accident, the system analyzes traffic conditions in the surrounding area to determine which response -- lane closures, light cycle adjustments, or driver advisories -- will restore order fastest. The problem is so complex that most computers can't keep up, but the Singapore team has developed efficient algorithms and data-mining strategies that generate predictions and select the best response within seconds. "The system will pick the highest-performance strategy, so you can implement that traffic control strategy on the real traffic network," says Der-Horng Lee, a civil engineer who led the project at the National University of Singapore. His country expects to deploy the system on its 300 kilometers of expressways by August 2005, he says.

A similar traffic-prediction system under development at MIT incorporates feedback -- the response of drivers to announcements of anticipated points of congestion. Traffic centers in Los Angeles and McLean plan to implement the program in 2004. Such systems will add intelligence to the roadways -- although researchers still haven't figured out how to add any to drivers.

Traffic Report

An excerpt from a Boston Magazine article. Full article can be viewed in the May 2002 issue.

...Late morning in a computer lab near the Kendall Square T Station, and Tomer Toledo is playing God.

The view from above shows that traffic is flowing smoothly, tiny boxes of yellow and blue and pink and red shuttling among the wishbone of the Central Artery. There are boxes sized like tractor-trailers and boxes sized like Hyundais, each with its own individual license plate number and driver profile and route. It is all very orderly, a ribbon of color and geometric shapes, until Toledo, a Ph.D. student at the Massachusetts Institute of Technology, presses a series of keys, closing two lanes. Then the boxes begin to funnel to the right and line up like hotels on a Monopoly board.

"I can generate an incident however I want and for as long as I want," Toledo says, satisfied. "I can close as many lanes as I want."

This is the advantage of a traffic simulator like MIT's MITSIMLab: Simulated drivers, of simulated automobiles, with simulated lives do not generally complain about two lanes of traffic being closed on a major expressway. By contrast, real drivers, of real automobiles, with real lives may incite riots. So MITSIM can aid in predicting patterns for a major project like the Big Dig without the distraction of dealing with real people, helping officials learn how to manipulate traffic in the tunnel's lanes using signs and ramp meters.

On the other side of the room, a graduate student named Srini Sundaram is using a red pen to draw a series of matrices in a reporter's notebook. What he is attempting to elucidate seems inexplicable enough without the abbreviations and arrows and percentages, and a diagram that looks like a crude outline of a gas grill.

What he is trying to do, he says, is predict the future.

This is something they call DynaMIT, a traffic estimation and planning system that plays on the basic notion that tie-ups can be calculated in advance, and that people will alter their behavior and therefore alter traffic patterns according to what an electronic sign tells them.

"When you give people guidance based on the weather, it's not going to change the weather," Sundaram says. "But if you give drivers certain information, it will change their decisions."...

Traffic Software Can Predict the Future

Have you ever been stuck in a traffic jam on the highway, or wished that you had the ability to avoid commuting headaches altogether? If your answer is “Yes,” you may be interested in DynaMIT – a real-time, dynamic traffic assignment system developed by the researchers at the Massachusetts Institute of Technology Intelligent Transportation Systems Program (MIT ITS). DynaMIT was developed to tell commuters, en-route, the quickest way to get to work.

How It Works
DynaMIT works as part of a traffic management system that collects data, compiles current driving conditions, and simulates future driver behavior. For example, DynaMIT may use data, like the number of cars on a roadway and the speed at which they are traveling, to predict driving conditions in the near future. Traffic researchers have, over the past two decades, developed and implemented methods, such as video surveillance cameras or strategically placed magnetic sensors, to collect traffic data. At the same time, researchers have developed sophisticated traffic simulators that show how traffic would be affected under a variety of conditions. DynaMIT is one of a handful of programs which utilizes the results of traffic research efforts, combining both collection and simulation information to predict specific roadway traffic conditions based on real-time data.

DynaMIT is currently being tested in Irvine, California -- one of the few cities that has incorporated sensor data into its traffic management system. The major roadways in Irvine have built-in magnetic sensors that detect the number of cars on a roadway at each sensor location. Researchers refer to this measurement of traffic as flow. Numerous studies have been done on traffic flows, and MIT ITS researchers have been able to use traffic flow research to produce a variety of simulated driving scenarios. In Irvine, real-time data from roadway sensors is fed into a traffic simulator every 30 seconds. The simulator then uses DynaMIT to compute the most likely scenarios. DynaMIT can produce traffic scenarios for any interval of time, up to 45 minutes into the future.

Applications
A predictive program, such as DynaMIT, is very desirable for traffic management centers across the U.S. Currently, data from magnetic or video sensors can only tell traffic managers what is happening at a particular moment on the road. DynaMIT provides managers with a way to determine what is likely to happen in the near future, based on the current conditions. DynaMIT’s information will, for example, allow traffic managers to alert drivers through computerized road signs and in-vehicle navigation or information systems that a traffic jam is likely to happen on Highway XYZ within 25 minutes, or that a faster route to the airport is available for the next hour by taking Exit 123.

The Federal Highway Administration (FHWA) sponsors the DynaMIT research project and plans on completing the on-line Irvine tests by early summer, 2002. The FHWA's long-term goal for DynaMIT is to integrate it into traffic management centers in busy U.S. cities in order to reduce traffic congestion and improve a driver’s commute.

Doctoral Graduates Awarded Thesis Prizes

Recent MIT CEE grads Joan Walker and Jon Bottom were awarded by the Transportation Science Section of the Institute for Operations and the Management Sciences (INFORMS) first and second place honors, respectively, for their doctoral thesis works. Prize Committee Chairman Ismail Chabini announced the award November 5, 2001, at the annual TSS meeting, held in Miami Beach.

Ms. Walker's thesis entitled, "Extended Discrete Choice Models: Integrated Framework, Flexible Error Structures, and Latent Variables," shared first prize honors with another thesis by student Alan Erera, from the University of California at Berkeley. Similarly, the second place prize was shared by MIT grads Andrew Armacost (Aero/Astro) and Jon Bottom.

The decision to have co-winners for both first and second prize was somewhat unique this year. In the past, one prize each was awarded for first place, second place, and honorable mention. Chairman Chabini described how the entries were judged: "The only criteria was quality, and all four theses were excellent."

Ironically, Ms. Walker, Mr. Bottom and Mr. Armacost worked in the same MIT graduate student office during their early doctoral research years. "It's a huge honor," exclaimed Ms. Walker on her prize, "and it's really nice to be recognized with Jon and Andrew."

Professor Moshe Ben-Akiva, research advisor to both Ms. Walker and Mr. Bottom, remarked, "It's fantastic that both Jon and Joan were recognized by INFORMS. Their work is first-class."

Traffic Model Has Human Touch

CAMBRIDGE-- For years, transportation planners building computer models of new highways have treated traffic as the equivalent of a fluid, a sort of viscous mass trying to flow through a series of channels. But everyone -- especially those who drive in Boston -- knows that drivers are for from being undifferentiated molecules. Some are grandparents. Some are cabbies. Some are truckers. And altogether too many are lane-hopping, finger-flipping, stoplight-running, tailgating speeders. Now the full range of human driving behavior, from crazed to courteous, and its impact on highway operations are being reflected in an elaborate new computer program. Developed by Massachusetts Institute of Technology transportation researchers, it is being used by builders of the $8 billion Central Artery/Third Harbor Tunnel project to find trouble spots that may need redesigning.

Instead of simply projecting theoretical flows, the MIT program depicts real stretches of highway, complete with uphill and downhill grades, and creates digital vehicles with specific destinations -- as many as 8,000 at a time -- to drive down them.

The program not only simulates different kinds of vehicles, such as trucks, buses, zippy new cars and logy old cars, it also mimics different kinds of drivers.

Based in part on observations of real-world drivers, the program assigns each vehicle's driver a "personality" including 15 different factors, such as propensity to speed, tailgate, ignore traffic signs and refuse to dawdle behind slowpokes.

The result is a computer program that comes far closer than previous attempts to mimicking the way traffic actually behaves, helping Artery-tunnel planners to figure out the best strategies.

"We have developed essentially a laboratory in which we can test new traffic-management technologies," said MIT civil engineering professor Moshe Ben-Akiva, chief of the Intelligent Transportation Systems Program, which has been working on the computer program for four years.

Added Sergiu Luchian, a top Highway Department official overseeing the Artery-tunnel: "It's been a good tool for us."

As different colored boxes zip across computer screens representing the Boston highway network of the future, Ben-Akiva's research crew tests such questions as how quickly after an accident a stretch of highway will jam up, or how far back motorists should be warned of an upcoming lane closure to minimize backups.

The MIT computer has proven particularly useful for Artery-tunnel engineers as they design a $54 million traffic control system for the seven miles of new highway. When complete, it will include everything from 500 full-color cameras to thousands of speed sensors embedded in the pavement, all beaming information back to a traffic "war room" in South Boston.

One example Luchian cited involves adjustable speed-limit signs on the new highways. In the case of clearing an accident from the Ted Williams Tunnel, Luchian said, Artery-tunnel engineers had planned to put the speed limit back up to 50 m.p.h. as soon as the tow truck was leaving the scene.

But running that scenario on the computer, they found that it would simply rush a flood of traffic ahead and create a new jam behind the tow truck. Now, they have revised their program to have the variable signs step up the speed limit incrementally to ease traffic out more smoothly.

"That was a very good outcome," Luchian said.

The computer also helped Artery engineers figure out the optimum number of toll booths to have in service when the new tunnel opened for commercial vehicles and taxis in December.

The computer accounted for the range of speeds at which different drivers get through tolls by considering the percentage of drivers who have their money out beforehand and those who fumble for change at the toll booth, delaying drivers behind them.

As the project moves forward to designing and building the depressed Artery downtown and the Charles River Crossing, Luchian said, officials will feed proposed highway and ramp designs into the MIT computer to see if they contain unexpected snarl points, and redesign them accordingly.

It's much easier...to fix design issues than to fix construction," Luchian said.

There is no shortage of rude, erratic drivers on the MIT computer. It is not unusual to see a few digital cars, warned that their lane will be closed a quarter-mile ahead, fly down that lane past the backup in the other lane and then try to muscle in ahead of the cars that have been waiting.

"It's actually annoying when you see some of these guys zooming down and trying to sneak in," Ben-Akiva said.

"Everyone sees some of their worst driving habits" reflected, agreed research associate Rabi G. Mishalani.

Many rude driving habits can be defined by mathematical formulas or bell-curve representations -- for instance, of the probability that a driver will stop for a yellow light. Two formulas that guide the MIT system express the probabilities for how long different drivers will wait behind a slow vehicle before changing lanes, and how much clear space they need to see in the next lane before deciding to move over.

To enhance the realism of the program, Ben-Akiva's team borrowed videotapes of traffic on Storrow Drive and the Central Artery from SmartRoutes, a Cambridge traffic monitoring company. The images of real cars were fed into a computer, then digitized to provide "Boston driver" profiles.

David Bernstein, a Princeton University engineering and transportation expert familiar with the MIT traffic forecasting system and sever others, said some others try to reflect driver behavior, but MIT researchers are well ahead of the curve in plugging real-world observations into the system.

Also, Bernstein said, the MIT system is unique because it can "predict the effects of variable message signs or variable speed limits or incident detection and management."

One drawback is that the computer's advice is only as good as the numbers fed into it. It runs on projections of traffic flow in 2010 along not-yet-built highways -- the type of forecasts that in countless cities and roadbuilding projects have proven off the mark. But it is flexible enough to test a range of traffic levels, Luchian said.

One future use for the computer, say Ben-Akiva and Mishalani, would be to advise the Artery-tunnel "war room" supervisors how to unsnarl jams of the future.

In their scenario, as a breakdown on the approach road to the Ted Williams Tunnel is beginning to jam traffic to the Expressway, the traffic masters punch in their plan for unsnarling it, such as a recipe of closing lanes or broadcasting a detour on message boards -- then have the computer run that scenario and tell them whether it will really improve traffic.

Because any highway system can be plugged into the computer, Mishalani said, it is possible other cities and states could use it to test traffic-management strategies.

Traffic Simulator Incorporates Driving Styles

MIT engineers have developed a state-of-the-art traffic simulator that actually mimics different drivers-aggressive, careless, timid or fast-and how they affect traffic flow.

The simulator enables the researchers to test various traffic management systems for the Central Artery/Harbor Tunnel (CA/T), the big construction project that is relocating the elevated portion of Boston's I-93 below ground and extending the Massachusetts Turnpike (I-90) to Logan Airport via a new tunnel.

The MIT research will help the CA/T system managers maintain a smooth flow of traffic on the new stretch of freeway by allowing them to test traffic management systems under simulated conditions that closely resemble the real thing. As a result, when the actual CA/T opens to the public it is likely that the best traffic management system will already be known and operational.

The traffic simulator, which runs on a workstation, is called the MIcroscopic Traffic SIMulator (MITSIM). Microscopic refers to the treatment of traffic as a set of individual vehicles, or particles, allowing each vehicle to move according to its own characteristics. The more common macroscopic simulator treats traffic like a fluid, assigning one set of characteristics to the entire stream of cars. MITSIM is more lifelike because it allows for differences in vehicles' movements as dictated by drivers' personalities.

"The simulator is designed to be very flexible, making it possible to incorporate various driver behavior models and a wide spectrum of traffic management system designs," said Rabi G. Mishalani, a research associate in the Intelligent Transportation Systems Program of the Center for Transportation Studies. It has more variable components than most other simulators and a graphical user interface (GUI) that allows the operator to watch the cars move through the simulated freeway on the computer screen.

When a researcher starts the simulation and activates the GUI, she sees a bright, multicolored ribbon on the computer screen. When she zooms in closer she sees a section of the new freeway, with layers of tunnels and roads, on- and off-ramps and little moving colored rectangles. Each rectangle represents a vehicle with its assigned peculiarities. They change lanes, exit hurriedly and sometimes even have accidents as they test the potential of a proposed traffic management system.

A good traffic management system controls lane use and traffic volume, determines toll collection placement and methods, and sets speed limits, among other things. It should also include a set of surveillance devices for collecting information about the vehicles, like speed and type, and be comprehensive enough to take into consideration changing weather and traffic conditions. Decisions about each of these components must be made in tandem, because each affects the other. MITSIM allows the system designers to fiddle with the components until they come up with the best combination.

For each traffic management system simulation on MITSIM, data about the number, types and destinations of vehicles are given to the computer program. Highway characteristics, like number of lanes and road surface, are also provided. However, vehicle characteristics, such as desired speed, are determined by mathematical models and assigned by the simulator itself.

"At the fundamental level, vehicle behavior is in the form of a set of mathematical relationships, each one invoked under certain conditions. The conditions under which each should be used are in the form of program instructions," Dr. Mishalani said.

As each vehicle enters the simulated Central Artery, it grabs a packet of vehicle characteristics that determines how it will act in certain circumstances. Not only does each vehicle have a size, type, occupancy level and destination, it also has driver characteristics. These include desired speed, propensity to yield to other vehicles, lane-changing behavior and route decisions. There's even a driver impatience factor that makes each driver's choices more realistic.

If MITSIM, with its little colored rectangles serving as vehicle substitutes, works well in its evaluation of traffic management systems, these MIT researchers will have helped to prepare the CA/T for our eventual use.

This work is sponsored by the Massachusetts Highway Department through Bechtel/Parsons Brinckerhoff. Professor Moshe Ben-Akiva of the Department of Civil and Environmental Engineering is the principal investigator.

Three from MIT are winners of Discover finalist awards

MIT researchers and their colleagues have won three finalists'prizes in the annual Discover Magazine award competition.

The prize-winning innovations were swift new audio-clipping software, a microscopic traffic simulator that predicts traffic flow based on drivers' real driving styles, and a technology known as PAN that transmits information among communications devices carried by an individual, including cellular phones and pagers.

The Discover Awards acknowledge "the creativity of the men, women and corporations/institutions who have reached superior levels of ingenuity." Each of the winners and finalists will be featured in the July issue of Discover. They also received all-expenses-paid trips to Disneyworld for themselves and a guest.

Discover Awards are given in seven categories: automotive and transportation, aviation and aerospace, computer hardware and electronics, computer software, environment, sight and sound. Five finalists are named per category; out of the five, one is chosen as overall winner for that category. An eighth category, "Editor's Choice," has only one finalist and winner. This category is for "any new innovation or technology that, by virtue of its `newness,' does not fit into any of the other categories."

TRAFFIC SIMULATOR
The traffic simulator chosen for a Discover finalist award was developed by Qi Yang as a component of both his SM and PhD theses (Mr. Yang received his doctoral hood last Thursday).

Mr. Yang's traffic simulator, which runs on a workstation, is called MITSIM, for MIcroscopic Traffic SIMulator. Microscopic refers to the treatment of traffic as a set of individual vehicles, or particles, allowing each vehicle to move according to its own characteristics. The more common macroscopic simulator treats traffic like a fluid, but anyone who spends more than five minutes in a car between, say, Quincy and Boston, can see that "fluid" does not describe the scene. MITSIM is more lifelike because it allows for differences among drivers' personalities, he said.

MITSIM already has a practical application: it helps researchers to test various traffic management systems for the Central Artery/Harbor Tunnel (CA/T), the sprawling "Big Dig" construction project that will eventually relocate the elevated portion of Boston's I-93 underground and extend the Massachusetts Turnpike to Logan Airport via a new tunnel.

"The simulator is designed to be very flexible, making it possible to incorporate various driver behavior models and a wide spectrum of traffic management system designs," said Rabi G. Mishalani, a research associate in the Intelligent Transportation Systems Program in the Center for Transportation Studies. It has more variable components than most other simulators and a graphical user interface (GUI) that allows the operator to watch the cars move through the simulated freeway on the computer screen.

When a researcher starts the simulation and activates the GUI, she sees a bright, multicolored ribbon on the screen. When she zooms in closer, she sees a section of the new freeway, with layers of tunnels and roads, on- and off-ramps and little moving colored rectangles. Each rectangle represents a vehicle with its assigned characteristics. They change lanes, exit hurriedly and sometimes even have accidents, just like real cars do.

For each traffic management system simulation on MITSIM, data about the number, types and destinations of vehicles are given to the computer program (for example, 524 vehicles driven by Bruins fans are headed south after midnight during a light blizzard with high winds). Highway characteristics such as the number of open lanes and conditions of the road surface are also provided.

"At the fundamental level, vehicle behavior is in the form of a set of mathematical relationships, each one invoked under certain conditions," Dr. Mishalani said.

As each vehicle enters the simulated Central Artery, it grabs a "packet" of vehicle characteristics that will determine its behavior on the road. Not only does each vehicle have a size, type, occupancy level and destination--it also has driver characteristics. These include the driver's desired speed, propensity to yield to other vehicles, lane- changing behavior and route decisions. There's even a driver impatience factor that makes each driver's choices more realistic.

Mr. Yang's work was sponsored by the Massachusetts Highway Department through Bechtel/Parsons Brinckerhoff. Professor Moshe Ben- Akiva of the Department of Civil and Environmental Engineering is the principal investigator.

ITS Home | Research | Publications | Events | Press | MITSIMLab | DynaMIT | Staff | Contact

Comments and questions to its-www@mit.edu

Photography Credits: photos used on this page can be found at FreeImages.