BLACKSBURG, Va., Sept. 27, 2012 – A couple of years ago, Hesham Rakha misjudged a yellow light and entered an intersection just as the light turned red. He got a ticket.
"There are circumstances, as you approach a yellow light, where the decision is easy. If you are close to the intersection, you keep going. If you are far away, you stop. If you are almost at the intersection, you have to keep going because if you try to stop, you could cause a rear-end crash with the vehicle behind you and would be in the middle of the intersection anyway," said Rakha, professor of civil and environmental engineering at Virginia Tech.
He is not trying to defend his action. Rakha, who is director of the Center for Sustainable Mobility at the Virginia Tech Transportation Institute, is describing his research. Since 2005, his research group has been studying driver behavior as they approach yellow lights in order to determine signal times for intersections that are safer, yet still efficient.
If the driver decides to stop when he or she should have proceeded, rear-end crashes could occur. If the driver proceeds when he or she should have stopped, a collision with side street traffic could occur. Although observation-based research shows that only 1.4 percent of drivers cross the stop line after the light turns red, more than 20 percent of traffic fatalities in the United States occur at intersections.
"If the yellow time is not set correctly, a dilemma zone is eminent," Rakha said.
"The dilemma zone occurs when the driver has no feasible choice," he said. "In other words, the driver can neither stop nor proceed through the intersection before the light turns red. This can also occur if the approaching vehicle is traveling faster than the posted speed limit and/or if the driver’s perception and reaction time is longer than the design one-second value."
In most cases, the yellow time is set for 4.2 seconds on a 45 mph road. The time is longer for higher-speed roads. "These timings are based on two assumptions," Rakha explains. "Namely, the driver requires one second to perceive and react to the change in signal indication and that the driver requires 3.2 seconds to stop from 45 mph at a comfortable deceleration level, assumed to be 3 meters per second squared (3 m/s2 ) or 10 feet per second squared."
Rakha's studies have used the Virginia’s Smart Road, located at the Virginia Tech Transportation Institute. The Smart Road has a signalized intersection with a controllable light. "We can study driver behavior by changing the signal when the driver is at a certain distance from the intersection."
The Center for Sustainable Mobility researchers have determined that half of drivers make the stop-go decision three seconds before the stop line and that of those that go, few clear the intersection before the light changes to red. In Virginia, if you are in the intersection when the light turns red, you are not running a red light. However, there is still risk. "Typically, an all-red interval is introduced to clear the intersection of all vehicles that enter the intersection legally before releasing any conflicting movements," Rakha said.
The specific findings from the Smart Road study are that 43 percent of the drivers, who crossed the stop line during the yellow time, were not able to clear the intersection before the light turned red. At 45 mph, it takes 1.5 seconds to clear a 30-meter (98.4 feet) intersection. "If the all-red interval is the minimum conventional one second, then there is a potential risk that the legal yellow-light runners would not be able to completely clear the intersection at the instant the side-street traffic gains the right-of-way," the researchers reported at the Transportation Research Board Annual Meeting in 2010.
People over 60 years of age have a longer perception-reaction time, so they have to brake harder to stop. But they are more likely to try to stop, compared to younger drivers. However, if they keep going, they are unlikely to clear the intersection, the researchers report.
"Even if yellow timing is designed properly to avoid a dilemma zone, someone driving above the speed limit could encounter a dilemma because it takes longer to stop from a higher speed. They could speed up, but our studies show that drivers who decide to keep going maintain their speed," Rakha said.
The research determined that the perception-reaction time is slightly longer than one second, but that driver deceleration levels are significantly higher than the deceleration level assumed for traffic signal design.
If the road conditions are poor, drivers react 15 percent slower because they are processing more information. "We thought there might be concern about whether following cars will be able to stop, but a research project that included following cars showed no difference in behavior," Rakha said. "Ability to stop on a wet surface was a consideration. Drivers realize they cannot brake aggressively because of the wet roadway surface." Deceleration level decreases by 8 percent. "In such conditions, you need a longer yellow," Rakha said.
He and his group have developed a novel procedure to compute the yellow and change interval duration that accounts for the risk of drivers being caught in the dilemma zone. Using this procedure they have created look-up tables for light vehicles at various speeds for dry roads and wet roads, so that traffic planners can set traffic signals that can be adjusted to roadway surface and weather conditions. They also created tables for the driving characteristics for different age groups.
"You could design the yellow time so that 90 percent of drivers – or 95 percent are not caught in the dilemma zone. But the higher you go, the longer the yellow time is, which increases the time cross traffic is stopped," he said.
Stopped traffic wastes fuel and increases air pollution. "From a fuel-consumption perspective, we want drivers to keep going if they can do so safely, rather than idle and use fuel and emit pollutants," said Rakha.
Also, drivers who know an intersection has a long red light might be more inclined to speed through a yellow when it is unsafe to do so.
One strategy might be more use of caution lights that tell drivers a green light is about to change, so the driver has a longer time to react. Such systems are now used on high speed roads, where the stopping distance is longer, and when the lighted intersection is not visible until the last seconds.
A future strategy that researchers are already working on is in-car display systems that can be customized to each driver's reaction time. "So one person receives a four-second warning of a light change, and another person receives a five-second warning. Or, instead of a warning, the system might just tell you to stop," said Rakha.
Most traffic planners and auto designers now accept that communication between traffic signals and vehicles will come to pass. But only four years ago, when Rakha submitted a journal article that suggested signals could communicate with cars, the article reviewers felt that this was unrealistic and infeasible.
Now research to implement vehicle to vehicle (V2V) and infrastructure to vehicle communication is ongoing. The Virginia Tech Transportation Institute is home to a Tier 1 University Transportation Center, where connected vehicle/infrastructure research is being conducted.
"If all the vehicles are communicating with each other, we might not need lights at all," said Rakha. "Cars would adjust their approach to an intersection like airplanes adjust their arrival at airport runways." V2V communication at intersections is one project in Rakha's group.
In the meantime, the results can be incorporated within traffic simulation software to enhance the modeling of driver behavior at the onset of a yellow indication. This software can then be used to design and test alternative dilemma zone mitigation strategies. The results can also be used in dilemma zone protection systems to provide better estimates of the dilemma zone boundaries.
The research has been presented at the 2010 and 2011 Transportation Research Board annual meetings. Articles published in 2012 are: