I love traffic signals. Consider what an extraordinarily simple solution they are to what many believe is the most complicated – and most dangerous – of all traffic situations: the intersection. There are hundreds of thousands of intersections just in the United States, and every day, millions of vehicles cross paths; despite rather overwhelming odds, most vehicles will pass successfully, thanks in part to those three colored lights.
Of course, it is not a perfect system; there are too many variables beyond the traffic signal that can affect the safety and efficiency of intersections, not the least of which is driver behavior. But by and large, the simple and effective way that those green, yellow and red lights control traffic flow through intersections is remarkable.
At the same time, I hate red lights. I freely admit that this is a completely contradictory statement, but like most drivers, I can’t stand to be stopped. I take it very personally when the light changes to red and I am forced to interrupt my drive, and waiting at a red light when there is no cross traffic is an exercise in impatient aggravation. I respect the red light, I even accept that it serves a useful and necessary purpose–I just prefer that it happen to someone else.
Knowing when to change the light to red, for how long and why, is the trick – and challenge – for traffic signal engineers around the world. When is a light change warranted? Will a change to red promote efficiency or will it only disrupt traffic flow? Understanding the effects of a red light and knowing how to implement it is the very essence of intelligent traffic control, and like all intelligent transportation systems, is dependent upon the delivery of accurate, reliable data.
There is no way to know how many intersections there actually are in the world, or how many of them are signalized. According to the US Federal Highway Administration’s Manual on Uniform Traffic Control Devices, the US has not collected that kind of information. The manual references a 2004 study by the Institute of Transportation Engineers that suggests, as a “rule of thumb,” assuming one signalized intersection for every 1,000 population: at the US’s current population of 318 million, this would mean 318,000 signalized intersections; by applying the same logic to the world’s population of over seven billion, we’d expect over seven million intersections around the world, servicing perhaps hundreds of millions of vehicles.
Keeping all of that traffic moving safely and efficiently is no easy task. The problem has existed as long as there have been roads; attempts to control it can be traced to manually-operated semaphores used in London as early as the 1860s. The introduction of the automobile only exacerbated the problem, and it quickly became apparent that a means of alternately assigning right-of-way at intersections was required to help prevent accidents.
The first recorded automobile traffic light appeared at a Salt Lake City intersection in 1912, and the first three-colored signal debuted in Detroit in 1920. Not long after, the first fixed-time signal controllers were introduced, and by 1924, Los Angeles had the first traffic signal network consisting of 31 timed intersections that could be controlled from a single, central location – the world’s very first traffic control center.
Photos courtesy of tennesseelawman.com,brinkster.com, AND Geographer at en.wikipediacom
Pictured above, left-to-right: 1.Early traffic control attempts used operated semaphores in London as early as the 1860s. 2.The first three-colored signal debuted in 1920. 3.Los Angeles’ ultra-modern traffic control center has come a long way since it was first created in 1924.
Signal timing was a revolutionary innovation because it introduced a level of automation to a process that became increasingly more difficult to manage. Manual systems required someone on-site to observe traffic conditions and change the lights accordingly; timing schemes presented a more cost-effective solution, saving municipalities both money and man-power.
Unfortunately, signal timing represents almost the antithesis of intelligent traffic. Conventional timed signals simply follow a pre-programmed daily schedule, and they operate at the same schedule no matter how heavy or light traffic may be. This means timed signals will change to red whether the change is warranted or not; and the red light won’t change no matter how many vehicles may be waiting. In other words, timed signals will distribute red lights in a way that is thought to be equitable, but they cannot determine when is the best time to change to red or whether a red light is really necessary, and this can lead to significant inefficiencies; in fact, the FHWA identifies poor or outdated signal timing as a contributing factor to traffic congestion and delay.
This is not to say that signal timing does not require intelligence. The mathematics required to implement an adequate timing schedule are impressive, and maintaining timing plans over time is surprisingly labor intensive, which is one of the reasons that transportation agencies with limited resources and tight budgets so rarely maintain them. But timed signals require no other information to operate, so they cannot, and do not, respond to actual traffic conditions.
The ability to gather real-time data and use that information to positively affect traffic flow is the central purpose of intelligent transportation systems. Thus, an intelligent intersection has the ability to monitor traffic conditions and respond to them in a way that will promote efficiency and increase safety.
There are three main categories of “intelligent” intersections: Traffic Actuated signals, which “activate” a signal change when a vehicle is detected; Traffic Adaptive signals, which can create timing plans based on real-time data; and Traffic Responsive signals, which use real-time data to adjust existing timing schedules. While each system has its own merits, the overall objectives are predominantly the same: to provide as much green light as possible to existing traffic; to distribute the green light as fairly to the intersecting directions of travel; and to reduce congestion and improve travel time by keeping traffic flowing smoothly through the intersection.
The kind of data required by each system can vary greatly from one deployment to the next, but in general, signal systems need accurate real-time vehicle presence and volume data to be effective; they also benefit from occupancy and density data and vehicle speeds, as well as headway and queue length. Together, these parameters paint a picture about traffic flow, which the FHWA, in chapter three of the Traffic Control Systems Handbook, says is “fundamental in analyzing intersection delay or capacity.”
To collect this data, agencies deploy vehicle presence detectors in some proximity to the intersection stop bar. Stop bar, or stop line, detection looks for the presence of vehicles and identifies gaps in vehicle presence that are sufficient in length to warrant changing the light to red. Mark Taylor, traffic signal operations engineer at the Utah Department of Transportation, says, “Stop bar detection allows for both calling and extending of the phase and improves efficiency by minimizing the cycle time needed to serve minor movement phases.”
The first stop bar detectors might actually have been the traffic police officers who at one time directed traffic by hand and later, manually operated traffic signals. These officers visually detected traffic and responded accordingly. One of the earliest attempts to automate detection came in 1928 with an audio sensor that consisted of a microphone mounted in a small box on a nearby utility pole; drivers would honk their horns to trigger the change to a green light. People living nearby complained about the noise, and the horn-triggered sensor quickly disappeared.
“It wasn’t until recently that non-intrusive detectors like radar have advanced to a point that they, too, reliably detect vehicles and bicycles at intersections.”–Mark Taylor, Operations Engineer at Utah Department of Transit
Much more successful were the pressure-sensitive detectors that were installed at about the same time. These consisted of a rubber pad set into a metal frame that lay flush with the surface of the road. Two metal strips embedded in the rubber served as electrical contacts which were brought together by the weight of passing vehicles to result in a detection. These detectors lasted for more than 40 years and led to the development of the most popular vehicle detector in the world, the inductive loop – a coil of insulated wire that is embedded within the surface of the road. Loops generate an electrical charge that is disrupted by the metal in vehicles that pass overhead.
“For a long time, and even today, inductive loops have been the gold standard of detection, due to its superior accuracy in detecting vehicles,” UDOT’s Taylor says. “It wasn’t until recently that non-intrusive detectors like radar have advanced to a point that they, too, reliably detect vehicles and bicycles at intersections.”
According to Taylor, there are two main benefits of stop bar detection: efficiency and performance measures. First, Taylor says accurate and reliable detections are an essential part of effective intersection control.
“Reliable stop bar detection is important for efficient signal operations,” Taylor says. “Effective detection that is accurate, flexible, easy to configure and capable of detecting vehicles and cyclists with abundant channels is the key to meeting a growing society’s transportation needs.”
Taylor says that when it comes to efficiency, it is also important to understand the relationship between detection zone size, vehicle extension (allowable gap) and maximum and minimum green times. “Maximum greens should be set high enough to gap out most of the time under light to moderate traffic, but low enough to maintain efficient cycle lengths under heavy traffic conditions,” says Taylor. “In general, shorter extensions reduce delays.”
According to Taylor, there are two main benefits of stop bar detection: efficiency and performance measures.
To accomplish this, UDOT uses the Wavetronix SmartSensor Matrix to create detection zones approximately 65 feet in length, beginning a few feet in front of the stop bar. This isn’t always possible with loops because larger loop zones are known to have sensitivity issues and can be cost-prohibitive. “The larger detection zone allows us to shorten the vehicle extension parameter by as much as two seconds,” Taylor explains. “If you assume 10 vehicles are waiting every two-minute cycle for just half a day, and if a person’s time is assumed to be worth $15 per hour, then the delay savings we achieve is approximately $30 a day, or $900 a month per intersection.”
Second, the data from reliable stop bar detection also provides departments of transportation with measures of effectiveness that Taylor says are critical to traffic management and planning.
“Agencies commonly struggle to monitor the health of vehicle detectors, frequently relying on complaints from the public to identify problems,” Taylor says. “Obviously, this is not ideal. At the same time, executive leaders and public officials are interested in program-wide signal performance and trends. Are signal operations getting worse? Getting better? They also want to know how an agency prioritizes resources and workload.”
The FHWA’s Signalized Intersections: Information Guide notes the challenge agencies face in providing outstanding customer service with limited resources: “Performance measures allow practitioners to assess the effectiveness of a signalized intersection or corridor.” Taylor says some of the data most useful as indicators of performance include lane-by-lane volume counts, volume-to-capacity ratios, split saturation and detector failures.
“By using effective detection that can also provide these MOEs, agencies will be able to improve mobility, increase safety and use resources more effectively,” Taylor says.
The principles of effective intersection control have changed very little in the last 100 years. An article titled “Small-area detection at intersection approaches” in the February 1974 issue of ITE Journal reads very much like any intersection-related article written in 2014; the scope, challenges and basic fundamentals remain very much the same.
In fact, the only thing that has changed significantly at the intersection is the quality and type of vehicle detection that is now available. Modern vehicle detectors provide a level of data that simply wasn’t possible just a few decades ago, and it’s opening up a realm of possibilities for the future of transportation that not long ago would have been written off as science fiction.
The principles of effective intersection control have changed very little in the last 100 years… In fact, the only thing that has changed significantly at the intersection is the quality and type of vehicle detection that is now available.
US Department of Transportation Secretary Anthony Foxx believes so strongly in the importance of data in developing tomorrow’s intelligent systems, he has launched the Data Innovation Challenge, giving interested innovators access to the agency’s data and an invitation to “revolutionize America’s transportation system.” In the May 2014 issue of Fast Company magazine, Foxx said planning to incorporate massive population growth into our transportation systems is one of the greatest challenges the USDOT faces, and it requires taking those systems into the 21st century in a bold way.
“I want the folks who use these to understand that we are embracing the technology to make them safer and move more efficiently,” Foxx said.
Mark Taylor fully agrees. As an experienced traffic signal professional, Taylor says he dreams of a day when traffic data moves almost symbiotically between the signal and individual vehicles.
“Using ITS with vehicle-to-vehicle and vehicle-to-infrastructure communications, the signal will know well in advance of the intersection when the vehicles will be arriving at the stop bar and even where they desire to go,” Taylor says. Such a system could incorporate GPS applications so that even bicyclists were detected when needed, and Taylor believes the result would be greater safety in addition to greater efficiency. “An intelligent system like this would help UDOT and other agencies achieve their goal of zero fatalities,” he says.
In the meantime, I have grown to appreciate those intersections that effectively detect vehicles and use the data they collect to intelligently change the light to red when it is absolutely necessary, and to maximize the green light as much as possible. There is a noticeable difference between timed intersections and those intersections that respond to real-time conditions, and while I’ll take my turn at the red light when I have to, I can’t help but admire the technology that works to keep traffic moving.
