There is an area near every signalized intersection in the world that has caused daily dilemmas for both drivers and transportation agencies alike. Traffic engineers have long recognized the dangers in the “dilemma zone,” the area of an intersection approach where incoming drivers facing a yellow light have just a moment to decide whether to proceed or to stop; if made incorrectly, this decision can lead to greater risk of right-angle or rear-end collisions and higher rates of red light running. Protecting drivers from this dilemma is a challenge, because for many years, traffic engineers were forced to compromise — they could either make intersections safe or efficient, but they could not do both.
SmartSensor Advance is revolutionizing the way traffic professionals look at the dilemma zone. The sensor uses radar to detect passenger vehicles up to 600 feet away from an intersection, and semi-trucks and other high profile vehicles up to 900 feet away. It then continuously tracks each vehicle, monitoring its speed and range to determine its estimated time of arrival at the intersection stop bar.
Wavetronix refers to this method of dilemma zone protection as dynamic ETA tracking. Unlike loops and other detectors that make a single detection at a fixed point and then assume conditions on the rest of the approach, SmartSensor Advance provides dilemma zone protection only to those vehicles which need it. As a result, Advance helps make intersections much safer without negatively affecting their efficiency.
The idea of dynamic ETA tracking is a relatively new one in traffic, and understanding how it works can be difficult, particularly when compared with the relative simplicity of loops, which have been the standard for dilemma zone detection and protection for decades. But understanding dynamic ETA tracking might be the key to improving intersection safety, and the answer to the question, “How do I make intersections both safer and more efficient?”
In dilemma zone situations, drivers have a split-second’s time to decide the safest — or perhaps least risky — course of action: run the risk of a rear-end collision by stopping; or a red light violation, even a high-speed collision, by continuing. Because driver behavior in these situations is impossible to predict, many systems simply provide dilemma zone protection based on the posted speed limit and the amount of yellow light time provided to drivers.
But vehicles that exceed the speed limit (and/or face shorter-than-recommended yellow times) may have literally no safe dilemma zone option at all; they are travelling too fast to stop safely, but don’t have enough yellow time to make it through the intersection either. The existence of red light cameras prepared to snap pictures of offending license plates influences driver decision-making adversely, causing some vehicles to hit the brakes in situations where continuing onward represented a far safer outcome.
Modern controllers can adjust and extend green time to allow for passage of cars in the dilemma zone — “dilemma zone protection” — although the trick is to identify which vehicles need protection in the first place. Installing dilemma zone protection may increase intersection safety and efficiency; however, ineffective dilemma zone protection takes away many of those safety and efficiency gains. Proper dilemma zone management requires an effective use of detection and analytics to identify and protect selected vehicles without causing unnecessary delays or increasing the likelihood of accidents.
For many years, traffic engineers were forced to compromise – they could either make intersections safe or efficient, but they could not do both.
Traditional dilemma zone protection has relied on inductive loops, triggered some distance away from the stop bar with notifications about vehicle presence and current speed. Since loop locations are fixed, traffic engineers needed to decide when to protect detected vehicles after they pass the loop location, especially given the fact that vehicle speeds will vary. One common method is to use 85th percentile and 15th percentile speed data as the upper and lower bounds for vehicle speeds.
There are three problems with dilemma zone protection using fixed loops. First, by design, selecting a statistical range of speeds to protect (whether 15th–85th percentile or some other range) excludes a non-zero number of outliers — vehicles who will receive no dilemma zone protection at all. Expanding the speed range to include some or all of those outliers only exacerbates the efficiency problems outlined below.
Second, vehicles will almost always change their speed — possibly accelerating or decelerating — as they approach the intersection, so original speed values recorded by loops at a given location become outdated almost immediately. Indeed, one group of sample loop sites was shown to provide accurate dilemma zone protection only 57 percent of the time due to changes in vehicle speed after passing the loop area. Other detection devices that also make a one-time detection at a specific location on the approach share this problem.
One group of sample loop sites was shown to provide accurate dilemma zone protection only 57 percent of the time due to changes in vehicle speed after passing the loop area.
Third, the attempt to be inclusive and cover vehicles travelling at different velocities leads to an overprotection problem that can greatly impact efficiency (see sidebar). In order to protect all vehicles, traffic engineers create a large dilemma zone that tries to protect both fast and slow-moving vehicles, but this creates situations where cars at certain speeds within the zone don’t require protection — the drivers face no dilemma because they can either stop or go safely. But because these vehicles are detected within the prescribed zone, a call to the controller to extend the green time is made even though it isn’t warranted.
Protecting vehicles that don’t need it results in hits to both efficiency and safety: efficiency suffers because the oversized dilemma zone sends unnecessary commands to the controller and adjusts the phase length when traffic does not require it, leaving side-streets waiting; it also causes unsafe conditions because it prevents the effective identification of gap-out opportunities between vehicles, where the controller can safely switch phases without leaving any vehicles in difficult situations. Fewer recognizable gap-out opportunities means controllers will ‘max out’ more often, forcing a phase switch whether cars are in dilemma zone situations or not. Inelegant phase switches may end up increasing the likelihood of accidents by leaving certain vehicles with difficult dilemma zone decisions, even in a system meant to avoid them, simply because the false protection calls took up all the available phase time.
Added together, these problems indicate traditional dilemma zone protection can create issues with safety and efficiency: some cars that need protection don’t get it; and some cars that don’t need it due to their speed receive it anyway.
To solve these problems, let’s look at what the key metric for dilemma zone protection should be. Since the available resource to control is expressed in time — remaining green time + yellow time — vehicle analysis should also be done using time: how many seconds will it take for each vehicle to arrive at the intersection, based on their current location and speed?
In reality, each vehicle has their own individual dilemma zone based on their unique speed and distance. Therefore, the key to providing accurate dilemma zone protection lies in recognizing individual dilemma zones without generalizations which simultaneously over- and under-protect different vehicles. Tracking how long vehicles will take to arrive at the intersection at their current speeds allows technicians to create a suitable dilemma zone range — say, three to five seconds — and then constantly compare vehicle ETA to that figure, taking into account speed changes as vehicles approach the intersection.
ETA is the most direct and useful metric for considering dilemma zone behavior. From the 2006 research paper Analysis of Dilemma Zone Driver Behavior at Signalized Intersections (Gates, Tim J., Noyce, David A., and Laracuente, Luis; University of Wisconsin–Madison), the authors note: “…the estimated travel time to the intersection at the start of the yellow interval was found to have, by far, the strongest effect on a driver’s likelihood to stop versus go through the intersection.”
Providing dilemma zone protection using ETA rather than speed or distance reflects the reality of vehicular travel: the zone in which faster vehicles should be protected is farther away from the intersection than slower vehicles because of the difference in speed.
Guesswork based on point speeds from a fixed location won’t take into account changes in speed, unless the ETA is being constantly updated at the same time.
Using dynamic ETA tracking, advance detection can protect the dilemma zone for cars travelling at any speed, not just at “typical” speeds in the middle of arbitrarily-chosen percentage boundaries. Each vehicle can have its individual dilemma zone calculated on the fly based on their expected arrival at the intersection — regardless of speed and location — and that value is updated as the vehicle changes speed throughout the detection zone.
Dynamic ETA tracking provides more accurate dilemma zone protection while also preventing the overprotection problem listed above. Each vehicle is protected only for the ETA boundaries that require it; otherwise, it is ignored, leading to more identifiable gap-out opportunities between vehicles. Intersections become both safer and more efficient; although dynamic ETA tracking won’t eliminate the tough dilemma zone decisions drivers may face, it will give drivers in the dilemma zone that extra second of travel to make the decision easier without disrupting the efficiency on side-streets or causing awkward max-out phase changes that can leave vehicles in difficult situations anyway.
