How do traffic light sensors work?

Inspired by How can you not know that traffic lights have sensors?, I got to wondering how exactly traffic lights sense the presence of cars. I know there are a variety of methods, like cameras mounted above the light or sensors embedded into the road.

Can any fill me in on the exact methods these systems use to detect a car, especially if it isn’t situated in exactly the right spot?

Induction loops.

Basically, a big metal detector.

Further to this, can you explain why my bicycle is invisible to about 90% of them? It is certainly metal.

Because your bicycle has a LOT less metal than a car does. You can try getting off and laying your bike down on the sensor; this gets your bike’s metal frame closer to the sensor which increases the signal strength due to the inverse square law.

The induction loops are the most common. Sometimes there are mechanical pressure plates, which used to be more common - when I was learning to drive, many of the lights in my town had very obvious pressure plates you drove over. Both systems fail to trigger if a driver doesn’t pull all the way up to the stop line, as bitched about in the pit thread. Both can have problems with bicycles or motorcycles not triggering them. You bicycle may simply not be enough metal to be detected by the induction loop, or weigh enough to activate the pressure plate.

Or, if you have a really expensive bicycle, it might be made of a less magnetic alloy, and be even less able to change the magnetic field of the induction loop.


“Here Kitty, Kitty, Kitty.” ~ Erwin Schrodinger ~

Back around '79, or '80, my stepbrother was coming home from work on his small motorcycle. It was about 2-3 in the morning and he pulled up to a light where they had recently installed a sensor. He sat ther for a minute or two, until he finally decided the light wasn’t going to change, so he made his left turn only to find a cop right behind him. The cop listened to his story, told him the proper thing to do was make a right turn and then a U-ey, he then gave him a ticket for running a red light. We figured his bike just didn’t have enough mass to trigger the sensor.

Electrical conductivity is the important characteristic here. A given mass of aluminum will be more easily detected than an equal mass of iron, all other factors being equal.

The why do they call them magnetic detectors?


With bicycles, it can also be an issue that the rider using the edge of the lane doesn’t ride over the sensor. If the municipality has created bike lanes on a street, they should have thought about this, but if not, the sensor may be placed where a typical bicyclist doesn’t go over it.

The sensitivities can be adjusted, too, but the trade off is that if you start making the thing sensitive enough to pick up bicycles, you might start getting the light activated by vehicles turning onto the street and passing by the sensor in adjacent lanes.

An article on detection of bicycles at traffic lights:

That article also mentions that it is the conductivity of the metal body passing over the loop which matters, not that it be made of ferrous metal.

Inductive loops have been around for a long time. It’s a mature technology. Works well, but it has some disadvantages:

  • Expensive to install.
    E- xpensive to repair (road must be ripped up).
  • Disrupts traffic during repair.

Because of these, many intersections are using proximity detectors mounted to the overhead poles. They look like small cameras. Some measure the infrared energy emitted by vehicles, while others are video cameras used in conjunction with video recognition software. They’re compact, cheap to purchase, and cheap to install. If repairs are needed, disruption of traffic is kept to a minimum.


Some states have added laws that allow motorcyclists to proceed through the intersection when that happens.

Here’s a story on the subject. Unfortunately, Texas isn’t one of them, as I have that problem at about half the lights around here.

That’s why I mentioned the era. It was when these things were pretty new and I suspect they were still working out the bugs. It was in Scottsdale, Az. and, in those days, the area was not as developed as today.

I used a traffic loop detector as part of an industrial control system once (it was one of many things to determine whether the big object underneath the chute was in fact a truck). The controller’s sensitivity was adjustable. At its most sensitive, it could pick up the metal in my steel toed boots. If you really want to detect a bicycle, it can do it. For traffic applications, the sensitivity is intentionally turned down so that a motorcycle will trip it but a bicycle won’t. If the motorcycle didn’t trip it, then it was adjusted wrong.

There are also pressure plate type sensors mounted in the road. You don’t see these around too much any more but they used to be fairly common. They’re pretty obvious when you see one, since it has a strip of steel about a foot wide going across the road.

The inductive loop will pick up a car over a wider area. The pressure plate relies on a car approaching the light to run over it. You don’t have to stop exactly on it. Same with the inductive loop, as long as you pass over the loop the traffic controller knows you are there.

Another type is microwave transmitters, which work like radar. They are mounted up by the light and periodically send a radio wave down towards the road. If it gets an echo, it triggers the light. I’m not sure how often these are used. I haven’t seen any in my area. You get basically the same thing when you go into the grocery store. The automatic door opener (the little black box above the door) works on the same principle.

Older traffic light controllers were very simple devices. The timer was just a motorized wheel which spun around and closed different switch contacts. Modern controllers are little mini computers which are programmable. They can be programmed to change how they switch the lights at different times of day. They are usually programmed to periodically cycle the lights even if no one comes by, just in case a motorcycle does stop in front of the light and doesn’t manage to trigger the switch. They shouldn’t get stuck there forever. Sometimes the controllers can all talk to each other in sort of a network topology. This allows multiple lights to be synchronized, which is important if you are trying to control the flow of traffic in a busy downtown area, for example.

There used to be a hill in Pittsburgh which had a radar speed detector attached to the controllers. Late at night, if you went down the road under the speed limit, you’d get green lights all the way down. If you went over the speed limit, you’d hit red lights all the way down. I don’t know if it still operates that way or not.

This is really interesting. What is the logic behind this? If bicycles are supposed to obey traffic laws, including red and green lights, why set up a system to make that impossible?

Practicality. It’s not that they don’t specifically want to be able to detect a bicycle, but if a bicycle can set off the sensor loop, then so can a truck one (or even two) lanes over. This is fine for some intersections, but not for others.

Because if the sensitivity was tuned high enough to detect a bicycle, it might also detect a semi driving through the cross street or in the opposing lane of traffic.

Anecdotally, I have discovered that a 4’ length of 3/4" copper piping laid directly over the loop will trigger it. That’s how I tested the parking gates at a former job.

ETA - What QED said.

Been in law enforcement 25 years and I’d like to ask that dick with a badge what he would expect someone to do at a freeway entrance ramp? My big ol’ Harley won’t trip the light at about 90% of acceleration ramps. Should one sit there for 3 hours until the timer shuts off after rush hour?

So the answer is simple all bicyclists must carry a 4 foot piece of pipe with them. I’m sure it would seem like a logical solution to our legislators.