Great video, as always, Nick. My neighbor and I requested a traffic study on our street trying to get the city to install a stop sign instead of a yield at the intersection in front of our houses. They used this device and sent us copies of the full traffic report. They didn't provide us vehicle types but we got all kinds of vehicle numbers and velocity data broken down by hour of the day. Really interesting.
Very cool, how specific did they get (wheel base/vehicle type?) Everything I talked about here I had to find on manufacture websites and patents, I've never seen the real thing.
These are used in the UK a lot. There are permanent traffic monitoring devices installed all around heavily populated areas that work on induction loops put under the surface of the road. You can see the cuts in the surface where the wires have been laid that also trail off to the side of the road to a small box, or lid to a device underground. These tube versions are very temporary and go to devices chained to lamp posts usually with small but thick steel containers and padlock.
These popped up a couple times on a road on my commute for a few days over the course of about a month, but I couldn't figure out what they were for. Thanks for the answer!
I might have missed this part, but can you explain how you can tell the difference between one car going fast with a long wheel base, and another car going slow with a short wheel base?
When the first axle goes over the sensors you can figure out it's speed, that lets you tell if it is a fast moving car that is long or a slow moving car that is short. Hope that helped.
Do the hoses use hydraulic pressure to activate switches in the box or a conductor wire in the hose itself? If using pressure how do they deal with the speed of sound when monitoring a large multi lane road (as there will be a limit on how fast the fluid can re-stabilize between cars)? If electric conductors how do they get them to not break down under the weight of constant commercial traffic?
I have never had my hands on a real traffic monitor so I am not sure but some one else here mentioned that there may be electronic sensor strips that can deal with signals from multiple lanes (and wouldn't suffer from propagation times). Very interesting idea, I hadn't thought about the rebound time of the system.
How do these work when the road has multiple lanes in the same direction? Does each lane count as its own sensor or is the entire length of the cord one big sensing element?
I wonder whether it is also possible to distinguish two cars passing by at roughly the same time, such that one of the tires of the other car passes one of the sensors in between the front and back tires of the first car. Especially when these sensors are placed on bicycle lanes.
I don't know, with smart enough software it might be possible to distinguish between passing cars provided they are both going at different speeds and are not exactly the same length, even then it is possible to generate confusing data. Elsewhere in the comments someone mentioned that there are sensor strips that are able to tell from which side of the road the signal is coming.
I think if you assume the vehicles being recorded have 4 wheels each, you should be able to differentiate 2 vehicles going in different directions because a vehicle's wheel would only ever be on 1 pipe at a time because of their spacing. If 2 vehicles' wheels where on a pipe at the exact same time, the second pipe would receive a different timing. A bit of extrapolation with the common types of vehicles and speeds should be able to clear this up. I imagine it's the same kind of extrapolation that speed cameras use using the doppler sensor, a single camera would appear to cover both lanes but ALWAYS flashes from behind (in the UK) so it doesn't blind you, and motorbike number plates are only on the rear. So the cameras would have to split apart traffic going which way and their speed using many tens/hundreds of samples per second.
you can't know how many axles a vehicle has. they have to make an assumption that a set of pulses is either a truck pulling a double axle trailer, or two cars close together. of if you want to say that a vehicle is three axles, then you have to make an assumption on timing. in other words, you have to analyze the data and group sets of pulses together based on gaps between them, and and an odd number of pulses probably means an odd number of axles, but it can also mean you have grouped your data incorrectly. Also these system totally falls apart when traffic is at a crawl. you have no hope for counting axles on an individual set of pulses if traffic is very slow. UNLESS you area also measuring the weight of the vehicle in your pulse. then you might be able to suss out a multi axles semi truck from a few closely driving cars or motor cycles.
I am curious to know if they can discern data from "confuse scenarios" such as two cars passing by each other over the detector... I guess there can be some multiple results on those cases. Also, I was wondering if the sensors could also measure the pressure over the hose. Seems like the cheapest sensor would be hidraulic, thus you could collect yet another type of info, I guess...
DoNotPushHere yeah that's what I never understood. They go over the WHOLE road here when I see them. How are they differentiating between the opposite directions of traffic? Speed information would seemingly be impossible to determine
10mintwo my only guess is that the data collected is filtered. Since they do not intend to track individuals, I think they might discard any conflictive data...
I'm sure that you will get some bad data from cars passing at the same time, but when you're looking at datasets that could easily be over 1000 cars an hour, it's not a big problem. I can't find any specific evidence, but I think the cable they use for the detector might be a thing called quantum tunnelling composite.
That's quite a surprise for me. I thought those hoses would rely on pure fluids, due to them being placed on...the street... open to the environment and any harsh conditions... So it's sort of a long array of pressure sensors?
DoNotPushHere: The hoses are just full of air. One end is sealed, the other has a pressure operated switch. When a tire squeezes (squashes) the tube, that creates enough of an increase in air pressure to operate the switch.
55Ramius just a guess, but a simple circuit like one based on a 555 could generate a continuous square wave. You must make sure it's attenuated and roughly matches the input impedance of your device before plugging it into a sound card, though.
I have the headphone output going to each switch (playing a continuous tone), the normally open contact of each switch goes to the left and right microphone inputs giving me 2 inputs.
I heard that in the days of the Commodore 64, people would connect directly to the pins in the parallel/serial ports to make their own sensors and collect data, but I have never seen anything like it done with modern PCs. Fascinating! Any chance of a video on that topic?
I had a Radio Shack color computer back in the 80's and tapped into the joystick cable to use as a sensing device. Photocell light meter thing. Could measure speed of a fan by holding photocell on one side and a light source on the other. Lots of uses for that joystick connect. But one day I tried feeding sound out, back into sound in. Shorted computer out and had to have it repaired. Never tried that again with those computers or a pc. I just bought a $2600 pc and I sure don't want to risk frying the sound card or worse. ps- still have coco2 and coco3 computers. Stored away for long time but want to mess with them again soon.
I've built a quick and dirty capacitive coupling to audio input and used Audacity as a very basic oscilloscope. Then I just bought an oscilloscope, but making one for 50 cents is hard to beat.
Belgian here. We see these all the time when there are new circulation plans upcoming. Seeing a large number of these over the city almost always means that soon enough there will be a new circulation plan, yet again (these happen every decade or so). I've always wondered a few things about these, though. How do they detect? I always assumed that it's pressure based. If so, can't you also try to calculate the weight of the vehicle passing by, or is the range in which the pressure changes just too small to have accurate results? Further, I've seen them on bicycle paths as well here. Are those different, as in they have different pressure? And lastly, what if they are spanned over multiple lanes, or even over bicycle lanes in the same direction? Is it possible to determine the lane the vehicle drove on? How accurate is it when two vehicles pass by (nearly) simultaneously? You almost exclusively see them going diagonally across here. That way, there's data for all four tires, not just the two axles! Well, that's assuming you could see the pressure difference between one and two tires in the first place. But then again, more length means a higher probability on simultaneous passings, so they must be able to detect multiple tires, right? (Edit: a word)
The diagonal method is perfect for removing ambiguity in the signals, I wish I had known about that earlier >_< As far as I know the ones in use here only detect presence or absence of a tire but elsewhere in the comments someone mentioned that there are sensor strips that are able to tell from which lane the signal is coming. With an air pressure sensor it might still be possible to tell how far along the car is, a car close to the sensor would produce a higher pressure spike than one at the far end of the tube but I don't know how accurate it could be.
You just found the fatal flaw of hose counters. They are really inaccurate on multi lane roads. As for seeing them about every 10 years in Belgium. I don't know your country's laws but in the united states its required every 10 years minimum that every agency that manages roads is required by the federal government to do traffic studies on every single road that they manage and send the studies to the F.H.A. (federal highway administration) or they can pull any funding that they are supporting.
They work on a per-lane basis so two cars at once isn't really an issue. For a large truck there would be an identifiable pattern, one single axle, a pair of axles for the cab and then one more pair for the trailer.
They're fairly regular in the Netherlands, I'd be surprised if they're not used in Germany. They're often temporary for traffic research, with a box on the side.
+Justin Koenig Interesting, they are used in Canada, and the US regularly, I've also seen them used in New Zealand. Do you know how traffic monitoring is done in Germany?
It's Germany, they just /know/ what their traffic is doing. Anything less would be inefficient. :) (Also, they are commonly used in Australia, they are being taken over by monitoring cameras though.)
This video explains how the data is used, but doesn't do what the title says - I still don't know how these things actually work. Are they air filled tubes? Some kind of sensor on the end?
I have never had my hands on a real one but the data crunching methods I showed would work for a simple hose + pressure switch as well as it would for a more complicated lane discerning sensor. Sorry I couldn't get my hands on a real one for this video :/
@@sgartner it doesn't. Still leaves you guessing. I had a good look at one in my neighborhood and i am pretty convinced it is hydraulic pressure: the end of the tube looks like a simple clamped hose.
If you ever see a metal plate embedded into the gutter wit 2 windows cut into it, it will probably be one of these. en.wikipedia.org/wiki/TIRTL It operates in a similar manner to the rubber hoses but uses IR beams instead.
My original idea for this video was to make a beam based sensor like the one you showed but I couldn't think of a way to make it both cheap and simple enough to set up on a busy road. Very cool, thanks!
In UK, there's a lot of buried induction loops for monitoring traffic - here's a perfect example: goo.gl/maps/SqeHATgxvLz - you see it goes to a box on the side of the road where data gets stored, maybe even remotely accessed.
We have a lot of those in Melbourne as well, especially on our highways. On some of our tollways, we have another type of induction system that uses 'studs', wireless transponders about the size of a hockey puck. They are buried in pairs for each lane, a couple of metres apart and are powered by a long life lithium battery that lasts for about 4 years. They transmit speed data in real time to a transceiver mounted on a nearby pole which then goes back to the control centre to give live traffic info. You can see them here in the middle of the lanes. They just look like patches in the road but the giveaway is how uniform they are. www.google.com.au/maps/@-37.8469599,145.0408599,3a,75y,174.51h,86.18t/data=!3m6!1e1!3m4!1sE7i3fZn6vM2wr1FQtzKPdA!2e0!7i13312!8i6656?hl=en
Wow, those are really inconspicuous. How do they get any feedback from induction when they're that small? Doesn't look like they'd detect motorcycles very well. I was driving at work earlier and found a complicated induction loop setup with a 4x4 arrangement, seems a bit overkill - goo.gl/maps/MC9cXp1nFJ12
Great video, as always, Nick.
My neighbor and I requested a traffic study on our street trying to get the city to install a stop sign instead of a yield at the intersection in front of our houses. They used this device and sent us copies of the full traffic report. They didn't provide us vehicle types but we got all kinds of vehicle numbers and velocity data broken down by hour of the day. Really interesting.
Very cool, how specific did they get (wheel base/vehicle type?) Everything I talked about here I had to find on manufacture websites and patents, I've never seen the real thing.
These are used in the UK a lot. There are permanent traffic monitoring devices installed all around heavily populated areas that work on induction loops put under the surface of the road. You can see the cuts in the surface where the wires have been laid that also trail off to the side of the road to a small box, or lid to a device underground. These tube versions are very temporary and go to devices chained to lamp posts usually with small but thick steel containers and padlock.
if you use analog readings instead of switches you may try to distinguish between signals from different vehicles that are too close to each other.
These popped up a couple times on a road on my commute for a few days over the course of about a month, but I couldn't figure out what they were for. Thanks for the answer!
Glad it was helpful!
We are for the same projects using Piezo electric sensor and loop based Vehicle classification free flow
thank you so much !!!!
excellent !
I might have missed this part, but can you explain how you can tell the difference between one car going fast with a long wheel base, and another car going slow with a short wheel base?
When the first axle goes over the sensors you can figure out it's speed, that lets you tell if it is a fast moving car that is long or a slow moving car that is short. Hope that helped.
Do the hoses use hydraulic pressure to activate switches in the box or a conductor wire in the hose itself? If using pressure how do they deal with the speed of sound when monitoring a large multi lane road (as there will be a limit on how fast the fluid can re-stabilize between cars)? If electric conductors how do they get them to not break down under the weight of constant commercial traffic?
I have never had my hands on a real traffic monitor so I am not sure but some one else here mentioned that there may be electronic sensor strips that can deal with signals from multiple lanes (and wouldn't suffer from propagation times). Very interesting idea, I hadn't thought about the rebound time of the system.
How do these work when the road has multiple lanes in the same direction? Does each lane count as its own sensor or is the entire length of the cord one big sensing element?
Edward Ciotic I've seen ones with extra ports for hoses that cover less than the full width of the road.
I wonder whether it is also possible to distinguish two cars passing by at roughly the same time, such that one of the tires of the other car passes one of the sensors in between the front and back tires of the first car. Especially when these sensors are placed on bicycle lanes.
I don't know, with smart enough software it might be possible to distinguish between passing cars provided they are both going at different speeds and are not exactly the same length, even then it is possible to generate confusing data. Elsewhere in the comments someone mentioned that there are sensor strips that are able to tell from which side of the road the signal is coming.
I think if you assume the vehicles being recorded have 4 wheels each, you should be able to differentiate 2 vehicles going in different directions because a vehicle's wheel would only ever be on 1 pipe at a time because of their spacing. If 2 vehicles' wheels where on a pipe at the exact same time, the second pipe would receive a different timing. A bit of extrapolation with the common types of vehicles and speeds should be able to clear this up. I imagine it's the same kind of extrapolation that speed cameras use using the doppler sensor, a single camera would appear to cover both lanes but ALWAYS flashes from behind (in the UK) so it doesn't blind you, and motorbike number plates are only on the rear. So the cameras would have to split apart traffic going which way and their speed using many tens/hundreds of samples per second.
Really interesting!
Thanks!
you can't know how many axles a vehicle has. they have to make an assumption that a set of pulses is either a truck pulling a double axle trailer, or two cars close together. of if you want to say that a vehicle is three axles, then you have to make an assumption on timing. in other words, you have to analyze the data and group sets of pulses together based on gaps between them, and and an odd number of pulses probably means an odd number of axles, but it can also mean you have grouped your data incorrectly. Also these system totally falls apart when traffic is at a crawl. you have no hope for counting axles on an individual set of pulses if traffic is very slow.
UNLESS you area also measuring the weight of the vehicle in your pulse. then you might be able to suss out a multi axles semi truck from a few closely driving cars or motor cycles.
I am curious to know if they can discern data from "confuse scenarios" such as two cars passing by each other over the detector... I guess there can be some multiple results on those cases. Also, I was wondering if the sensors could also measure the pressure over the hose. Seems like the cheapest sensor would be hidraulic, thus you could collect yet another type of info, I guess...
DoNotPushHere yeah that's what I never understood. They go over the WHOLE road here when I see them. How are they differentiating between the opposite directions of traffic? Speed information would seemingly be impossible to determine
10mintwo my only guess is that the data collected is filtered. Since they do not intend to track individuals, I think they might discard any conflictive data...
I'm sure that you will get some bad data from cars passing at the same time, but when you're looking at datasets that could easily be over 1000 cars an hour, it's not a big problem.
I can't find any specific evidence, but I think the cable they use for the detector might be a thing called quantum tunnelling composite.
That's quite a surprise for me. I thought those hoses would rely on pure fluids, due to them being placed on...the street... open to the environment and any harsh conditions... So it's sort of a long array of pressure sensors?
DoNotPushHere: The hoses are just full of air.
One end is sealed, the other has a pressure operated switch.
When a tire squeezes (squashes) the tube, that creates enough of an increase in air pressure to operate the switch.
Could you detail a bit on how you connected to computer to use audacity? If audio port, just what is happening when switch closes? Isn't that a short?
55Ramius just a guess, but a simple circuit like one based on a 555 could generate a continuous square wave. You must make sure it's attenuated and roughly matches the input impedance of your device before plugging it into a sound card, though.
I have the headphone output going to each switch (playing a continuous tone), the normally open contact of each switch goes to the left and right microphone inputs giving me 2 inputs.
I heard that in the days of the Commodore 64, people would connect directly to the pins in the parallel/serial ports to make their own sensors and collect data, but I have never seen anything like it done with modern PCs. Fascinating! Any chance of a video on that topic?
I had a Radio Shack color computer back in the 80's and tapped into the joystick cable to use as a sensing device. Photocell light meter thing. Could measure speed of a fan by holding photocell on one side and a light source on the other. Lots of uses for that joystick connect. But one day I tried feeding sound out, back into sound in. Shorted computer out and had to have it repaired. Never tried that again with those computers or a pc. I just bought a $2600 pc and I sure don't want to risk frying the sound card or worse. ps- still have coco2 and coco3 computers. Stored away for long time but want to mess with them again soon.
I've built a quick and dirty capacitive coupling to audio input and used Audacity as a very basic oscilloscope. Then I just bought an oscilloscope, but making one for 50 cents is hard to beat.
what is the pressure applied to the hose to actuate the switch?
Minimal. You can test the counter by hitting the hose with a hammer.
Belgian here. We see these all the time when there are new circulation plans upcoming. Seeing a large number of these over the city almost always means that soon enough there will be a new circulation plan, yet again (these happen every decade or so).
I've always wondered a few things about these, though. How do they detect? I always assumed that it's pressure based. If so, can't you also try to calculate the weight of the vehicle passing by, or is the range in which the pressure changes just too small to have accurate results? Further, I've seen them on bicycle paths as well here. Are those different, as in they have different pressure? And lastly, what if they are spanned over multiple lanes, or even over bicycle lanes in the same direction? Is it possible to determine the lane the vehicle drove on? How accurate is it when two vehicles pass by (nearly) simultaneously?
You almost exclusively see them going diagonally across here. That way, there's data for all four tires, not just the two axles! Well, that's assuming you could see the pressure difference between one and two tires in the first place. But then again, more length means a higher probability on simultaneous passings, so they must be able to detect multiple tires, right?
(Edit: a word)
The diagonal method is perfect for removing ambiguity in the signals, I wish I had known about that earlier >_<
As far as I know the ones in use here only detect presence or absence of a tire but elsewhere in the comments someone mentioned that there are sensor strips that are able to tell from which lane the signal is coming.
With an air pressure sensor it might still be possible to tell how far along the car is, a car close to the sensor would produce a higher pressure spike than one at the far end of the tube but I don't know how accurate it could be.
You just found the fatal flaw of hose counters. They are really inaccurate on multi lane roads. As for seeing them about every 10 years in Belgium. I don't know your country's laws but in the united states its required every 10 years minimum that every agency that manages roads is required by the federal government to do traffic studies on every single road that they manage and send the studies to the F.H.A. (federal highway administration) or they can pull any funding that they are supporting.
Someone else mentioned that there are multi-lane hoses that have sections to distinguish traffic. I imagine they wouldn't be too hard to make.
I don't think I've ever seen one of those in Turkey before. Are they expensive to set up and operate?
+Misclickable I'm not sure how expensive they are, they are only used here when planning changes to the local roads or traffic signals.
actually i've seen them couple of times in Turkey. i'm sure now you will see because you know what are they looks like :)
Where have you seen them? Maybe I've seen them on highways, but didn't notice they were used for this purpose.
How does it know overtaking? Or 16 wheeler trucks?
They work on a per-lane basis so two cars at once isn't really an issue. For a large truck there would be an identifiable pattern, one single axle, a pair of axles for the cab and then one more pair for the trailer.
never seen these in germany
ever
They're fairly regular in the Netherlands, I'd be surprised if they're not used in Germany. They're often temporary for traffic research, with a box on the side.
+Justin Koenig Interesting, they are used in Canada, and the US regularly, I've also seen them used in New Zealand.
Do you know how traffic monitoring is done in Germany?
Pretty standard in the UK too, at least Cambridge
It's Germany, they just /know/ what their traffic is doing. Anything less would be inefficient. :)
(Also, they are commonly used in Australia, they are being taken over by monitoring cameras though.)
In germany, they are embedded into the road
This video explains how the data is used, but doesn't do what the title says - I still don't know how these things actually work. Are they air filled tubes? Some kind of sensor on the end?
I have never had my hands on a real one but the data crunching methods I showed would work for a simple hose + pressure switch as well as it would for a more complicated lane discerning sensor. Sorry I couldn't get my hands on a real one for this video :/
Here's a video explaining how they work. It's air.
czcams.com/video/Y4RUnJ0EiFk/video.html
@@sgartner it doesn't. Still leaves you guessing. I had a good look at one in my neighborhood and i am pretty convinced it is hydraulic pressure: the end of the tube looks like a simple clamped hose.
If you ever see a metal plate embedded into the gutter wit 2 windows cut into it, it will probably be one of these. en.wikipedia.org/wiki/TIRTL It operates in a similar manner to the rubber hoses but uses IR beams instead.
My original idea for this video was to make a beam based sensor like the one you showed but I couldn't think of a way to make it both cheap and simple enough to set up on a busy road. Very cool, thanks!
In UK, there's a lot of buried induction loops for monitoring traffic - here's a perfect example: goo.gl/maps/SqeHATgxvLz - you see it goes to a box on the side of the road where data gets stored, maybe even remotely accessed.
We have a lot of those in Melbourne as well, especially on our highways. On some of our tollways, we have another type of induction system that uses 'studs', wireless transponders about the size of a hockey puck. They are buried in pairs for each lane, a couple of metres apart and are powered by a long life lithium battery that lasts for about 4 years. They transmit speed data in real time to a transceiver mounted on a nearby pole which then goes back to the control centre to give live traffic info.
You can see them here in the middle of the lanes. They just look like patches in the road but the giveaway is how uniform they are.
www.google.com.au/maps/@-37.8469599,145.0408599,3a,75y,174.51h,86.18t/data=!3m6!1e1!3m4!1sE7i3fZn6vM2wr1FQtzKPdA!2e0!7i13312!8i6656?hl=en
Wow, those are really inconspicuous. How do they get any feedback from induction when they're that small? Doesn't look like they'd detect motorcycles very well. I was driving at work earlier and found a complicated induction loop setup with a 4x4 arrangement, seems a bit overkill - goo.gl/maps/MC9cXp1nFJ12
Your units are wrong, the wheel bases are 4, 7 and 9 respectively
foobargorch lol, almost. they are 3, 6 and 8.
Ha! It took me until now to realize what you two were talking about >_< I even checked my math to make sure I hadn't messed it up somehow.