This is new?
The traffic sensing bit is ancient - they had those black rubber bars across the road to sense traffic movement near junctions decades ago.
A Swiss traffic management and transport economics expert believes a combination of queue management and computing can help solve the gridlock that plagues the modern city. Dr Dirk Helbing of ETH Zurich, a professor of sociology specializing in modeling and simulation, says “self organizing” traffic control systems, using …
The traffic sensing bit is ancient - they had those black rubber bars across the road to sense traffic movement near junctions decades ago.
> they had those black rubber bars across the road
The black rubber tubes are actually a census point. They are connected to a box chained to a pole somewhere close, and are used for gathering information about road use for planning purposes - they can determine class of vehicle, speed, wheelbase, etc.
But you're right that VA junctions are old news - only the sensors are predominantly buried into the road.
The temporary rubber tubes placed on the surface and hooked up to a recorder are indeed census points, but I don't think that's what the post referred to. Before the buried inductive loop system was used, traffic lights were controlled by pneumatic sensors. These were metal tubes with rubber tops, buried in the road just before the lights. As a car arrived at the lights it would signal the local controller, which could then trigger a change.
I've no idea if they survived the transition to centralized computer control, or if that had inductive sensors from the beginning.
A fascinating plunge into traffic sensors!
These days, the primary detection method is inductive loops. A long time ago, rubber bars with a hose in, a more permenant version of the sensus point, were popular.
These days, more advanced sensors are getting common, for instance radar sensors that can monitor multiple lanes without having to replace on resurface. Also infra red sensors for more accurate vehical profiling, height etc.
Another recent inovation are little wireless capsules, containing an electronic compas, that detect car type/speed/count based on the vehicals disruption of the earths magnetic field. The highways agency trialed them a couple years ago, and found them much faster than loops to fit, but battery life in field remains to be proven.
Another exotic technology is multi microphone sensors which use traffic noise to give a car, lane position, speed etc. Less accurate than loops but cheaper.
The UK uses automatic traffic control in cities. The individual lights all report back to a central computer which optimises traffic flow in quite a deterministic way.
It is much cheaper to link the lights to a central station because you can just use BT phone lines.
Of course, you could do self organising by linking the lights via a central hub at the same cost, but if you do that, you lose the redundancy anyway, so there doesn't seem much point.
IMHO, real progress could be made by harnessing data from peoples GPS devices to let you know where they were trying to go. Perhaps you could send instructions back to the GPS, routing people evenly accross varius possible routes. If this happens, it will come from the GPS/mapping industry, not the traffic lights people.
"The UK uses automatic traffic control in cities. The individual lights all report back to a central computer which optimises traffic flow in quite a deterministic way."
...unless a politician determines that the traffic flow will be slower than a donkey with three gout-ridden legs and a wooden pegleg, of course.
Remember the day the Congestion Charge came in? Now compare it to how it is currently.
Or when Red Ken order traffic lights be added to Hyde Park Corner to totally bu&^er up the flow of traffic.
Take 100 cars on a stretch of road in a traffic jam
Easily remove100 cars from road using this formula,
Litre of fuel £2.50
Now there's no traffic!
When might this happen, probably next year.
Yes - because there is REALLY a viable alternative to cars for every situation.
Not every city has London's public transport system.
I agree, on the principle the tax payers will pay for the extra 2 - 3 hours a day I lose by using train or bike and the 1/2 day I'd lose using the bus.
Oh sorry, did I mention I don't live in a city?
If you don't live in a city, then presumably traffic jams are not an issue.
Tell that to the M1
I had a right go at TfL for a junction near Hyde Park Corner (Brompton Rd/South Ken Rd) recently - the three sets of lights there are out of sync and you can only get two cars through at a time.
The response I got is that they are actively managed and its working perfectly.
So apparently it is intentional .. ??!!
> it is intentional
When the traffic industry says "managed", don't take that to mean "does something I consider useful".
Traffic management is about affecting driver behaviour. The actual intent might be different to what you or I would consider sane. That intent might or might not actually do some good somewhere.
Was I the only child who used to jump on the strips to try & get the lights to change? (Traffic was MUCH lighter in those days! :)
The sensors always controlled the light as well as collecting data. People used to reverse over them to make the lights change green.
Many in Australia don't use phone lines. They have a tiny solar panel for power and a wireless internet connection to the central control. Cameras monitor major roads & intersections. Human beings can change the lights for miles around, to push traffic around trouble spots.
> Using sensors to measure variables like the amount of traffic already in
> a road section, how quickly it is moving, and how long to the next change
> of lights, Dr Helbing’s approach is designed to respond to unexpected events.
Traffic systems have been doing that for decades.
If you look at an approach to a set of lights, there will generally be two sensors embedded in the road - either a diamond or a rhombus shape, depending on which manufacturer put them in. There may also be IR sensors on top of the signal heads.
Each junction has its own local controller, and is also networked to all the others in the area.
I've not been near the traffic industry for about 15 years, but all this was established technology when I was doing it...
Get the minimum timings wrong and they can still go pear-shaped. My village has two sets of lights networked together. In the ground are the diamond-sensors, and what looks like IR cameras on top of at least one of the sets of lights.
The first set of lights controls the direction of traffic over a single-lane bridge - no room for two lanes of traffic, so it's uni-directional. About 50m after the bridge is the central-village T-junction traffic lights. The problem is that folks coming over the bridge cannot see if there is space for them to cross, so if traffic is backed up at the T-junction, then the queue to get through that can back up onto the bridge so that no-one can leave town. Unfortunately, one cycle of that means that nothing moves to leave town, so the lights assume nothing is trying to leave town (probably thinking everyone is parked, rather than queuing), so then fix that light to red - therefore no-one can leave town and the queue backs up on this reasonably busy route to elsewhere. The only way out of this scenario is for someone to run the red light over the bridge which is rather unsafe as they won't know if someone is coming at them at high speed (have seen the occasional good spirited soul get out of their car further back in the queue, run up to the bridge to stop traffic coming in to let some traffic out safely over the red so that there can be movement over the sensors to reset the system). This situation happens reliably at around 7.25am Mon-Fri, despite emails to my local controller.
Another flaw is that the induction sensors are sometimes not sensitive to pick up a bicycle and the "camera" sensors occasionally even miss a motorbike meaning two-wheel users end up stuck at the lights until eventually a car pulls up behind them (and, yes, when cycling I *DO* stop at red lights, ok?!)
I've had this on a bike too. Once I even had it in a Nissan Micra. Had to move back and forward until the sodding thing detected me!
Living in Watford, as I do, it's occurred to me that a set of traffic lights at four or five main entrances, to stop too much traffic in the town centre, then all the other lights could be removed, and all the other junctions turned into mini roundabouts, and there'd basically be no jams nor congestion..
The fuel savings alone would be massive.
Not all towns are the same, but this one is, the rule being "Let the traffic out as quick as you can, and stop it coming in until there's a free ride."
I'm surprised no-one's tried it.
Watford is an interesting place. The main loop around the centre has some interesting queuing. Late at night all the approach roads to the central loop (which is really a huge roundabout going all the way around the main shopping centre) are red and the loop itself is green. When you approach the traffic lights coming on, they switch to green if no traffic is coming the other way, let you onto the main loop, then you can drive around as far as you want, usually with greens all the way. As traffic gets busier, the main loop changes mode. All the entrance move to green in order, letting traffic on one stop to queue up, then letting the entire ring make progress. You normally have to stop once waiting to get on the loop, once again waiting for the loop to move, then often get most of the way around the loop in a single move. As traffic gets busier still, the control of traffic getting onto the loop tries to keep the jams out of the loop and keep the loop moving. In general it worked very well for the year that I lived there.
It's occurred to others too and works very well around Maastricht.
Most times lights are controlled by traffic sensors. During rush hour times and other periods of heavy traffic, the peripheral lights hold traffic back and some roads have unidirectional lights installed purely for flow control activated.
It's odd the effect a ring-road blockage has around Watford. Backs up all the way onto the M1 and the M25 if Exchange Road is closed for more than 30 minutes. But how do you fix Bushey Arches?
How do I fix bushey arches.
The same as I fix Vicarage Road. (Stop everyone at the harvester roundabout, and let them queue all the way back up towards Croxley green tube. Queues on downward sloped hills use less fuel, people just lift their foot off the brake. And then build a mini roundabout out at the end of the croxley park to let people out of the town via the back of it, towards Ricky past Camelot.)
Bushey arches is the same. The whole congestion thing is caused by the one way system round Mothercare/B&Q. If there were no lights at all there, and only a light on the dual carriage way down from the motorway, one stopping traffic by Cost Co (already there,) and one stopping people at the bottom of bushey hill from getting past the nine hole course, and lastly one stopping them coming in on that road from harrow. Then hey presto. No traffic.
Like the motorway junction on the M1 (J28 Northbound I think.) You just stop people entering the system.
Watford has full on dual carriageway all round it. A single trafiic light at the Dome Roundabout ,watford side, stops all incoming, and a small chiswick type flyover to let the cars queue by Hunton bridge and at M1 J5 instead of by asda. Hey presto.
As for traffic on the way out. Dig down onto the foot path at hunton bridge and another filter flyover to allow out.
No more than ten traffic lights. All it takes is a set of speed cameras to measure traffic density.
Waxing is the approved technique I believe.
> But how do you fix Bushey Arches?
Nuke it from orbit. It’s the only way to be sure.
I always thought that the main issue at BA could be alleviated by taking out the lights at the Saab showroom and putting in a crash barrier (maybe 30 yards long or so) separating the two lanes heading towards Century park
People in the right lane opposite the Saab showroom, and coming from BA past the Mercedes showroom would be able to flow more easily, hopefully making BA itself less congested.
All the technology needed exists, including sensors and computers and a bit of software.
Sadly all that happens in most cases is that it moves an already marginal system with little contingency in case anything happens, even closer to the absolute maximum, thereby reducing the available contingency even further.
And when the supervisory system fails (as it inevitably will), the result is extended gridlock.
Nothing new here, they already knew this back in the days of The Italian Job (the proper one).
<---- For the Italian Job reference alone :)
Sure, VA signals have been around for decades, and many of those systems count vehicle throughput. The back end of those systems however is a legacy of their age, so they don't pass around a lot of information and the network is often partitioned into small groups of sites with pre-determined coordination settings. My experience with signal control systems is limited though and I'd be surprised if there aren't more advanced systems outside my experience already in use.
I'm interested in the statement "For example, the system would detect that traffic on one section of road is slowing down or has stopped, and divert it to alternate routes to prevent the backfill causing gridlock." Are they suggesting that the system dynamically and forcibly changes drivers' route choices? How do you do that? The seppos aren't going to like that one.
Well, normally, I would just jack up the OSPF cost, but that doesn't work with the current automobile-traffic systems. No idea how the current lights would enforce that a path was unusable, so there would have to be other technology applied for that bit.
Sigh...as if taxes aren't high enough.
I see lots of comments about "traffic lights have been doing that for ages" - yes, but only locally. What this guy is proposing is a city-wide model, hence the huge amounts of maths involved, which is not driven by clock variations as most sensing lights are, but by actually, city-wide need (I would caveat that one, no idea if sensors can pick up bicycles etc).
He seems to have worked out the maths for a city-sized green wave which may actually work. I know of some green waves in Belgium (Hasselt) which only are green as long as you break the locally imposed speed limit, no doubt to generate revenue through the traffic cameras which append to be installed at every single light. Zürich is also a place that loves speed cameras (typically mounted in places where you should have your eyes on the road instead of on your speed, especially in 30 km/h zones), but it is a good place to try this out. First of all, the peak hour mess around Bellevue makes it easy to measure effect (just measure queue length along the lakeside), and they already have all the required sensors and controls in place.
London is another good example, with the added benefit that it is large enough to just take a section and try it. Anything that improves the West End and around Hyde Park corner (to name but two of many, many bottlenecks) is worth the money IMHO..
Another good mess to sort out is the so-called "ring of steel" around Amsterdam - the main reason people use the totally overloaded trains there is because using a car to get in from outside during peak hours is even worse..
I've seen bicycle specific sensors but only in painted bicycle lanes so far.
"Well, normally, I would just jack up the OSPF cost,"
You could always try route poisoning. Especially as more and more people have traffic aware GPS.
"I would caveat that one, no idea if sensors can pick up bicycles etc"
I'm sure some can. Most, however, cannot even pick up a motorcycle. I would know, the number of times I've been stuck at a red light for an extended (i.e. abnormal, much longer than I have ever waited in a car) period of time, to have the lights change a few seconds after a car pulls up behind me.
I now avoid such junctions on the bike when I know about them.
There are some that can pick up bikes. Back in my student days I used to cycle past a car park and regularly made a slight detour to set off the car park entrance sensor and get a ticket.
I was wondering how many times I could do it before the computers would think the car park was full of cars that never chose to leave.
Oxford's bus/cycle lanes have a pedestrian-crossing-style push button at the head of the queue. This allows cyclists to gain the advantage of the bus lane's priority traffic lights even when no bus is coming.
Regarding sensors picking up (motor)bikes, I recently saw a do-it-yourself article on one of the "maker" sites that simply involves taping a couple of neodymium magnets to the frame or pedals. I don't cycle in urban areas (too scared), and so haven't tried it, but it may be worth trying.
On the "green wave", Sheffield has the exact opposite (red wave?). Getting from Hillsborough to J34 M1 (or the other way) on the main road/new ring road, travelling at the speed limit in quiet traffic, will have you stopped at over 3/4 of the lights. At night, the lights will change to red as you approach them. The only way to avoid being stuck at otherwise empty junctions is to exceed the speed limit by about 8mph - which most people do, obviously.
> What this guy is proposing is a city-wide model
Existing traffic systems have been using a city-wide model for decades.
He *might* have come up with a clever new algorithm. He probably hasn't - most such "innvoations" have a fundamental flaw that has been overcome many times. But what he has *not* done is to create something entirely new - a city-wide, flow-modelled control system with feedback loops of varying sizes is in existence in most major cities around the world, and has been for a very long time.
 My favourite is the single signal head: every year, some educator uses his influence to get in front of the bods at one of the big traffic companies to show off his latest protege's idea. This idea will inevitably be a signal head with a single lens, and a multi-coloured illuminator behind it. This used to be separate bulbs and a bit of optics, now it's assuredly LED. He will go through all the supposed benefits of simplified maintenance etc. Eventually, one of the audience will crack, and utter that fateful observation "red-green colour blind"...
> I've seen bicycle specific sensors
Standard vehicle sensor loops are supposed to be sensitive enough to pick up bicycles.
I know of one in Southampton that is clearly out of tune, because it doesn't detect my bike.
> I'm sure some can
When properly set up, they all can. It's part of the design spec.
Some of them are not properly tuned, and so the discrimination is not as tight as it should be.
> taping a couple of neodymium magnets to the frame or pedals.
I wouldn't bother with that.
The sensor loop is a parallel tuned circuit - you've got a big coil of wire in the road, and a capacitor across it. The whole is driven by an oscillator - you've got a high-Q filter, so when in tune, there is negligible current flowing.
When something metal passes over the loop, the circuit loses tune, so current starts flowing. This is a very sensitive detector.
Over time, the components and road environment age, and you'll get some drift in the tuning. I did read about an idea to allow the loop driver to auto-tune, but I've no idea if that ever made it into production - it's a long time since I've done anything in this field.
Swindon's "Northern Orbital" road has the same problem - as you enter this long stretch of 40mph dual carriageway with at least 10 junctions of traffic lights, if you're in a pack you're quite likely to hit greens but the second you hit a red you will continue to get them unless you speed. The problem is that the red opens a gap which the neighbouring lights take to mean "hey, I can go red now", which will then happen as you approach. That's why everyone does 50-60 down that stretch.
Now they're talking about using funky speed cameras that trigger the lights to go red if someone is approaching at greater than the speed limit - fun all round as you'll be damned if you do, and damned if you don't.
I use a car park with a weekly ticket.
Sometimes there's a bug in the system, so they leave the exist gate open for everyone to go out without presenting their ticket.
Then next time it won't let you in because it thinks you're still in there. You have to take a single-use ticket to be allowed in, then waste 15 mins at the office getting a credit for it.
I have never seen a traffic queue at a broken set of traffic lights as long as the queue when those same lights were fully functioning. Rip them all out and save money. Rip up all the one way system as well and shorten the journeys.
Huh? You want everyone to revert to Italian driving? No thanks.
Actually that's not as daft an idea as you might think. I just finished Tom Vanderbilt's "Traffic: Why we drive the way we do" which was a hell of an interesting look at all the problems of traffic. Even if you don't agree with him, it makes you think about it all.
There was an interesting case a while ago in Bangor, North Wales. The council re-modelled the junction outside the railway station. Afterwards, there were terrible traffic queues.
A few months later, the traffic lights failed and the queues disappeared. The locals all asked for the traffic lights to be removed. In typical council fashion, they ignored the local's request and fixed the traffic lights.
If it's right outside the railway station, it's entirely possible the signals were installed to improve pedestrian access to the station.
Many years ago, Bristol City Council spent a load of money doing studies and re-designing the Three Lamps junction leading out of Bristol on the A4 because it was a notorious bottle-neck.
Apparently the contractors they hired chucked down some temporary lights whilst they were working and the junction had never flowed more freely!
They told the Council who said "stick with our design" and, when they'd finished, it was as much of a bottle-neck as ever...!