Mapping pedestrian traffic light timing in Sydney, Australia | Hacker News Reader

Mapping pedestrian traffic light timing in Sydney, Australia

jakecoppinger.com

79 pointsjakecopp51 comments

Transport for NSW, the government agency which controls traffic signal timing in Sydney and NSW, has an excellent stated goal of increasing walking and cycling trips – and reducing pedestrian wait times at intersections.

However, there is no public data on traffic light timing in Sydney or NSW.

I built this little React site to crowdsource measurements and inform where positive changes could be made - Sydney focused, but works anywhere in the world if you'd like to make use of it/fork it!

I wonder if the $75 fine for jaywalking and NSW Police's love for issuing these fines are related to how long the timings are.

Just cross back from the crossing. It’s legal (?not illegal) to jaywalk in NSW if you are more than 20m from the nearest designated crossing, if you are not interfering with traffic (section 234). I’ve never met anyone who has received a fine for crossing at the wrong time?

Most folks in the city just walk through the red light when there's no traffic - I've seen them fine folks in waves around Central and Town Hall.

As an European which spend 4 years in Sydney recently I was always surprised at the opposite. Every after going out at midnight, Hurds of people would be waiting for a red light with no car/traffic in sight for a mile. I always jaywalk, feel particularly upset that a computer is telling me to wait for a save time, without context, while I can obviously tell it's save.

Recently discussed this behaviour in Sydney, remember it was particularly bad around Surry hills

"just walk through the red light when there's no traffic"

You can't blame them, it really is damn infuriating when nothing is coming. Same applies as a driver, if I could have back all the time I've waited for lights to change I reckon I'd be years younger. Sydney's traffic light system is brain-dead.

Incidentally, years ago I was at a trade launch by a well known company and I found myself sitting next to a person who was one of the programmers for the traffic light system and I made the mistake of mentioning how bad the system was. Wow, he nearly took my head off, it was such as sensitive subject. He then proceeded to lecture me about how difficult the problem was and how good the local effort was compared with many other comparable cities.

I've lived in the Sydney CBD (Haymarket) for 10 years. I don't own a car because I don't need one. I jaywalk at every opportunity many times every day. I've never been fined for jaywalking.

One time I was crossing the street with my wife near Town Hall against a red pedestrian light and I was smiling because of the conversation. We were walking toward two uniformed NSW Police officers whom I failed to notice as I was looking out for traffic when crossing the street. One of them, in her twenties, snapped at me like I was a naughty child (I'm around 50) with "you have the nerve to smile while illegally crossing? I could fine you you know". I didn't want a fine so stopped smiling and apologised and continued on my way.

That said occasionally there is a coordinated session where they hang out at specific intersections waiting to fine people but they don't try to hide themselves (usually there's a parked police motorcycle on one corner of the intersection, etc.)

When you're a uni student, they don't hesitate, they just fine you.

Thanks for making this! I’ve found that as a pedestrian in Sydney, up to half my walking time is spent waiting for traffic lights, and have always wondered what could be done about it.

"...up to half my walking time is spent waiting for traffic lights,"

Amen. ...And often when no traffic is coming.

When I moved from Europe to Sydney, it was striking how much traffic lights deprioritised walking. Visitors have independently commented on similar things.

Its not just timing, but also inefficient phases. I requested the dataset for phases, but after months was given a totally different dataset. I gave up.

Australia is at least 40 years behind in traffic management.

There is no traffic lights priority for public transport. Melbourne's tram avg speed is 10-15 kph. If you power walk you can go faster than the tram. It just sits there waiting on traffic lights more than moving.

There is no synced traffic lights. On long avenues you just stop every 200m because once the light goes green when you arrive at the next intersection the light there goes red. It seems it is synced, but to the opposite it should be doing.

Sydney's traffic light system (Sydney Coordinated Adaptive Traffic System, or SCATS) is supposedly "a recognised worldwide market leader in intelligent transport systems" and has been exported to a bunch of cities around the world. It does (supposedly) prioritise public transport in phase timings.

https://en.wikipedia.org/wiki/Sydney_Coordinated_Adaptive_Tr...

> On long avenues you just stop every 200m because once the light goes green when you arrive at the next intersection the light there goes red. It seems it is synced, but to the opposite it should be doing.

I recall a link a while back (from HN?) that demonstrated that it's basically impossible to sync green traffic lights in both directions. Any attempt at syncing will priotitise one direction over the other. I can't find the source, and it was so long ago that it's very likely that I'm misremembering.

Re. SCATS, I used to work at the RTA as a consultant for another company back in ~2000 when it was still based as Eveleigh. My understanding is and has always been that they were the ~first to put loops under the road to count traffic but the other elements of the system were not that amazing. They still had a PDP live, and also ran the Sydney portion of the emergency services radio network from the same data center. A control system would show real time traffic densities. We were adding a PTZ-capable remote video on demand component circa 2000 using a combination of 128K ISDN and fiber. I wrote some kind of last minute multiplexing PTZ data control bridge kludge in perl. The idea was that, after the traffic system detected an anomaly, the control system operators could "dial in" and "look around" to "determine" the cause of traffic issues and create an appropriate response using their bridgedeck mounted joysticks straight out of the 80s. We sold some to the Philippines and I got to go, as an 18 year old, and party hard with the locals after we signed a deal with the mayor of Manila. Fun times. Corrupt as knifespoon.

These days I assume most people/companies/governments source their traffic flow info from Google because it has more 'active' sensors in so far as every car has at least one mobile phone. Telcos could and should also sell this type of data in an anonymous format. Large toll roads track individual vehicles with a combination of wireless billing systems and numberplate ID so they are effectively able to provide a similar solution for those high value roads.

> it's basically impossible to sync green traffic lights in both directions.

For a bi-directional green wave, intersections have to be spaced at intervals consistent half the cycle time (times the intended travel speed) of the traffic lights. In that case opposing vehicle streams will always meet each other at an intersection, thereby minimising the green time required along the green wave and maximising the green time available for handling crossing and turning traffic.

If an intersection deviates from that mathematically optimal location, it means that there'll be less of an overlap between the green times for the two opposing direction, meaning that there's less green time left over for cross traffic.

And since in real life, intersections are rarely spaced at the mathematically optimal interval, in practice this means that eventually you have to give up and just favour one direction depending on the time of the day. (To some extent you can try to salvage things by varying the assumed speed between intersections in order to keep the travel time from intersection to intersection constant, but there's only a limited usable speed range between too slow and the legal speed limit.)

And if you attempt to have multiple green waves crossing each other, you introduce even more scheduling constraints which might leave no room for a perfect solution.

Germany does it very well. Until I make a turn, I do not wait at red lights more than once or twice.

Melbourne is the same. Buttons that change the timing are rare or maybe even non-existent. I can’t think of any intersections where pressing the button does anything other than activate the pedestrian lights.

I have never seen an intersection where beg buttons favor pedestrians and not drivers, because beg buttons give engineers the latitude to establish red crosswalks unless requested. So help you if you arrive just in time to miss your activation cycle and have to wait a full light cycle for the crosswalk.

I've experienced plenty of beg buttons that don't do anything other than activate the crossing light when that direction next goes green for cars. But also plenty where pressing the button takes immediate steps for the pedestrian: if the direction is already green for cars, the walk sign is turned on; otherwise, the current cycle ends and any intermediate cycles proceed with minimum duration; non-dominant directions will have longer minimum cycle time if the pedestrian button has been pressed.

In my experience, if the intersection processes car traffic dynamically, it will also process pedestrian traffic dynamically, and generally favor pedestrians. If the intersection has static timing, it's not going to do anything different for pedestrians. I recall driving in Milwaukee near UWM two decades ago, every morning I could count on the lights showing exactly the same thing at the same time. If I left my apartment at the same time, I'd get the same sequence of missed or made lights. Most other places I've driven had car sensors (usually loops in the pavement, but cameras are cheaper and less maintenance now), so there's no schedule to be learned, although the behavior is predictable if you learn the cycles and observe the stimuli. (Although, sometimes sensors are broken or miswired; there was an intersection near one of my employers where the cyclist button was connected as an left turn lane occupancy sensor; resulting in consistently wrong behavior. it's also not uncommon for a disabled vehicle to be located above a loop, or sometimes a construction / utility vehicle to be parked above a loop; some control systems can adapt and filter that out, but many don't.)

I don’t see beg buttons in Sydney that turn green mid-phase — where are these? You always have to wait for a complete phase (which can be shortened, with no traffic and no area phasing) before the pedestrian signal is allowed to turn green again.

The only partial exception is pedestrian-only crossings and then only if you are in the part of the phase where enough time still is available to return cars to green.

I've seen quite a few intersections (Germany) that only feature one car road, but a crossing pedestrian/bike street that after pressing the beg button immediately begin to switch the light, resulting in a consistent, short wait (apart from signal priority stuff for busses). But anything more complicated, nothing comes to mind.

https://bicycledutch.wordpress.com/2016/06/21/traffic-lights... has plenty examples of things more complicated, for example:

“In the past, buses always got priority. Now every bus is tagged and the computer knows all the time-tables of every line. The system knows exactly when which bus needs to arrive at a certain intersection. The installation checks whether a bus is too early, on time, or running late and sets the lights accordingly.”

“On the cycleways we always have double loops, separated by 75 centimetres. That way we can measure speed and direction by combining the data the computer receives from the two loops. It also means we don’t need to reserve green time for a cyclist riding against traffic, away from the intersection.” And the pushbutton? “For cyclists that’s not really necessary. They’re only there for reassurance. You often see in the reports that people don’t even bother to push the button.”

But there the priority is pretty much everything else other than the beg button.

I can't help but wonder if such aggressive timings discussed in the article are good for people with disabilities. People often panic when rushed by a change to red.

I also wrote a blog post on beg buttons in the Sydney CBD: https://jakecoppinger.com/2022/12/sydney-cbd-is-bringing-bac...

(194 comments on HN: https://news.ycombinator.com/item?id=34058004)

> Buttons that change the timing are rare or maybe even non-existent

Buttons are old tech. https://www.flir.com/products/trafione/:

“The FLIR TrafiOne uses thermal imaging and Wi-Fi technology to adapt traffic signals based on the presence detection of vehicles, bicycles and pedestrians, even in total darkness or adverse weather.”

I don’t know how smart that version is, but there are versions that predict whether pedestrians have the intention to cross the street. https://www.tugraz.at/en/tu-graz/services/news-stories/tu-gr...:

“The innovative camera-based system recognises the intention of pedestrians to cross the road and switches to green automatically. What’s more, it optimises the traffic flow further, as Horst Possegger from the Institute of Computer Graphics and Vision explains by way of two examples: “The green phase can be extended in the case of large groups of persons, who require more time to cross the road. And if persons leave the waiting area before the lights have turned to green, this is also passed on to the lights. The traffic lights subsequently don’t switch to green and there are no unnecessary waiting times for motorised traffic.”

[…]

“Using the current configurations, our system signals that wish to cross three to four seconds before the button is pushed,” says Possegger.”

Feels like you could do that somewhat cheaper with just visual cameras and any off-the-shelf YOLOv# detection model.

For the pedestrian problem at least, it seems even cheaper to just use a button. That also completely bypasses the problem of trying to detect if they intend to cross. Is there really much of a disadvantage? The pedestrian has to be in a specified area beside the crossing already to use it, so likely within arms reach of a pole already. But I can see it being useful for cars and bikes - as a cyclist I've hit a good number of times the magnetic sensors clearly just don't work for a bike, and had to wheel over to the pedestrian crossing to press the button to get the lights to change.

And I expect "person detection" to be much easier with an IR image, there's probably not many places where human body temperature isn't noticeably higher than the background should make it more accurate to pick out. And if you don't need accuracy of temperature reading, an IR camera isn't really any more expensive than a visual camera - it's the same CCD just with a different filter on top. And it completely bypasses any problems of darkness.

I also wouldn't be surprised if the actual camera unit was a near negligible cost of installing something like this, so even if it was multiples of the price, optimizing for that may be the wrong target if it affects accuracy at all.

> Is there really much of a disadvantage?

Granted it's probably nothing too big in absolute terms, but does feel a little unfair, because vehicles usually get a few seconds head start for free due to being able to call a signal as soon as they reach the location of the green extension detector, which can be a few tens of metres in advance of the stop line.

(On the other hand yes, I do acknowledge the because "turning lanes" don't really work for pedestrian traffic, it's often hard to detect pedestrians much in advance because you'll have no idea which way they'll want to turn at the intersection. But it still feels a little unfair :-) )

And unless the default state of an intersection is "red for all directions", switching the lights will always take a few seconds even in the best of cases. By giving the traffic lights controller some advance warning of somebody approaching, it's possible to switch the traffic lights just in time so that that user doesn't have to stop (if traffic conditions allow it of course – with heavy traffic it's unavoidable that somebody will have to wait eventually).

With semi-flexible traffic light designs (where phase length and whether a phase is called at all depends on traffic conditions, but otherwise they still always cycle through the phases in a fixed sequence), there are also situations where a signal can only turn green (if called) until a certain point in the cycle no later than X (in which case it'll remain green until let's say for example X + 10 seconds). If you miss that time X, you'll have to wait a whole signal cycle until the signal can turn green again. If it just happens that you're reaching the intersection at a time X + 5 seconds, it means that if you could have called the signal while still six seconds away (i.e. at X - 1 second), you still could have crossed successfully (because in that case the signal would have been green between X and X + 10), whereas without that possibility you can only press the button at X + 5 seconds and now have to wait one whole signal cycle.

In the USA, the Manual of Uniform Traffic Control Devices has an (I think) reasonable requirements for crossing durations but almost no further recommendations about maximum wait times. This results in completely idiotic timing programs of the signals everywhere that encourage people to cross when cars have a green light. If you want to change it to be more pedestrian friendly, it can cost $20k+ because the signals are programmed in a 40 year old language that only a very small group of highly paid traffic engineers know how to write.

I happen to be a traffic engineer who operates signals in the USA. We may not have the most advanced technology but I can assure that we don't not write code to operate our signals. We do enter different parameters that govern the amount of time that you have to wait.

In most situations it comes down to if you want to favor vehicles or if you want to favor pedestrians. As everyone has mentioned it is difficult to accommodate both as they would like.

Also as government worker I can assure we are not highly paid. We make a decent living but there is actually a shortage of people right now that want to do the work.

I trust my sources (mainly, a city planner that's been working here for 30 years). Certain engineering specialists out here sometimes make absolutely absurd amounts of money because they work on contracts with such limited scope, the equivalent of 600k or more a year.

Also, I think the notion of having to balance pedestrians' interests with drivers' is a cultural thing that filters down from the MUTCD and into traffic engineering. Drivers create almost all the risk associated with our road system yet their interests are prioritized highly. Traffic engineers are bound to this deadly model that prioritizes drivers, because road construction projects are not given government funds (which they often completely rely on) unless they conform to MUTCD.

Sorry one last thing. I do try to ride my electric scooter to work. :)

> the signals are programmed in a 40 year old language that only a very small group of highly paid traffic engineers know how to write.

This is interesting. What language is this, or more generally, where can I read more about this?

Traffic light systems are designed using software such as LINSIG, TRANSYT, CORSIM/TRANSYT-7F or VISSIM.

Per wikipedia different countries use different systems over time:

e.g. Sydney: 3,400 traffic signals co-ordinated by the Sydney Co-ordinated Adaptive Traffic System (SCATS). Designed and developed by RTA, the system was first introduced in 1963 and progressively developed since then. By October 2010, SCATS was licensed to 33,200 intersections in 144 cities across 24 countries worldwide, including Singapore, Hong Kong, Dublin, Tehran and Minneapolis and Detroit.[19][23][24]

https://en.wikipedia.org/wiki/Traffic_light_control_and_coor...

If you are super bored. Here is a manual that is VERY detailed explanation on programming traffic signals. I give this to new engineers and technicians to look at when they first start.

Chapters 5 and 6 get into the actual programming of the signal.

https://ops.fhwa.dot.gov/publications/fhwahop08024/index.htm

edit: Whops. Just realized theyre talking about the US. Disregard my comment.

I'm not entirely sure what OP is talking about, but I think they use a system called SCATS.[0] Looks like the system is actually pretty widely used in some other countries as well.

[0] https://www.scats.nsw.gov.au/solutions/products

> If you want to change it to be more pedestrian friendly, it can cost $20k+

Per set of lights?!

I could be wrong. I can't locate the source I found originally. Online chatter for a city planning forum said around 20k. This very old information says it's a lot less, but I don't think these numbers would be accurate for a small town with only enough money to fix their signals piecemeal by contractors, or for one that involves smarter programming that works around MUTCD. https://www.itskrs.its.dot.gov/its/benecost.nsf/ID/215f723db...

Don't forget the infuriating situation where infra designers/whomever decide "we can't put a sidewalk here", or governments allow construction to close a sidewalk for years on end.

Why it's acceptable to put up signs saying "don't cross here" and instead require that people cross the road 3 times to go a few dozens of feet is beyond me.

> Don't forget the infuriating situation where infra designers/whomever decide "we can't put a sidewalk here", or governments allow construction to close a sidewalk for years on end.

My favorite instance of this happened when the town decided to improve the dodgy sidewalk that followed a busy road. The sidewalk was in rough shape, and while it was walkable, it would have been impossible for anyone with a mobility impairment.

Improving the sidewalk meant that it was closed for almost 8 months. During that time, I had to walk at the edge of the vehicle travel lane. As winter approached, they finally finished construction and I enjoyed the beautiful condition of the new sidewalk...for about two months. After the first snow storm, it became impassible and, for reasons I never found out, the town never plowed it again. So, I had the option to travel back in traffic, or trudge through the snow. The situation ended up being a factor in me giving up and buying a car.

This is awesome, and I'd add that in doing safe streets advocacy in SF I have learned that to make changes like these it's helpful to record the entire signal timing for all phases to make more concrete recommendations for changing timing. In other words you'll have to take time from other signal phases to increase ped timing and without that data it's hard to advocate for change. Here is one example of a full phase visualization: https://glitch.com/edit/#!/aframe-ped-signals?path=README.md...

This is actually one of the most annoying things I’ve noticed in Brisbane - the lights for walking take absolutely forever. To the point that everyone will j walk because it can take 5 minutes for a light to change; it makes getting around the city so frustrating as a person coming from Northern Europe

If I were to figure out a localised instance would you collect data? I don't go into the CBD that often is all.

edit: or just do data collection for Brissie directly:

https://betterintersections.jakecoppinger.com/?lat=-27.5&lon...

You don't have to set up another instance - this site works for anywhere in Australia (or the world - but maybe their traffic lights are too different.

Currently I just default the view to Sydney but this could be modified to view where a user is (rough IP geolocation), and I've got a work in progress PR for storing the lat/lon/zoom in the URL so you can bookmark/share locations.

I guess the question I have is what you intend the main interface experience to be for people - the form or the map?

I imagine there's a few trick things that could be done on the map page like nailing the coordinates, doing the timer --- and then bouncing over to Google Forms via the power of pre-filled links.

Definitely would prefer the map (and associated pages/possibly form in future) to the be main interface.

The Google Form makes it very easy to bootstrap collecting emails (reliably) to prevent any spam that may be submitted, but it's long and confusing to new users (already got that feedback).

Ideally in future you could click on an intersection on the map, use a timer in the page (could optimise the UX), associate an email address with a one time code (on first load/if no cookie) then submit (with the OSM node ID already filled in so I don't have to manually update that).

So that would be the long term roadmap - if it gets enough traction and I have time!

I'm surprised traffic light timing is a focus for increasing cycling trips.

Bikes can use the same timing as car traffic (we go when the cars go). I know the "bike paths" of sydney have their own little green cycling man, but I mostly ignore those.

The bike lanes are infuriating. with multiple areas requiring the crossing of 3 lanes of traffic where the bike lane is on the right side going into an intersection, and on the left on the other side of the intersection.

I (and many other cyclists I've seen) are regularly just cycling in car lanes because the bike lane is incompressible in some areas. Often, it is only for a block or 2 until you can get back into the bike lane safely, and stay in the lane for at least a few blocks.

I often wonder if the person who is in charge of designing bike lanes in Sydney even has a bike.

I've been meaning to time the traffic lights where I am, most are really good for pedestrians, but there are some here and there that take an age to change. I even changed my walk to work as the shortest path has a couple of particularly slow crossings, making a slightly longer route quicker as it only has zebra crossings.

Are we documenting “silent walks”, where timing is flexibly maximised into whatever the car traffic is doing? They’re a tiny percentage overall, but on quiet t-junctions they can last hours in the dead of night. “Silent” because the accessibility sounds deactivate until the phase changes.

I’d very much like to see for Melbourne!

You can add data for intersections in Melbourne! The default view is Sydney but it works everywhere.

PRs welcome for setting initial landing coordinates to rough geolocation!

Reply on news.ycombinator.com